Marijuana (Cannabis sativa) and its primary psychoactive component, δ-9-tetrahydrocannabinol (Δ9-THC), have long been known to disrupt cognition in humans. Although Δ9-THC and other cannabinoids disrupt performance in a wide range of animal models of learning and memory, few studies have investigated the effects of smoked marijuana in these paradigms. Moreover, in preclinical studies, cannabinoids are generally administered before acquisition, and because retention is generally evaluated soon afterward, it is difficult to distinguish between processes related to acquisition and retrieval. In the present study, we investigated the specific effects of marijuana smoke and injected Δ9-THC on acquisition versus memory retrieval in a mouse repeated acquisition Morris water-maze task. To distinguish between these processes, subjects were administered Δ9-THC or they were exposed to marijuana smoke either 30 min before acquisition or 30 min before the retention test. Inhalation of marijuana smoke or injected Δ9-THC impaired the ability of the mice to learn the location of the hidden platform and to recall the platform location once learning had already taken place. In contrast, neither drug impaired performance in a cued task in which the platform was made visible. Finally, the cannabinoid-1 (CB1) receptor antagonist N-(piperidin-1-yl)-5-(4-chlorophenyl)-1-(2,4-dichlorophenyl)-4-methyl-1H-pyrazole-3-carboxamide HCl (rimonabant) blocked the memory disruptive effects of both Δ9-THC and marijuana. These data represent the first evidence demonstrating that marijuana impairs memory retrieval through a CB1 receptor mechanism of action and independently of its effects on sensorimotor performance, motivation, and initial acquisition.
Marijuana (Cannabis sativa) produces a constellation of effects in humans, including alterations in perception and mood, intoxication, euphoria, increased heart rate, physical dependence upon chronic use, and cognitive impairment (Pacher et al., 2006; Ranganathan and D'Souza, 2006). During the more than 40 years since Δ9-tetrahydrocannabinol (Δ9-THC) was first identified as marijuana's primary psychoactive ingredient (Gaoni and Mechoulam, 1964), great strides have been made in understanding its actions in the brain. Δ9-THC and related chemicals, known as cannabinoids, produce their psychoactive effects by acting at the cannabinoid-1 (CB1) receptor in brain areas associated with learning and memory and elsewhere (Herkenham, 1991). Particular interest has focused on the effects of cannabis on cognition, because both naturally occurring and synthetically derived cannabinoids disrupt performance in a variety of rodent spatial (Lichtman et al., 1995; Lichtman and Martin, 1996; Mishima et al., 2001; Nava et al., 2001) and operant (Heyser et al., 1993; Hampson and Deadwyler, 1999) models of learning and memory.
Although most studies investigating the consequences of cannabinoids on learning and memory administer Δ9-THC or synthetic cannabinoids through various routes of injection, marijuana is typically smoked by human users. Because marijuana contains more than 60 cannabinoid components in addition to Δ9-THC, as well as several hundred noncannabinoid chemicals (Turner et al., 1980), it is plausible that these other constituents may elicit a distinct spectrum of pharmacological effects, alter the pharmacological effects of Δ9-THC, or a combination. Additionally, it is possible that the route of administration (injection versus inhalation) could modify behavioral effects. However, virtually no animal studies have investigated the consequences of smoked marijuana on learning and memory in laboratory animals. Therefore, the primary aim of the present study was to investigate the consequences of exposure to marijuana smoke on learning and memory, as assessed in mouse Morris water-maze tasks. For comparison, we evaluated the effects of injected Δ9-THC in this same paradigm.
