CB1 receptor signaling regulates social anxiety and memory


Corresponding author: Y. Litvin, PhD, 1230 York Ave. Box 165, New York, NY 10065, USA. E-mail: ylitvin@rockefeller.edu


The endocannabinoid (eCB) system regulates emotion, stress, memory and cognition through the cannabinoid type 1 (CB1) receptor. To test the role of CB1 signaling in social anxiety and memory, we utilized a genetic knockout (KO) and a pharmacological approach. Specifically, we assessed the effects of a constitutive KO of CB1 receptors (CB1KOs) and systemic administration of a CB1 antagonist (AM251; 5 mg/kg) on social anxiety in a social investigation paradigm and social memory in a social discrimination test. Results showed that when compared with wild-type (WT) and vehicle-treated animals, CB1KOs and WT animals that received an acute dose of AM251 displayed anxiety-like behaviors toward a novel male conspecific. When compared with WT animals, KOs showed both active and passive defensive coping behaviors, i.e. elevated avoidance, freezing and risk-assessment behaviors, all consistent with an anxiety-like profile. Animals that received acute doses of AM251 also showed an anxiety-like profile when compared with vehicle-treated animals, yet did not show an active coping strategy, i.e. changes in risk-assessment behaviors. In the social discrimination test, CB1KOs and animals that received the CB1 antagonist showed enhanced levels of social memory relative to their respective controls. These results clearly implicate CB1 receptors in the regulation of social anxiety, memory and arousal. The elevated arousal/anxiety resulting from either total CB1 deletion or an acute CB1 blockade may promote enhanced social discrimination/memory. These findings may emphasize the role of the eCB system in anxiety and memory to affect social behavior.

Sociality is essential for the survival of all vertebrates. Social skills facilitate evolutionarily crucial functions such as reproduction (Pfaff 1999), hierarchy formation (Blanchard et al. 1995), pair bonding (Donaldson & Young 2008), maternal behavior (Carter & Altemus 1997) and threat communication (Litvin et al. 2007). Sociality is profoundly affected by fear and anxiety (Haller et al. 2004; Klugmann et al. 2012). Both fear and anxiety are adaptive emotions that protect the animal from external threats. Conspecifics may pose a particularly salient threat to an organism, especially during mating periods or times of limited resources. Thus, social behaviors such as investigation, recognition and discrimination are essential for the survival and procreation of a species, and are modulated by fear and anxiety. The identification and study of neural underpinnings of social behaviors in laboratory animals are relevant for determining the mechanisms of human disorders with abnormalities in sociality, such as autism spectrum disorder, social anxiety, depression and pathological aggression.

For centuries, marijuana and hashish have been utilized as euphoria-inducing psychotropic compounds. The psychoactive component of the cannabis plant, delta-9-tetrahydrocannabinol (THC), exerts its effects on physiology and behavior primarily through activation of the cannabinoid type 1 (CB1) receptor (Howlett et al. 2002). The endogenous endocannabinoid (eCB) N-arachidonylethanolamine (AEA) and 2-arachidonoylglycerol (2-AG) are synthesized and released postsynaptically to act as retrograde messengers at presynaptic CB1 receptors located on glutamatergic and gamma-aminobutyric acid (GABA)ergic axon terminals (Katona et al. 2001, 2006) to suppress neurotransmitter release (Wilson & Nicoll 2001), which in turn control synaptic plasticity (Katona & Freund 2012). In line with this function, eCBs bind to CB1 receptors in distinct brain regions to modulate behavioral processes related to cognition, stress and emotionality (Hill & McEwen 2010; Riebe & Wotjak 2011), all of which profoundly affect sociality.

The eCB system in general and CB1 receptors in particular have been implicated in stress, fear and anxiety, as well as in associated memories and their extinction (Hill et al. 2010b; Marsicano et al. 2002; Metna-Laurent et al. 2012; Steiner & Wotjak 2008). Specifically, CB1 receptors are involved in stress habituation processes (Hill et al. 2010b) through a process that likely involves the coordinate actions of glucocorticoids and glutamate neurotransmission (Hill et al. 2010b; Kamprath et al. 2009).

