Animal experiments provide good opportunities for brain research. However, the data are often ambiguous.
2.1. Alterations in Cognitive Functions upon Direct Action on CB Receptors and after the Deletion of CB Receptors
The data obtained in studies on rodents indicate that the ECS modulates specific aspects of learning and memory. Thus, Pério and co-authors showed that SR141716 (rimonabant, a selective CB1R antagonist) dose-dependently improved the performance of the social recognition task in rats and also attenuated the deficit exhibited in aged mice and rats performing the same task [
52]. Consistent with these data, an improvement in memory performance for object recognition in CB1R knockout mice (-/-) compared with CB1 (+/+) wild-type mice was found [
53]. Also, the results of the study by Lichtman in which rats were trained to perform a task in a radial eight-ray maze showed a better task performance after the administration of SR141716 than in the control [
54]. The beneficial effect of rimonabant on memory was confirmed in the work by Wolff and Leander [
55]; rimonabant also reversed THC- or anandamide-induced memory deficits [
56] and attenuated sleep deprivation-induced memory impairment in rats [
57]. Recently, Ghazvini and colleagues also revealed a positive effect of rimonabant: this CB1R antagonist improved the methamphetamine-induced impairment of object recognition and social behavior [
58]. Another selective CB1R antagonist, AM251, may attenuate short- and/or long-term memory deficits in the inhibitory avoidance test [
59]. On the other hand, WIN 55,212–2 (WIN), a potent CB receptor agonist, impaired recognition memory in rodents [
60,
61] or showed no effects on methamphetamine-induced impairment of object recognition and sociability [
58].
However, the effects of improving memory during CB1R blockade were not observed when animals performed the tasks where the participation of working memory was necessary [
62,
63,
64]. Thus, when studying the effect of cannabinoid drugs on the learning of rats, which had to memorize a different sequence of three items, SR141716A, a CB1R antagonist, as well as the CB1R agonists anandamide and CBD, did not affect the speed and accuracy of the performance of this task. In contrast, THC and the long-acting endogenous ligand analogue R-methanandamide caused a dose-dependent increase in the error rate and a slower response. SR141716A (1 mg/kg) removed the effects of THC and R-methanandamide. Thus, the CB1R antagonist SR141716A in a dose effective for blocking the action of THC and R-methanandamide, by itself, did not affect the performance of the task in the working memory test [
62]. When training rats in the test for spatial memory in the Morris water maze, the effects of improving memory with CB1R blocking were also not observed [
51].
In contrast, systemic administration of THC, WIN-2, and CP55,940 (CBR agonists) impaired working memory in rats; interestingly, unlike the listed drugs, anandamide (CB1R agonist) and CBD had no visible effect on working memory [
65]. In the studies by Lichtman and colleagues, when CP55,940 was injected into the hippocampus, this drug dose-dependently reduced the accuracy of task performance, although did not increase the time of its execution; therefore, the authors suggested that the effects of THC, WIN-2, and CP55,940 on working memory were mediated through the CBR in the hippocampus [
66,
67]. In another early work, the effect of the synthetic CB1/CB2 receptor agonist HU-210 on learning and memory consolidation was studied; two variants of the Morris maze, with a fixed position of the platform hidden in the water and with a visible platform, were used. The administration of HU-210 60 min before training at doses of 50 and 100 μg/kg/daily for four days
disrupted learning only in a more complex task (with a hidden platform). In contrast, at a dose of 25 μg/kg, HU-210
facilitated training in any platform position. Thus, different doses of this CBR agonist oppositely affected the learning in a complex task in the Morris maze. Importantly, as noted by the authors, the CBR agonist HU-210 at doses of 50 and 100 μg/kg caused tigmotaxis, which is observed on increased arousal; therefore, the effect of HU-210 in this case may be mediated not by direct action on CB receptors but by other mechanisms [
68].
Learning the task in the Morris water maze is dependent on the hippocampus; this task is used to investigate spatial navigation and memory. It is interesting that a single injection of an extremely low dose of THC (0.001 mg/kg) significantly affected the performance of the task by mice in the complex Morris water maze test 3 weeks later. THC-injected mice showed both longer escape latencies and lower scores in the execution of this test compared to their matched controls, indicating the induction of cognitive deficits [
69].
The long-term administration of the CBR agonist WIN-2 led to a deterioration in the performance of the task of recognizing a new object (NORT) in mice [
70]. Besides, in a study using functional imaging (PET), this long-term introduction of WIN-2 affected brain metabolism and functional connection between the hippocampus and the prefrontal cortex, between the thalamus and the prefrontal cortex, and between the hippocampus and the perirhinal cortex, i.e., between the structures involved in memory processes. The injection of AM 251, an antagonist of CB1R, removed the disturbances in the NORT task in mice [
71].