The Morris water maze has been a particularly useful tool to investigate both the physiological function of the endocannabinoid system (Varvel and Lichtman, 2002; Varvel et al., 2005a) and the impact of exogenous cannabinoid administration on learning and memory (Mishima et al., 2001; Varvel et al., 2001). Repeated injections of either Δ9-THC (da Silva and Takahashi, 2002) or the potent cannabinoid receptor agonist HU-210 (Ferrari et al., 1999) before each acquisition session have been shown to disrupt the ability of rodents to learn the location of a hidden platform that remains in a fixed location. One limitation of these acquisition tasks is that they require several days of training and repeated drug administration, raising the possibility that drug tolerance could lead to an underestimation of acute cognitive-impairing effects. For example, the disruptive effects of Δ9-THC in a delayed non-match-to-sample procedure in rats were shown to undergo tolerance following repeated drug administration (Hampson et al., 2003). Using a repeated acquisition Morris water-maze procedure, we have previously shown that Δ9-THC and other cannabinoid agonists produce CB1 receptor-mediated performance deficits when the location of the target was changed each session (Varvel et al., 2001; Varvel and Lichtman, 2002). However, these experiments were not designed to focus on any particular aspect of cognition, because the mice were administered the drugs 30 min before the acquisition trial and they were assessed in the retrieval test only 30 s later. Accordingly, the second objective of the present study was to distinguish between the effects of cannabinoids (i.e., marijuana and Δ9-THC) on learning (i.e., acquisition) and memory retrieval.
To delineate between these two processes, we adapted a repeated acquisition Morris water-maze task. Specifically, the hidden platform was placed in a randomly selected position of the water maze before each session, and the mice were given five acquisition trials to learn the platform location, followed 60 min later by a 60-s probe trial in which the platform was removed from the tank to assess retrieval memory. The dose-response relationships of marijuana and Δ9-THC were evaluated independently on acquisition and on retrieval. The noncompetitive N-methyl-d-aspartate (NMDA) receptor antagonist dizocilpine maleate (MK-801) served as a positive control. Finally, to determine whether the CB1 cannabinoid receptor was involved, the mice were pretreated with rimonabant before exposure to marijuana smoke, or administration of Δ9-THC or MK-801.
Materials and Methods
In total, 72 male C57BL/6J mice (The Jackson Laboratory, Bar Harbor, ME) served as subjects in these experiments. The mice were housed four per cage in a temperature-controlled (20–22°C) environment, with a 12-h light/dark cycle. Food and water were available ad libitum in their home cages. The Institutional Animal Care and Use Committee at Virginia Commonwealth University approved all experiments.
Δ9-THC, marijuana (C. sativa, containing 5.19% Δ9-THC, 0.37% cannabidiol, 0.23% cannabichromene, and 0.20% cannabigerol), ethanol-extracted marijuana (placebo material), and N-(piperidin-1-yl)-5-(4-chlorophenyl)-1-(2,4-dichlorophenyl)-4-methyl-1H-pyrazole-3-carboxamide HCl (rimonabant; SR141716) were obtained from the National Institute on Drug Abuse (Bethesda, MD). MK-801 was purchased from Tocris Cookson Inc. (Ellisville, MO). Δ9-THC and rimonabant were dissolved in a vehicle consisting of a 1:1:18 solution of ethanol/Emulphor (Rhone-Poulenc, Princeton, NJ)/saline. MK-801 was dissolved in a saline solution (0.9%). All injections were given through the i.p. route of administration. In the inhalation experiments, different quantities of marijuana were prepared by mixing marijuana with placebo material at the appropriate ratios, keeping the total weight of all samples at 200 mg.
The water maze consisted of a large galvanized steel pool (1.8 m in diameter, 0.6 m in height) half-filled with water (22°C), with the top of a white platform (10 cm in diameter) submerged 1 cm below the surface of the water. A sufficient amount of white paint (proline-latex flat) was added to make the water opaque and to render the platform virtually invisible. Visual cues (i.e., cardboard cutout letters and shapes) were attached to the walls of the laboratory as well as attached to the sides of the tank (i.e., five sheets of laminated paper with black-and-white geometric designs). An automated tracking system (Columbus Instruments, Columbus, OH) analyzed the swim path of each subject, the path length and latency to find the hidden platform, and the time spent in the specified target areas.