The present series of experiments were aimed to assess and compare the effects of pharmacological blockade, with those of constitutive genetic knockout (KO) of CB1 receptors (CB1KOs), on social anxiety and discrimination in two ethologically relevant animal models. Using this approach, we aim to compare and discriminate between manipulations of acute eCB tone with potential long-term developmental changes that occur in non-inducible KO mice.

Materials and methods


Male C57BL/6J mice were bred in house, weaned at postnatal day 21 and group housed (n = 5 per cage). Animals were individually housed 1 week prior to the commencement of experiments and between experiments. Animals were 8–10 weeks old at the time of testing. The CB1KO mice used in this study were originally generated and backcrossed to a C57/BL6J background (Zimmer et al. 1999) and were provided by the National Institute of Mental Health. The CB1KOs and their wild-type (WT) controls were bred from heterozygote mothers and CB1KO fathers and WT breeding pairs, respectively. All the heterozygote mothers used were derived from the same pool of animals that underwent backcrossing on C57/BL6J mice obtained from Charles River. All breeding of heterozygote mothers with either CB1KO or WT fathers was done in the same animal facility at the same time under a timed breeding schedule. Thus, the KO and WT animals used in this study were not littermates, but were all bred from mothers generated from the same backcrossing and raised in the same facility, in the same colony room under the same environmental conditions. All testing was conducted during the dark phase of the light/dark cycle.

Stimulus mice were all outbred CD1 mice aged 10–14 weeks. Mice for the social discrimination were castrated 1 week prior to testing.

Social investigation

The ‘social investigation’ model is a modified version of the ‘social partition’ test chamber, used by our lab (Litvin et al. 2011; Murakami et al. 2011) and others (Berton et al. 2006; Elliott et al. 2010; Krishnan et al. 2007; Kudryavtseva 2003) (see Fig. 1, top). The social investigation apparatus is a 50-cm long, 12-cm wide and 50-cm high box constructed of white Plexiglas except the front, which is clear to permit observation. It is divided into three equal sections by small marks on the clear Plexiglas front wall (16.66 cm each). All testing was conducted during the dark cycle under dim red light. Intruder mice were introduced into the right side of the chamber for 5 min with an empty perforated box in the left corner of the apparatus (‘habituation’ to the apparatus). Following this 5-min ‘pre-exposure’ period, a novel male was introduced into the left side of the chamber inside the perforated box for an additional 5 min ‘post exposure’. The 10-min testing duration was recorded on a digital video recorder for later analysis by an observer who was blind to group assignment.

Figure 1.

(a) Experimental timeline and (b) apparatus (top: social investigation and bottom: social discrimination). Social investigation consists of a 5-min ‘pre-exposure’ in which experimental mice are placed in the chamber with an empty stimulus box (left), after which they are exposed to a 5-min ‘post-exposure’ in which a novel male conspecific is introduced into the stimulus box. Social discrimination consists of four ‘habituation trials’ of 5-min exposures to the same two castrated mice followed by 15-min exposures to two empty stimulus boxes. The 5-min trial consists of replacing one of the mice with a novel conspecific and testing for social discrimination.

Experimental mice were exposed to a 5-min habituation period to the apparatus including the empty perforated box in which the novel mouse was later introduced in order to control for behaviors or location preferences that may be exhibited irrespective of the social stimulus (‘pre’ period). Analysis was performed on the frequencies and durations of exhibited behaviors during the 5 min while the novel mouse was present (‘post’ period) minus those displayed during the 5-min habituation (‘pre’ period). All groups were exposed to the novel conspecific during the post period (see Fig. 1). We divided the four groups into two rounds of experiments: WT were run with KO and vehicle were run with animals that received AM251 (CB1 antagonist).