Interestingly, the effects of cannabinoids on cognitive performance in animals change with age, with stronger negative effects being observed in the pubertal phase compared to the adult [
72,
73,
74]. In view of this, Verrico and colleagues carried out a research of the effect of THC on working spatial memory and object recognition memory in adolescent monkeys. Monkeys were injected intravenously with THC, and the chronic effect of the drug on the efficiency of working memory was studied depending on the time of testing (delay time) after the last injection. In control animals, a clear increase in the accuracy of performing the task for spatial working memory was found with a greater delay (71 h), while in THC animals with this delay, a dose-dependent decrease in performance accuracy was observed. However, regarding the object recognition tasks, which do not have an emotional component, no THC effects have been identified in adolescent monkeys [
75]. In adolescent mice, the chronic administration of THC led to immediate and long-term impairments in the performance of object recognition and working memory tasks; there was also an increase in anxiety of animals. Likewise, chronic administration of WIN-2 to rats of late adolescence led to a memory deficit in the object recognition test; on the contrary, the administration of THC to adult mice caused only an immediate but not long-term deterioration in object recognition and working memory [
76]. Thus, these data support the notion that adolescence is a vulnerable period and that long-term exposure to THC during this period can adversely affect the cognitive function and behavior. Interestingly, the co-administration of THC with CBD prevented both cognitive and behavioral disturbances caused by THC in adolescent mice [
73].
It was also demonstrated that exposure to THC in adolescent rats (35-45 days) can cause deficits in short-term memory in adult animals tested in the Y-maze [
74]. Interestingly, in this study, exposure of juvenile rats to THC induced a malfunction of the kynurenine pathway in the adult brain, specifically increasing the level of kynurenic acid in the medial prefrontal cortex.
In contrast to the results of studies of spatial and nonspatial memory on animals in the pubertal phase, no significant effect of chronic administration of THC was found on adult rodents [
77]. Moreover, it was recently found that in adult rats in an object recognition test, acute and chronic administration of THC (at a dose of 1.5 mg/kg, i.p., but not 0.75 and 3.0 mg/kg) improved cognitive performance; at the same time, an increase in the expression of dablcortin (a protein associated with microtubules) and the brain growth factor BDNF, as well as neurogenesis in the hippocampus were observed [
78]. This work on adult animals convincingly showed a parallel increase in cognitive functions and markers of plastic processes in the brain under the influence of THC in a specific dose.
Thus, in tests for object recognition and spatial memory, the authors obtained opposite results regarding the effects of CBR agonists on cognitive performance depending on the age of the animals and the dose of the administered drugs.
The cognitive function can also be enhanced by activating CB2 receptors; for example, this activation restored the impaired behavior of rats in hippocampus-dependent tests. Thus, it was determined in a recent study by Abd El-Rahman and Fayed whether the D-galactose-induced impairment of cognitive behavior in ovariectomized female rats can be restored via CB2R activation [
79]. The authors examined rats using the novel object recognition and the Morris water maze tests and found the return to normal behavior in both cases by injection of the CB2R agonist AM1241.
Interestingly, aging animals showed improved cognitive performance under the influence of THC [
80]. The authors demonstrated that a low dose of THC reversed the age-related cognitive decline in mice aged 12 and 18 months when performing a hippocampus-dependent spatial memory task in the Morris water maze. This effect was accompanied by an increase in the expression of synaptic marker proteins and in the density of spines in the hippocampus. The restoration of transcriptional gene patterns in this structure was also observed. Besides, the expression profiles of these genes in 12-month-old mice treated with THC were very similar to those without THC in mice at the age of 2 months. The transcriptional effects of THC were critically dependent on CB1R on glutamatergic neurons, since their inhibition blocked the positive effects of THC. The authors suggested the optimistic conclusion that the restoration of CB1 signaling in the elderly may be an effective strategy for treating age-related cognitive impairment.
It is noteworthy that although some early works have provided evidence of selective deficits in the hippocampus-dependent memory under the influence of cannabinoid drugs [
67,
76,
81,
82], in a recent study, it was found that a low dose of the CBR agonist WIN-2 (1 mg/kg) and URB597 (a potent selective inhibitor of FAAH, 0.2 mg/kg) improved avoidance memory consolidation [
83]. Analysis of the results allowed the authors to conclude that the effects of WIN-2 on memory consolidation were mediated predominantly by CB1R activation, with the involvement of CB2R.