The smoke exposure system, developed and used in this laboratory, has been described previously (Varvel et al., 2005b; Wilson et al., 2006). The smoke was drawn through a 27.5-cm length of Tygon tubing to the manifold at a flow rate of 1.0 l/min using a vacuum pump and flow regulator. The amount of time required to burn a 200-mg marijuana sample was 1 to 2 min. A solenoid was used to alternate the flow of smoke and fresh air to the animals every 8 s to mimic puffing. The subjects were placed into holding tubes that fit snugly into a manifold, consisting of six ports for a nose-only exposure. Tygon tubing, containing 0.5 g of glass wool fiber to sequester the smoke, was connected to the exhaust of the manifold. If the substance ceased to burn at any time, it was lit again until completely consumed.
Training. Morris water-maze training consisted of three phases: an acclimation session, fixed platform training, and repeated acquisition training. Upon each exposure to the pool, the mice were gently lowered into the water facing the wall. Before the first training day, the mice were given a single 300-s acclimation session to the pool without the platform present. During initial training, a fixed platform procedure was used in which the subjects received 8 days of acquisition training with the platform remaining in a single location. Each training day consisted of four 120-s trials, with a 10-min intertrial interval. Each trial began from one of four start positions, balanced among and between mice. If a mouse failed to find the platform within the 120-s trial, it was manually guided to it, and mice were allowed to remain on the platform for 30 s before removal from the tank. In between trials and at the end of each session, each mouse was placed in a heated cage containing dry paper towels.
After fixed platform training, the mice received 10 to 15 days of repeated acquisition training, in which the platform was placed in one of 32 possible locations in the tank before each session. Each subject received five 2-min trials per day, from different and balanced start positions. The same four starting points described above were used (one position was used twice); no mouse was run from the same starting position on two consecutive trials. Positions immediately in front of the hidden platform or in the direct center were not used as starting points. Successful performance for each repeated acquisition session was defined as swimming to the platform in less than 30 s on two of the last three trials. To qualify for testing, mice were required to achieve this criterion on three of the previous four training sessions and to maintain this level of performance throughout the study. Between each test session, the mice were given at least one training session in which they were required to swim to the platform in less than 30 s on two of the last three trials. Mice that failed five training days in a row were excluded from the study, even if they previously passed the criteria for testing.
Once mice had reached training criteria, an experiment was conducted to assess the effects of interposed delays on retrieval performance. Following completion of the fifth trial of an otherwise identical training session, the platform was removed from the tank, and the subjects were given a 60-s probe trial after a variable delay (1 min or 1, 2, 6, or 18 h). Only those mice that achieved acquisition criteria for that day were tested for retrieval. The primary dependent measure of interest was the percentage of time that the animals spent in the target zone, an area immediately surrounding the target location (300% larger than the actual target, making up approximately 8% of the surface area or the maze), compared with the amount of time spent in a control zone located directly opposite from the target zone. The starting position for the retrieval task was along the side of tank halfway between the two zones. Other measures that were recorded were the latency to target (i.e., the area in which the platform was previously located), the path length to target, and average swim speed.
Water-Maze Testing. In all drug tests, the probe trial was given 60 min after the fifth acquisition trial. A quasi-Latin square design was used in which the location of the hidden platform was counter-balanced in all quadrants of the tank. No more than two subjects under each condition were assessed from any single platform location, and each subject was assessed from a different platform location for each treatment. Mice were administered Δ9-THC (1, 3, and 10 mg/kg) or vehicle or exposed to smoke from marijuana (50, 100, and 200 mg) or placebo either 30 min before repeated acquisition or 30 min before the probe trial to assess learning and memory retrieval, respectively. As a positive control, we evaluated the effects of an NMDA receptor antagonist, MK-801 (0.1 and 0.2 mg/kg), on memory retrieval. Furthermore, we examined whether the memory-disruptive effects of marijuana, Δ9-THC, and MK-801 were CB1 receptor-mediated by treating mice with 3 mg/kg rimonabant 5 min before agonist administration. A minimal washout period of 72 h and at least one successful training session was given between treatments, and mice received between one and five training sessions between test sessions.