Behavioral analyses included indices of general arousal, spatial preference within the apparatus and ethologically inspired measures of fear and anxiety using the program Hindsight. The DVDs were each scored twice: once for spatial preference and general arousal as deduced by transits—crossing of lines marking the far, medium and near thirds of the apparatus, and a second time for exhibited behaviors. Spatial measures included far—duration of time spent in the far compartment relative to the stimulus; medium—duration of time spent in the medium compartment relative to the stimulus; near—duration of time spent in the near compartment relative to the stimulus and contact—duration of time spent in contact with the chamber in which the stimulus was contained. Other dependent behavioral measures analyzed in each test situation included freezing—complete cessation of movement other than respiration; freezing is considered a ‘passive coping’ defensive strategy. Risk assessment is the combined measure of both stretch approach—forward ambulation with flat back and stretched neck—and stretch attend—standing on all four paws with flat back and stretched neck orientated toward the stimulus (novel male conspecific). Risk assessment is considered an ‘active coping’ defensive strategy. Grooming is the movement of forepaws or tongue over the body.

Social discrimination

Mice were tested for social discrimination 4 days following social investigation testing. All animals received the same drug/vehicle as in the social investigation test (see Fig. 1, bottom). The social discrimination consisted of exposing mice five times (trials 1 through 5) within the same apparatus as used for the social investigation (see above) in a binary choice test (Choleris et al. 2006). In this test, two cylinders (4 inches tall and 10 inches in diameter made of aluminum wire mesh; modified Office Depot pencil holders) each containing a stimulus mouse were placed in two corners of the apparatus. Each trial lasted 5 min, and the trials were repeated at 15-min intervals, during which empty cylinders were placed in the apparatus. In trials 1–4, the same two stimulus mice in identical locations within the social investigation apparatus were used, whereas in trial 5, one of the two stimulus mice was novel, whereas the other was the same familiar mouse, and in the same location, as in trials 1–4. The familiar mouse that was switched for the novel mouse at trial 5 was randomized across subjects. Importantly, in all trials, the number (two) and location of stimulus mice that the experimental animal received were kept constant to reduce possible confounding effects due to changes in the amount of stimulation received in the five trials. Importantly, we refer to ‘novel’ as the position where the novel mouse on trial 5 will be placed and ‘familiar’, as the position where the familiar mouse will be placed on trial 5. While testing, the mice were left undisturbed in the room, and their behavior was videotaped for subsequent behavioral analysis. At the onset of the dark phase on the day of testing, experimental and stimulus mice were moved to a darkened holding space next to the testing room and were left undisturbed for 90–120 min. Stimulus mice were placed into clean cylinders 5 min before being placed into the testing apparatus. New cylinders were used for each test so that the novelty factor of the cylinders remained constant throughout the experiment. The cylinders were thoroughly cleaned with 10% ethanol and then they were paper towel- and air-dried. All stimulus mice were castrated so that fear and aggression do not factor into the motivation of the test mice to sniff the stimulus mice.

Behavioral analyses included location preference relative to the cylinders.


To pharmacologically determine a role for eCB signaling on social investigation and discrimination, we used the CB1 receptor antagonist AM251 (5 mg/kg; Tocris-Cookson, Minneapolis, MN, USA) dissolved in dimethyl sulfoxide, Tween 80 and physiological (0.9%) saline (1:1:8, respectively). We chose the dose based on several published studies that have established it as producing a clear anxiety-like profile in mice (e.g. Naderi et al. 2008; Sink et al. 2010). All intraperitoneal (ip) injections for the pharmacological experiments were administered 30 min prior to testing.

Statistical analysis

Social investigation

Statistical analyses were conducted on the frequencies and durations of behaviors within particular segments of the apparatus (spatial preference) during the 5 min ‘post’ period minus those during the 5 min ‘pre’ period (‘delta duration’ or ‘delta frequency’). Frequency of line crossings during the ‘pre’ period was scored as a measure of general locomotion/arousal. Grooming was assessed during the ‘pre’ period. The KO and WT groups were assessed using two-tailed t-tests, as were vehicle and AM251 groups. Separate t-tests were conducted as opposed to a two-way analysis of variance (anova) because of the difference in methodology used [i.e. genetic vs. pharmacological study, which entailed no injections and (the stressful) injections, respectively]. Significance was determined at P < 0.05. Group numbers for the social investigation were KO = 17; WT = 15; AM251 = 20 and Veh = 14.