In addition to the hippocampus, the medial prefrontal cortex (mPFC) was found to be a critical site for CB1R-dependent modulation of acquired fear responses [
84,
85,
86]. However, in the experiments of these authors on rats, not a context, but a certain signal (smell or sound) was used as a conditioned stimulus, and the reaction was considered as hippocampus-independent. Exposure to odor, previously associated with an electrocutaneous irritation, increased the burst activity in a subpopulation of neurons in the mPFC [
87]. When a CB1R antagonist was injected into mPFC, the acquisition of a conditioned freezing reaction was blocked, which was associated with impaired neuronal bursting activity in this area of the neocortex and a decrease in LTP in the synapses of afferent fibers from the basolateral amygdala to PFC [
84,
85]. These data indicate that CB1R signaling at amygdala-mPFC synapses is involved in the coding of the fear response to olfactory conditioning.
The results of a recent work by Pires and colleagues [
88] confirm the facts obtained in early experiments. Using the Morris maze and chronic (up to 22-29 days) administration of WIN-2 (2 mg/kg, i.p.) in different groups of mice, they studied its effect on different phases of memory, learning (with the injection of the drug before the test for working memory) and consolidation (after this test), with parallel assessment of gene expression in the hippocampus and the prefrontal cortex. Insignificant cognitive impairments were found only in short-term working memory, which interfered with learning; however, long-term memory (consolidation) was not disturbed. Besides, an increase in the expression of DAGL-α, an enzyme for the synthesis of 2-AG, and a decrease in the level of MAGL, its degradation enzyme, were found in PFC in animals that received WIN-2 before training; at the same time, mice injected after training to assess memory consolidation, showed opposite changes. By the authors’ oppinion, minor cognitive impairments caused by the administration of WIN-2 may be associated with a possible increase (above normal) in the concentration of 2-AG in PFC. For genes associated with AEA metabolism, no correlation was found between molecular and behavioral data [
88].
In a number of studies on the effect of activation of the ECS on learning and memory, the neural activity in the hippocampus and, in parallel, the temporary coordination of this activity by the field theta rhythm were analyzed. In particular, Robbe and Buzsáki [
82] showed that the synthetic CB1R agonist CP55940 caused cognitive deficit in rats performing a spatial task of delayed alternation in a modified T-maze and decreased the power of theta, gamma, and ripple oscillations in the hippocampus. In these experiments, the activity of place cells forming the internal “spatial map” was not disturbed; thus, no “remapping” was observed, but the binding of the activity of the place cells to the phase of theta wave was significantly deteriorated. The temporal coordination of cell ensembles was also impaired in short time intervals (<100 ms). The authors believe that cannabinoids can impair memory primarily by disturbing the temporal dynamics of hippocampal neurons, regulated by theta rhythm [
82]. Thus, according to the authors, cannabinoids do not change the representation of space but disrupt the coordination and synchronization of the activity of hippocampal cell ensembles, which encode information and thereby disrupt spatial memory.
Interestingly, the aforementioned study by Marsicano and colleagues [
28] demonstrated a new mechanism for astroglial control of synaptic plasticity and memory through the D-serine-dependent modulation of NMDA receptors. The authors found the memory impairment in mutant mice lacking CB1 receptors on astroglial cells (GFAP-CB1-KO), when they had to recognize new objects in the L-maze; a decrease in LTP in the hippocampal CA3-CA1 synapses in vivo and in vitro was also observed. The activation of CBR by the administration of the WIN-2 agonist increased the intracellular astroglial Ca
2+ level and the extracellular level of the co-agonist of synaptic NMDA receptors, D-serine, in hippocampal slices. Accordingly, in in vivo experiments, GFAP-CB1-KO mice exhibited a lower occupancy of the D-serine binding site. The administration of 5 μM WIN-2 selectively increased the level of D-serine; at the same time, LTP impairment and memory disturbance were completely prevented in GFAP-CB1-KO mice [
28]. Thus, the activation of astroglial CB1R controls the activity of NMDA receptors and hippocampal LTP by regulating the synaptic level of D-serine, a signaling amino acid.
As regards the role of CB2 receptor activation in modulating cognitive functions, interesting results were obtained by Manzanares and co-authors [
89]. This work clearly showed that memory (as assessed by the hippocampal-dependent passive avoidance test) of CB2R knockout mice was impaired compared to wild-type animals. The selective CB2R agonist JWH133 was shown to improve memory consolidation, while the CBR antagonist AM630 worsened memory responses. Later, Kruk-Slomka and Biala [
90] showed that JWH133 at a low dose (0.5 mg/kg) had no effect on learning but enhanced the consolidation of long-term memory in the passive avoidance test. At the same time, JWH133 at higher doses (1 and 2 mg/kg) improved both the acquisition and consolidation of long-term memory. Subsequently, similar results were obtained by Alarcon and colleagues [
88]. At the same time, it was found in another work on CB2 receptor knockout mice that hippocampal-dependent long-term contextual fear memory was impaired, while hippocampus-independent cued fear memory was normal. In contrast to CB2 receptor knockout, acute blockade of CB2 receptors by AM603 in C57BL/6J mice did not affect memory [
91]. Thus, it can be assumed that CB2R ligands are of particular importance in the formation of long-term hippocampus-dependent memory. It should also be noted that the specific effects of CB2R ligands on cognitive processes seem to be quite complex and still cannot be exactly assessed.