All 72 mice learned the fixed platform location task; however, only 65% of these subjects achieved criteria in the repeated acquisition task. Thus, 47 mice in total were used for the repeated acquisition studies. A group of 14 mice was used in the memory decay, Δ9-THC dose-response on memory retrieval, and rimonabant-Δ9-THC experiments. A second group (n = 14 mice) was used in the Δ9-THC disruption of repeated acquisition experiment. A third group (n = 12 mice) was used to assess the effects of marijuana and rimonabant in combination with marijuana on repeated acquisition and on the memory retrieval. The fourth group (n = 7) was used to assess the effects of MK-801 and rimonabant in combination with MK-801 on memory retrieval. In all of the data presented throughout this manuscript, the mice exhibited stable performance on training days or when tested with vehicle.
Cued Water-Maze Task. The minimal dose of each drug that fully disrupted the retrieval task was evaluated in a cued Morris water-maze task to identify potential sensorimotor or motivational deficits. In the cued version, the platform was made visible by placing a black rubber stopper (height, 3 cm; radius, 1.5 cm), which extended approximately 2 cm above the surface of the water, on top of the submerged platform. To reduce the total number of mice needed for this study, mice that completed the repeated acquisition testing were used for the cued task. During training, the subjects were given two 2-min trials per day from different and balanced start positions. The latency and path length to the cued platform were scored. Subjects that achieved the training criteria of swimming to the platform in less than 20 s on two of three consecutive days were administered drug, and they were assessed in a single trial the following day.
Two-way ANOVAs were used to analyze the repeated acquisition experiments in which the factors included drug treatment and trials. One-way repeated measure ANOVAs (in experiments in which all the mice completed all drug conditions) and one-way ANOVAs (in experiments in which not all the mice completed all drug conditions) were used to analyze the effects of treatment on time spent in the target and the control target areas. Dunnett's test was used for post hoc comparisons in experiments in which treatment groups were compared with a single control group (e.g., dose-response and repeated acquisition experiments). The Tukey test was used for making multiple post hoc comparisons in the rimonabant experiments. Planned comparisons were conducted between the target zone and control zone in the experiments assessing memory, and for all experiments in the cued platform task. For all analyses, differences were considered statistically significant at p < 0.05. The group data are presented as the mean ± S.E.M.
The ED50 values for marijuana- and THC-induced disruption of acquisition and recall performance were determined by least-squares linear regression followed by calculation of 95% confidence limits (Bliss, 1967). For both tests, the data were transformed to % maximal possible effect, using the following formula: % maximal possible effect = 100 × [(Emax – Xi)/(Emax – Emin)]). The dependent measure for the acquisition experiments was escape latency, and Emax is control (i.e., placebo smoke or vehicle injection) performance on trial 1, Emin is control performance on trial 5, and Xi is trial 5 performance for each respective drug-treated mouse. The dependent measure for the retention experiments was percentage of time spent in the target zone, and Emax is control performance, Emin is predicted random performance (i.e., 8%, which reflects the percentage of the total surface area of the pool encompassed in the target zone), and Xi is performance of each respective drug-treated mouse.
Performance during the Retrieval Test Is Delay-Dependent. The path lengths and latency to swim to the hidden platform during repeated acquisition training are shown in Fig. 1. Mice exhibited significant decreases in path length [F(4,140) = 16; p < 0.001] and latency to find the platform [F(4,140) = 17; p < 0.001] across the five acquisition trials, signifying that they learned the location of the platform. The path lengths were significantly reduced in trials 3, 4, and 5 compared with trial 1 (Fig. 1A). Likewise, the escape latencies were significantly decreased in trials 2, 3, 4, and 5 compared with trial 1 (Fig. 1B). The average swim speed was 17.4 ± 0.4 cm/s, and there were no significant differences in swim speed across trials.