Social discrimination

Statistical analysis entailed two-tailed t-tests between AM251/Veh groups and KO/WT groups for investigation ratios (arcsin square root transformed; graphs illustrate investigation percents) and discrimination index. Two-way repeated measures anova was used to analyze investigation durations. Post hoc Student–Newman–Keuls tests were used to assess significance, determined at P < 0.05. Group numbers for the social discrimination were KO = 14; WT = 16; AM251 = 13 and Veh = 11.


Social investigation

AM251-treated and KO mice displayed enhanced anxiety-like behaviors in the social investigation task compared to their respective controls, although with several differences. The KO animals displayed elevated arousal/locomotion, whereas the animals treated with the CB1 antagonist AM251 displayed enhanced grooming.

The t-test showed a significant difference (t(32) = 2.16; P = 0.037) in spatial preference between animals that received AM251 (mean 29.5 ± 13.93 seconds) and Veh groups (mean −12.14 ± 15.7 seconds). Two-tailed t-test showed a tendency of KO animals (mean 20 ± 15.1 seconds) to display avoidance toward the novel conspecific (t(30) = 1.78; P = 0.08) when compared with WT animals (mean −12.62 ± 8.97 seconds).


In order to control for arousal/locomotor effects, the number of transits between the three compartments of the apparatus during the ‘pre’ period was analyzed (Fig. 2). Veh/AM251 showed no significant differences in transits (i.e. arousal/locomotion; two-tailed t-tests). However, CB1KOs showed significantly more transits, indicating higher arousal/locomotion levels than their WT counterparts (t(30) = −2.04; P = 0.05).

Figure 2.

Transits and Grooming. Number of transits between the far, medium and near compartments of the social investigation apparatus during the 5-min pre-exposure (top). Pre duration (in seconds) of grooming behavior (bottom). Data represent mean ± standard error. **P < 0.01 (two-tailed t-tests).


Results from a t-test showed that AM251 significantly elevated grooming levels during the ‘pre’ exposure when compared with Veh-treated animals (t(32) = 2.73; P = 0.013) (Fig. 2). The effects of acute CB1 antagonism on grooming have been previously reported using rimonabant (Tallett et al. 2007, 2008). Results from a t-test showed no significant differences between KO animals and their WT counterparts.

Freezing (passive coping)

Results from a t-test show that AM251 significantly elevated ‘delta freezing’ levels when compared with Veh-treated animals (t(32) = 3.29; P = 0.002) (Fig. 3). Results from a t-test show that KO animals showed significantly elevated levels of ‘delta freezing’ when compared with their WT counterparts (t(30) = 3.35; P = 0.002).

Figure 3.

Freezing and Risk Assessment. Post minus pre duration (in seconds) of freezing behavior (top). Post minus pre duration (in seconds) of risk assessment (RA) behavior (bottom). Data represent mean ± standard error. **P < 0.01 (two-tailed t-tests).

Risk assessment (active coping)

Results from a t-test showed no significant differences between AM251-treated animals and their Veh counterparts in ‘delta risk assessment’ (Fig. 3). However, results from a t-test showed significantly elevated levels of ‘delta risk assessment’ of KO animals when compared with their WT counterparts (t(30) = −3.17; P = 0.003).

Social discrimination

Both AM251-treated mice and KO mice displayed enhanced social discrimination compared with their respective controls, implicating a role for CB1 receptors in social memory.

Figure 4.

Investigation percent and Discrimination Index. Percent of time spent investigating the novel mouse during test 5 (top). Discrimination index (Novel sniff − Familiar sniff)/(Novel sniff + Familiar sniff) during test 5 (bottom). Data represent mean ± standard error. *P < 0.05 (two-tailed t-test).

Investigation percent and discrimination index (Fig. 4)

The two commonly used measures of social discrimination are investigation percent and discrimination index. A two-tailed t-test indicated that mice administered AM251 had a higher investigation percent during trial 5, when a novel stimulus mice was presented, compared with vehicle-treated animals (t(22) = 2.621; P = 0.016). The KO mice also had significantly higher trial 5 investigation ratios compared with WT animals, as assessed using a two-tailed t-test (t(28) = 2.177; P = 0.038).

Discrimination index = (Novel sniff − Familiar sniff)/(Novel sniff + Familiar sniff) is another measure of the ability to discriminate a novel conspecific on trial 5. Both KO and AM251 animals displayed significantly elevated discrimination indices when compared with their respective controls (vehicle compared with AM251: t(22) = 2.523; P = 0.019 and WT compared to KO: t(28) = 2.113; P = 0.044).

Time in novel mouse chamber

Investigation ratio results are mirrored by the amount of time AM251-treated mice spent in the novel stimulus mouse compartment compared with the familiar stimulus mouse compartment during test 5 (data not shown). Two-tailed t-tests showed a trend toward significance for mice treated with AM251 to spend a higher proportion of time in the novel mouse compartment compared with vehicle controls (t(22) = 1.768; P = 0.091); however, there was no difference between KO and WT mice for this measure (t(28) = 1.195; P = 0.243).

In addition, we did find that there was no main effect of test (F4,88 = 1.428; P = 0.232) or interaction of treatment and test (F4,88 = 0.155; P = 0.960), but there was a significant main effect of drug treatment for the total amount of time test mice spent in both the stimulus mice compartments (as opposed to the center compartment) when assessed using a two-way repeated measures anova (F1,88 = 6.025; P = 0.022). Student–Newman–Keuls post hoc test showed that AM251-treated mice spent significantly longer in the mice compartments than did vehicle controls (q(22) = 3.471; P = 0.023), which may indicate that the CB1 antagonist does in fact increase interest in social stimuli, although this is not supported by the fact that these mice do not sniff the stimulus animals more than vehicle-treated animals (discussed below). The amount of time KO mice spent in stimulus mice compartments did not differ in comparison to WT mice (F1,112 = 0.718; P = 0.404). Similarly, there was no main effect of test number (F4,112 = 0.673; P = 0.612) or interaction between genotype and test (F4,112 = 2.261; P = 0.067).

Total investigation

A two-way repeated measure anova of total investigation durations from trial 1 through trial 5 showed that there was no effect of drug treatment (F1,88 = 1.028; P = 0.322) or genetic condition (F1,112 = 0.311; P = 0.582) on the total amount of time mice spent investigating social stimuli, nor was there an interaction between the drug treatment or genetic condition and test (F4,88 = 1.465; P = 0.220 and F4,112 = 0.608; P = 0.658, respectively) (Fig. 5, top). These findings suggest that the enhanced social discrimination performance showed by AM251-treated and KO mice was not secondary to changes in interest of social stimuli or enhanced interest in novelty per se. There was a significant effect of trial for both the experiment examining the effects of AM251 (F4,88 = 4.790; P = 0.002) and CB1KO (F4,112 = 4.811; P = 0.001), although there was no significant interaction between factors of drug treatment or genetic condition and test number.

Figure 5.

Total Investigation and Novel Investigation. Total (top) and novel (bottom) investigation durations. ‘Novel’ refers to the location where the novel stimulus mouse will be placed in trial 5. Data represent mean ± standard error. *P < 0.05, **P < 0.01. Significant differences represent between-trial differences of AM251 (left) and CB1KO (right) (Student–Newman–Keuls post hoc test).

For experiments examining the effects of AM251, a Student–Newman–Keuls post hoc test indicated that investigation durations at test 1 are significantly higher than at tests 3 (q(23) = 5.210; P = 0.004) and 4 (q(23) = 3.941; P = 0.033), showing normal habituation of the mice to the stimulus animals. Investigation durations at test 5 were also significantly higher than at test 3 (q(23) = 4.576; P = 0.009) and there was a trend toward significance at test 4 (q(23) = 3.307; P = 0.056), indicating dishabituation to the novel stimulus mouse presented during test 5. Similarly, normal habituation−dishabituation to stimulus mice was found in the CB1KO mice experiment. Student–Newman–Keuls post hoc tests show significantly higher investigation durations during test 1 compared with test 4 (q(29) = 5.575; P = 0.001) and during test 5 compared with test 4 (q(29) = 4.905; P = 0.004).