As shown in Fig. 1C, the percentage of time spent in the target zone during the probe trial significantly decreased as a function of time delay [F(4,39); p < 0.01]. In particular, mice spent less time in the target zone following the 6- and 18-h delay than in the 1-min delay condition (p < 0.05). In contrast, no significant effect was found for the time spent in the control zone, which did not differ from chance values (i.e., 8%). Subjects spent significantly more time in the target zone than in the control zone, following the 1-min and 1-, 2-, and 6-h delays.
Mice were challenged with 0.1 and 0.2 mg/kg MK-801, which has already been established to impair Morris watermaze acquisition in rats (Ramírez-Amaya et al., 2001), and they were assessed in the retrieval task. Subjects were administered the drugs 30 min after the repeated acquisition session and given a retrieval probe trial at 60 min. Whereas vehicle-treated subjects spent significantly more time swimming in the target zone than in the control zone during the retrieval test, MK-801 led to a significant decrease in the amount time spent swimming in the target zone [F(2,12) = 6.9; p < 0.01] (Fig. 1D). In particular, the 0.2-mg/kg dose of MK-801 led to a significant decrease in the amount of time spent in the target zone compared with the saline treatment (p < 0.05). Additionally, this dose of drug significantly reduced the swim speed [F(2,12) = 4.3; p < 0.05] (data not shown). In contrast, the drug failed to elicit significant effects on the time spent in the control zone (p = 0.10). As can be seen by the representative swim traces in Fig. 1E, control mice spent a considerable percentage of the probe trial swimming in the target zone and surrounding area, whereas the MK-801-treated mice displayed no bias to any region in their swim patterns.
Marijuana and Δ9-THC Disrupt Repeated Acquisition. Two-way ANOVAs revealed significant statistical interactions between marijuana smoke and acquisition trial for both path length to platform [F(16,140) = 2.2; p < 0.01] (Fig. 2A) and latency to platform [F(16,140) = 2.1, p < 0.05] (Fig. 2B). Inhalation exposure to marijuana smoke failed to affect swim speed (p > 0.41; data not shown). Subsequent one-way repeated ANOVAs for each treatment condition indicated significant decreases in path length to target and escape latency data across the five acquisition trials under the naive, placebo, and marijuana 50- or 100-mg conditions (p < 0.05), indicating that the mice learned the task. However, mice exposed to smoke from 200 mg of marijuana failed to exhibit any improvement across the five acquisition trials for both the path length to target (p = 0.18) and escape latency (p = 0.18). The ED50 (95% CI) value of marijuana was 80 (65–99) mg of marijuana, which contained 4.2 (3.4–5.1) mg of Δ9-THC before burning.
The effects of vehicle and Δ9-THC (1, 3, and 10 mg/kg) on repeated acquisition assessments are also shown in Fig. 2. Two-way ANOVAs revealed that Δ9-THC dose-dependently disrupted acquisition, as reflected by significant main effects of drug on both path length to platform [F(3,52) = 7.7; p < 0.001] (Fig. 2C) and the latency to platform [F(3,52) = 7.8; p < 0.001] (Fig. 2D). Subsequent one-way repeated ANOVAs for each treatment condition indicated that the path lengths to target were significantly longer (p < 0.05) after treatment with 3 or 10 mg/kg Δ9-THC than with vehicle. Likewise, the latencies to target were significantly increased (p < 0.05) following treatment with 10 mg/kg Δ9-THC compared with vehicle. Although there was a significant drug effect on the swim speed [F(3,52) = 5.8; p < 0.01], none of the treatment groups differed from vehicle. This significant effect occurred because the mice swam significantly faster following 3 mg/kg Δ9-THC than following 10 mg/kg Δ9-THC (data not shown). The ED50 (95% CI) of Δ9-THC was 4.4 (2.4–7.0) mg/kg.