Novel investigation

A two-way repeated measure anova analysis of novel investigation durations (or the mouse that will be replaced by the novel stimulus during the habituations) for the CB1 antagonist experiment failed the test for equal variances; therefore, we examined within-group differences using a one-way repeated measure anova (Fig. 5, bottom). Both novel investigation durations for animals administered AM251 and vehicle had a significant effect of trial (repeated measures anova on ranks χ2 = 22.646, df = 4, P < 0.001; one-way repeated measures anova F4,54 = 2.683, P = 0.045, respectively). Student–Newman–Keuls post hoc analyses showed that animals given AM251 did not show significant habituation from trial 1 to trial 4, but did significantly dishabituate from trial 4 to trial 5 (q(12) = 4.195; P < 0.05), whereas animals administered vehicle did not show statistically significant habituation or dishabituation.

For the KO experiments, two-way repeated measures anova of the amount of time mice spent sniffing the novel stimulus (or the mouse that will be replaced by the novel stimulus during the habituations) also shows a significant effect of trial (F4,112 = 3.348; P = 0.013), but neither for genotype (F1,112 = 1.064; P = 0.311) nor for an interaction of genotype and trial (F4,112 = 0.679; P = 0.608). Student–Newman–Keuls post hoc tests for novel stimulus investigation show a trend toward habituation from test 1 to test 4 (q(29) = 3.604; P = 0.058), followed by dishabituation between test 4 and test 5 (q(29) = 4.898; P = 0.007).

Familiar investigation

There was a significant effect of trial on familiar stimulus investigation during the CB1 antagonist social discrimination experiment as analyzed using a two-way repeated measure anova (F4,88 = 4.870; P = 0.001) (data not shown). There were no significant effects of treatment (F1,88 = 2.211; P = 0.151) or of interaction (F4,88 = 1.565; P = 0.191). Investigation durations of the familiar stimulus were significantly higher at trail 1 compared with trail 3 using Student–Newman–Keuls post hoc analyses (q(23) = 4.432; P = 0.012) and trial 5 (q(23) = 5.466; P = 0.002).

For KO experiments, a two-way repeated measures anova of the amount of time mice spent sniffing the familiar stimulus also shows a significant effect of trial (F4,112 = 3.480; P = 0.010), but not of genotype (F1,112 = 0.0395; P = 0.844) or of interaction (F4,112 = 0.522; P = 0.720). Student–Newman–Keuls post hoc tests for familiar stimulus investigation also show habituation from trial 1 to trial 4 (q(29) = 4.849; P = 0.008).


Our present results delineate the behavioral effects of CB1 receptor inactivation on arousal, social anxiety and memory by utilizing convergent pharmacological and genetic approaches in two behavioral models. Our approach compares between the manipulation of acute eCB tone, and the developmental effects that occur as a result of constitutive CB1 deletion. We show that these approaches produced similar behavioral profiles, namely anxiety-like behaviors in the social investigation procedure and increases in social memory in the social discrimination task with some differences that can attest to the fundamental differences between these manipulations. It should be noted that although our KO and WT animals were bred from heterozygote mothers, which were all from the same founder line, they were not littermate paired. While this could introduce additional variability and interpretive issues into the study, the convergence of our pharmacological and genetic data attests to the fact that the data obtained herein with the CB1KO mice are likely owing to the absence of the CB1 receptor and not owing to non-specific changes in the mice due to genetic drift or other issues related to breeding.

Our current findings extend previous reports of a role for the eCB system in mood and memory (Hill & McEwen 2010; Lafenetre et al. 2007; Marsicano & Lafenetre 2009; Marsicano et al. 2002; Metna-Laurent et al. 2012) and conform to reports showing a specific role for the CB1 receptor in these processes (Atsak et al. 2012; Metna-Laurent et al. 2012; Rodgers et al. 2005; Terranova et al. 1996). Our findings of increased cognitive abilities of social discrimination in the CB1KO mice are consistent with previous reports of increased cognitive performance in several other tasks, such as object recognition (Maccarrone et al. 2002; Reibaud et al. 1999), contextual fear conditioning (Jacob et al. 2012) and active avoidance memory (Degroot & Nomikos 2004; Martin et al. 2002).

Our results are also in line with studies showing no effect of acute CB1 antagonism on general activity (Rodgers et al. 2005) while enhancing grooming (Tallett et al. 2007, 2008).