Marijuana and Δ9-THC Disrupt Retrieval. The effects of exposure to smoke from marijuana (50, 100, and 200 mg) on retrieval are shown in Fig. 3A. A one-way ANOVA revealed that marijuana smoke dose-dependently disrupted recall of the target zone location [F(4,24) = 5.1; p < 0.01], exposure to marijuana (100–200 mg) smoke led to significant decreases in the amount of time spent in the target zone compared with exposure to placebo smoke (p < 0.05). In contrast, exposure to marijuana smoke failed to affect the amount of time spent in the control zone (p = 0.35). The ED50 (95% CI) value of marijuana was 72 (47–110) mg, which contained 3.7 (2.4–5.7) mg of Δ9-THC before burning. Subjects in the naive or placebo conditions exhibited significant recall of the platform location, as indicated by the observation that they spent significantly more time in the target zone than the control zone (p < 0.05). Whereas mice exposed to smoke from the ethanol-extracted marijuana (placebo) focused much of their swimming in the target zone region, the mice exposed to marijuana smoke swam throughout the tank with no bias to a particular region, as shown in Fig. 3B.
Likewise, a one-way ANONVA revealed that Δ9-THC dose-dependently decreased the percentage of time that subjects spent in the target zone location [F(3,39) = 5.3; p < 0.01] (Fig. 3C). Again, vehicle-treated subjects spent significantly more time in the target zone than in the control zone during the memory retrieval test. In particular, the mice exhibited a significant decrease in the amount of time spent in the target zone when treated with 10 mg/kg Δ9-THC compared with when treated with vehicle (p < 0.05). As shown in Fig. 3D, the mice did display a bias to any region in their swim patterns when treated with Δ9-THC compared with the vehicle treatment. However, Δ9-THC failed to affect the amount of time spent in the control zone (p = 0.32). The ED50 (95% CI) value of Δ9-THC was 2.5 (1.5–4.1) mg/kg.
Assessment of Sensorimotor and Motivational Deficits. Although the disruptive effects of Δ9-THC and marijuana during the retrieval test are consistent with the notion that cannabinoids impair memory, an alternative explanation is that the drugs disrupted performance-related factors, such as the ability to see the visual cues, impaired swimming ability, or decreased motivation to swim to the platform (Brandeis et al., 1989). To address these possibilities, the mice were given the minimal dose of each drug that disrupted the retrieval task, and they were then evaluated in a cued version in which the platform was made visible by placing a visible object on it. The effects of each drug are shown in Fig. 4, A to C, for the path length to the cued target, latency to cued target, and swim speed, respectively. In the cued Morris water-maze task, 0.2 mg/kg MK-801 failed to elicit significant effects on either path length (p = 0.49) or latency (p = 0.11) to the platform, although it still significantly reduced swim speed (p < 0.001). Neither marijuana (100 mg) nor Δ9-THC (10 mg/kg) significantly affected the path length to platform, latency to platform, and swim speed measures.
Evaluation of CB1 Receptor Mechanism of Action. As can be seen in Fig. 5A, mice treated with the vehicle-marijuana combination did not spend significantly more time in the target zone location than in the control zone location (p = 0.80). However, rimonabant pretreatment blocked the disruptive effects of marijuana on performance during the retrieval test (p < 0.05; see Fig. 5B for swim traces). Similarly, rimonabant blocked Δ9-THC-induced disruption in the retrieval task [F(3,33) = 4.7; p < 0.01] (Fig. 5C). During the probe trial, subjects spent significantly more time in the target zone than in the control zone when treated with vehicle. However, there were no significant differences in time spent between the target and control zones when the mice were treated with Δ9-THC. Rimonabant completely blocked these disruptive effects of Δ9-THC during the retrieval test (p < 0.05; see Fig. 5D for swim traces). The rimonabant-vehicle condition did not significantly differ from the vehicle-vehicle condition. In contrast, none of the treatments affected the percentage of time spent in the control zone, which again did not differ from chance levels (p = 0.21). Finally, rimonabant failed to reverse the disruptive effects of MK-801 (data not shown).