Social investigation

The social investigation paradigm was developed with the premise that anxiety levels selectively modulate social behavior in rodents; the time spent in social investigation provides an inverse measure of anxiety. The eCB system is clearly involved in the processing of anxiety (Lafenetre et al. 2007), as observed in various behavioral models. The stress response is typically activated when an organism is faced with present or perceived challenges to homeostasis; the eCB system interacts with the stress response at multiple brain regions to affect behavior (Hill & McEwen 2010).

In the present series of experiments, constitutive KO and pharmacological antagonism of CB1 receptors produced anxiety-like behaviors in a social investigation test, as indicated by increases in avoidance, freezing and risk assessment toward the unfamiliar conspecific. There were no significant differences in any behaviors during the 5-min pretest, suggesting that the effects are specific to social cues and not to novelty. These results compliment previous findings outlining the involvement of CB1 in sociality. A previous report indicates that the effects of CB1 on social anxiety are pronounced in a more stressful context (Haller et al. 2004). In addition, deletion of cortical and striatal GABAergic CB1 receptors, but not those expressed on dopamine D1-expressing striatal medium spiny neurons, has been reported to increase social exploration, whereas deletion of CB1 in glutamatergic cortical neurons resulted in decreased exploration (Haring et al. 2011). In apparent contrast, a report shows that when CB1 receptors were deleted from neurons expressing D1 receptors, mice displayed anxiety-like behaviors in a social investigation paradigm (Terzian et al. 2011).

Interestingly, our results indicate distinct differences between pharmacological and genetic approaches. Although mice that were injected with AM251 showed increased passive coping, namely avoidance and freezing when compared with vehicle-injected animals, CB1KOs showed elevations in both passive (freezing) and active (risk assessment) defensive coping strategies. These results suggest that although acute pharmacological blockade induces pronounced fear, constitutive KO promotes a state of fear intermixed with anxiety.

Social discrimination

Social anxiety facilitates learning and memory of social bonds, which enable the formation and maintenance of social structures essential for survival and propagation of a species. The eCB system regulates stress and emotionality, both of which are central to memory formation. Specifically, eCBs are involved in stress habituation processes (Hill et al. 2010a; Patel & Hillard 2008) that likely facilitate the ability to distinguish between stimuli that are worthy of memory consolidation from those that are rather forgotten. Endocannabinoids interact with glutamate to regulate stress habituation (Kamprath et al. 2006, 2009; Riebe et al. 2012); thus, elevations in fear in animals in which CB1 signaling is perturbed may be owing to elevated glutamate signaling. In line with this hypothesis, our results show that disruption of CB1 by genetic or acute pharmacological means leads to elevated levels of social recognition, which may be facilitated by interference with stress habituation. In apparent contrast to our findings, a recent study shows no difference between CB1KOs and WT mice in short- and long-term social and object recognition memory (Jacob et al. 2012). Despite the contrary, in this social recognition experiment, control animals displayed high levels of discrimination, and thus further enhancements in social discrimination were unlikely to be detected owing to a ceiling effect (see Phan et al. 2011). Additionally, the authors employed a social recognition protocol whereby test animals were exposed to a juvenile for 4 min and short- and long-term memory was tested after a retention interval of either 1 or 24 h by re-exposure to the familiar juvenile together with a novel animal. Recognition was determined if the animal showed preference for the unfamiliar animal. Here, we employed a procedure whereby test animals are exposed to the same two mice for 4 × 5 min trials, each separated by 15-min intervals before testing with one novel and one familiar stimulus mouse. Thus, our procedure utilizes consecutive habituation trials as part of a clear process involved in learning to discriminate between conspecifics. As such, our procedure is better suited to determine the involvement of eCBs in the process of discrimination and recognition.

General discussion

These findings may be understood in light of the importance of eCB signaling in the amygdala, a central locus involved in fear processing and expression. Current theories suggest that there is ‘ demand’ role of eCBs during fear processing, whereby eCBs are produced in response to highly aversive stimuli and facilitate the recovery from a state of fear back to a state of normal functioning (Marsicano & Lutz 2006; Riebe et al. 2012). Disruption of eCB signaling increases anxiety and potentiates fos activation in the amygdala, suggesting that there is an eCB tone in the amygdala that constrains anxiety and gates activation of this structure (Hill et al. 2010b; Newsom et al. 2012; Patel et al. 2005). Consequently, blockade of this pathway could increase anxiety (Hill et al. 2012). Increased activation of the amygdala as a result of reduced eCB tone or deletion of CB1 receptors would thus increase vigilance and threat detection, and result in anxiety-like behavior. This process may be related to structural changes within the basolateral nucleus of the amygdala, as eCBs have been shown to regulate synaptic plasticity in this structure, likely affecting emotionality (Azad et al. 2004; Kodirov et al. 2009; Marsicano et al. 2002). In addition, recent studies have employed local injection approaches to directly modulate CB1 function in the amygdala (Campolongo et al. 2012; Kamprath et al. 2011). In the study by Kamprath et al. (2011), local blockade of CB1 in the central nucleus of the amygdala by AM251 regulated postsynaptic responses and potentiated fear in an auditory fear conditioning paradigm. Furthermore, eCBs have also been shown to regulate activity in the medial amygdala (Krebs-Kraft et al. 2010), a nucleus centrally involved in sociality (Gabor et al. 2012). As such, in addition to general effects on arousal that may affect memory systems in the basolateral nucleus (as shown here by changes in transits in CB1KOs in the social investigation task), disruptions in eCB signaling may affect mechanisms in the medial amygdala involved in social discrimination per se.

Enhanced glutamate release has been suggested to regulate fear responses observed in CB1KOs (Kamprath et al. 2006). A recent study showed the involvement of amygdalar GABAergic CB1 receptors in passive coping, whereas cortical glutamatergic CB1 receptors regulate active coping (Metna-Laurent et al. 2012). In addition, it was reported that administration of THC exerted a biphasic control of fear coping strategies, with lower and higher doses favoring active and passive responses, respectively. These observations may suggest that the predominant site of action of AM251 may be cortical CB1 receptors and may also indicate a dose-dependent effect on behavior. To test this hypothesis, further studies using anatomical ablation and/or pharmacological intervention are required.

We suggest that both eCB tone and acute regulation serve to gate emotional behaviors in response to conspecifics, underlying key functions that are evolutionarily highly adaptive and are most likely conserved in mammalian species, including man. Consistent with this, gene variants in components of the eCB system in humans are related to differential amygdala and striatal activation in response to social threat and social reward stimuli (Chakrabarti & Baron-Cohen 2011; Chakrabarti et al. 2006; Domschke et al. 2008; Hariri et al. 2009). Furthermore, a polymorphism of fatty acid amide hydrolase (eCB-degrading enzyme) that results in elevated AEA levels shows blunted amygdala activation to threat (Hariri et al. 2009). Moreover, activation of CB1 receptors through consumption of THC dampens amygdala reactivity in humans in response to social threat (Phan et al. 2008), consistent with the high rate of self-medication and cannabis dependence in individuals with social anxiety (Buckner et al. 2006a,b, 2008, 2012a,b,c,d). In accordance, it is plausible that deficits in eCB signaling may predispose individuals to social anxiety, and that cannabis use may occur as a means of increasing CB1 receptor activation and reducing social anxiety.

Another possibility that could account for the enhanced social memory involves a noradrenergic pathway. Increased norepinephrine release in the brain is known to facilitate the consolidation of emotionally salient memories (McIntyre et al. 2012). The CB1 receptors are found on noradrenergic terminals and disruption of these receptors increases norepinephrine release in limbic regions (Srivastava & Lutz 2012; Tzavara et al. 2003). Thus, it is possible that disruption of eCB signaling could enhance noradrenergic release during emotionally salient events, such as social interaction, which would enhance social memory and discrimination.


These results indicate that a deregulation of the eCB may have profound effects on social anxiety and memory. By its very nature, the eCB system may be involved in psychopathologies and conditions with symptoms of social deficits, such as autism, aggression and social anxiety.


We would like to thank Hanna Sofia Salm and Hilary Lambert for excellent technical assistance and behavioral scoring.