For decades, marijuana has been the most widely abused illicit drug in the Unites States (Johnston et al., 2006). A common untoward effect of this drug is the impairment of both learning (acquisition of new information) and memory (retention and retrieval of previously learned information). Although the effects of Δ9-THC in animal models of cognition have received considerable attention over the years, there have been few, if any, reports of cognitive disruptions induced by inhalation exposure to marijuana smoke in rodents. Moreover, most animal studies investigating the cannabinoid-induced memory impairment fail to distinguish between the disruptive effects of cannabinoids on acquisition and memory retrieval processes. The results of the present study are the first of which we are aware that evaluate acquisition and memory retrieval in mice that are either exposed to inhaled marijuana smoke or given an injection of its primary psychoactive constituent, Δ9-THC. Direct comparisons between the effects of Δ9-THC and marijuana smoke are important because of suggestions that non-Δ9-THC constituents of marijuana may alter the pharmacological profile of Δ9-THC (Carlini et al., 1974; Zuardi et al., 1982; Russo and Guy, 2006) and because of possible differences due to the routes of administration.
In the present study, exposure to marijuana smoke before acquisition impaired the ability of mice to learn the position of a hidden platform. A subsequent experiment also demonstrated that exposure to marijuana smoke 30 min before the retention test impaired the ability of mice to return to the location where the platform had been previously located, suggesting impaired retrieval of recently learned spatial information, independently of effects on acquisition. Thus, two different aspects of cognition (acquisition and retrieval) were impaired by exposure to marijuana smoke. Furthermore, this study provides further evidence that marijuana smoke produces disruptive effects on learning and memory independently of sensorimotor or motivational confounds, because performance of the cued version of the task was not disrupted at a dose that disrupted memory. It should be emphasized that the total amount of Δ9-THC in the plant material does not reflect the actual dose administered to each animal, because the majority of smoke passes through the system and is trapped in the filter. Previous work from our laboratory comparing the absorption of Δ9-THC in mice exposed to marijuana smoke has shown that levels of Δ9-THC in the brain and plasma increased systematically with increased quantities of marijuana (Wilson et al., 2006). Brain levels following exposure to smoke from 200 mg of marijuana were roughly equivalent to those seen following intravenous administration of 1 mg/kg Δ9-THC. Importantly, the effects of marijuana on retrieval in the present experiment were blocked by pretreatment with rimonabant, demonstrating a CB1 receptor mechanism of action.
Likewise, Δ9-THC disrupted both the acquisition and retrieval of recently learned spatial information independent of effects on acquisition with no apparent sensorimotor or motivational confounds. The potent impairment of acquisition is consistent with a number of previous demonstrations of Δ9-THC's ability to disrupt working memory in a variety of tasks, including the delayed alternation T-maze (Jentsch et al., 1997; Nava et al., 2000) eight-arm radial maze (Lichtman and Martin, 1996), and previous water-maze tasks (Varvel et al., 2001; da Silva and Takahashi, 2002; Fadda et al., 2004). Interestingly, acquisition and retrieval were disrupted by similar doses of either marijuana or Δ9-THC. Based on the calculated ED50 values, each respective drug was equipotent in disrupting acquisition and retrieval. In contrast, the retrieval of well established (i.e., long-term) memories are relatively impervious to Δ9-THC-induced impairment (Varvel et al., 2001; da Silva and Takahashi, 2002), although the present results show that retrieval of recently learned information (i.e., short-term memory) is just as sensitive to Δ9-THC-induced impairment as is acquisition. As with other cognitive effects of Δ9-THC, this retrieval deficit seems to be mediated via CB1 receptors, because it was reversed by rimonabant pretreatment.
The experiments reported here required modification of the water-maze task to distinguish different aspects of repeated acquisition, or delayed match-to-place performance. Due to the novelty of these procedures, we performed a number of initial experiments to characterize performance of drug-naive mice and to test their response to the NMDA antagonist MK-801, which produces well-known cognitive impairments, as a positive control. We have shown previously that mice well trained in a fixed platform location task (four trials per day for 8 days) can retain a memory of the platform location for a remarkably long time, at least 9 weeks (Varvel et al., 2005a). However, under the repeated acquisition procedure used in the present study, performance in the memory retrieval task decays over a period of hours, and it is at chance levels by 18 h after acquisition. These findings suggest that memory systems underlying retention in the repeated acquisition task are very different from those underlying retention when the fixed platform procedure is used. The most obvious explanation accounting for the differences of cannabinoid-induced disruption of memory retrieval between these two procedures is that mice are given substantial training in the fixed platform procedure (i.e., four trials per day over 8 days) compared with the repeated acquisition task procedure (i.e., five trials in which the platform is placed in a different location for each session). Accordingly, the mice trained in repeated acquisition task are unlikely to consolidate the daily platform location into long-term memory. This explanation is consistent with the hypothesis that Δ9-THC selectively impairs short-term memory as opposed to long-term memory processes.
The similarities between the effects of marijuana smoke and Δ9-THC administration support the hypothesis that the effects of marijuana on memory are predominantly mediated via its Δ9-THC content. Specifically, both treatments were equally effective in disrupting acquisition and retrieval, and rimonabant blocked the effects of both treatments. Although the cannabidiol, cannabichromene, cannabigerol, and other non-Δ9-THC cannabinoid constituents could conceivably contribute to the effects of marijuana on cognition, it should be noted that only trace amounts of these compounds were in the present sample. In a human study, altering the concentration of cannabidiol and cannabichromene in marijuana failed to affect the study participants' performance in cognitive tasks, neurophysiological measures, and subjective effects (Ilan et al., 2005). In contrast, in a rat water-maze task increasing the relative amount of cannabidiol was found to antagonize the memory-impairing effects of Δ9-THC (Fadda et al., 2004) as well as blunt the anxiogenic effects of marijuana in humans (Zuardi et al., 1982). Similarly, the marijuana constituent Δ9-tetrahydrocannabivarin elicited CB1 receptor antagonist effects in both in vitro [i.e., CP 55,940-stimulated guanosine 5′-O-(3-thio)triphosphate binding and the effects of WIN-55,212-2 in the mouse isolated vasa deferentia] and in vivo (i.e., Δ9-THC-induced analgesia and hypothermia) assays (Pertwee et al., 2007). However, the effects of other cannabinoids present in marijuana at higher concentrations than naturally present in the plant material have yet to be systematically evaluated on cognitive processes. Thus, further research examining the impact of other marijuana constituents is warranted.
Together, the results of the present study demonstrate that marijuana smoke as well as its primary psychoactive constituent, Δ9-THC, impairs the retrieval of recent memory, independently of its effects on initial learning, sensorimotor performance, or motivation. The effects of both treatments on retrieval were mediated through the CB1 receptor. Furthermore, the particular water-maze procedures used in these experiments seem well suited to distinguish the effects of drugs on processes related to acquisition versus memory retrieval and for investigations of the mechanisms responsible for marijuana-induced cognitive deficits.
This work was supported by National Institute on Drug Abuse Grants DA015683 and DA002396.
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
ABBREVIATIONS: Δ9-THC, Δ9-tetrahyrocannabinol; CB1, cannabinoid-1 receptor; HU-210, Δ8-tetrahydrocannabinol dimethyl heptyl; NMDA, N-methyl-d-aspartate; MK-801, 5H-dibenzo[a,d]cyclohepten-5,10-imine (dizocilpine maleate); rimonabant, SR141716, N-(piperidin-1-yl)-5-(4-chlorophenyl)-1-(2,4-dichlorophenyl)-4-methyl-1H-pyrazole-3-carboxamide HCl; ANOVA, analysis of variance; CI, confidence interval; CP 55,940, (1R,3R,4R)-3-[2-hydroxy-4-(1,1-dimethylheptyl)phenyl]-4-(3-hydroxypropyl)cyclohexan-1-ol; WIN-55,212-2, R-(+)-[2,3-dihydro-5-methyl-3-[(morpholinyl)]-pyrolol[1,2,3de]-1,4-benzoxazinyl]-(1-naphthalenyl)methanone.
- Received January 8, 2007.
- Accepted June 22, 2007.
- The American Society for Pharmacology and Experimental Therapeutics