An important role of the ECS in cognitive functioning was revealed in the work by Busquets-Garcia et al. [
92], where an original learning model, the so-called
mediated learning, was used [
93]. A typical initial behavioral procedure in this model is sensory preconditioning, where pairs of two minor stimuli (smells, light, tones, gustatory stimuli) are accompanied by the classical conditioning of one of them with an aversive or appetitive unconditioned reinforce. As a result of these associations, subjects avoid or prefer a stimulus that has never been clearly combined with a conditioned stimulus [
94,
95]. Sensory preconditioning involves three different, sequential processes. First, an incidental association is formed between low-significant stimuli during the preconditioning phase; second, direct association of one of initial signals with the reinforce stimulus enhances its salience during the conditioning phase; third, the presentation to the subject of any of the initial signals (directly associated with the conditioned stimulus or never associated with it) reveals the retrieval of direct or mediated memory, respectively. It should be noted that the behavior of animals in natural life is more often associated precisely with mediated learning based on previous experience; the same applies to human behavior [
93,
96]. Busquets-Garcia et al. [
92] used this model of incidental learning and found that this learning was impaired in CB1R knockout mice (CB1R-KO). In this investigation, wild-type and CB1R knockout mice were preconditioned with pairs of stimuli: smell–taste (banana (+) and almonds (-) as smells; sucrose (+) and maltodextrin (-) as taste), followed by conditioning one of two stimuli, pleasant or unpleasant; then a test stimulus was presented that was different from the conditioned one (with indirect learning) or the same content (with direct learning). The authors have convincingly shown that in CB1R-KO mice the mediated learning was impaired, while direct learning was preserved. This demonstrates that CB1Ps are essential for this type of wildlife training. At the same time, control mice showed no significant difference in the two learning models, classical and mediated. Interestingly, CB1R knockout mice (CB1R-KO) exhibited impaired mediated learning regardless of the sensory modality of the test stimulus. This study also provided evidence that the activity of cholecystokinin-containing CB1R expressing GABAergic hippocampal neurons plays a crucial role in mediated learning. The authors ultimately concluded that fine regulation of hippocampal GABAergic interneurons via CB1R can explain how humans and animals integrate and associate a variety of randomly occurring low-salience signals so that, as a result, they develop a seemingly unreasonable attraction or aversion to specific objects, places, or people [
92]. Thus, the use of nonstandard strategies in the study of the ECS can reveal its specific role in cognitive behavior.
Interesting results were also obtained by the research group of Bénard & Marsicano who showed the dependence of cognitive deficits caused by CBR agonists on mitochondrial CB1 receptors [
97]. In their study of hippocampus-dependent memory, it was demonstrated that the synthetic cannabinoids WIN-2 and HU210, administered intrahippocampally, cause acute memory impairment in mice during the recognition of new objects in an L-maze. Genetic removal of mitochondrial CB1 receptors in hippocampal neurons prevented cannabinoid-induced impairment in mitochondrial motility, synaptic transmission, and memory formation, which was accompanied by the normalization of mitochondrial respiration and ATP production. Thus, the data of these authors evidenced that bioenergetic processes occurring in mitochondria of hippocampal cells operate as subcellular regulators of cognitive functions mediated by CB1 receptors [
97].
2.2. Changes in Cognitive Functions upon Modulation of Metabolism of eCBs
Modulating the levels of the eCBs (i.e., anandamide and 2-AG by (the) pharmacological blockade of their catabolism) is a promising approach to the treatment of AD. The inhibition of the two main endocannabinoid hydrolase enzymes, FAAH and MAGL, enhances the level of the endocannabinoids accessible for interaction with their receptors. Most importantly, this manipulation augments no relevant side effects (for details, see reviews [
33,
98]).
In the work by Yasar and colleagues [
99], the effects of URB597 (a FAAH inhibitor) and WY14643 (an agonist of PPARα) on the learning of rats in the hippocampus-dependent passive avoidance task were investigated. The drugs were injected before or immediately after the training session (to assess the effect on memory acquisition and consolidation, respectively) or before a test conducted 24 h after the training session to determine their effect on memory retrieval. URB597 and WY14643 induced significant improvement in learning. This facilitation was blocked by MK886, a PPARα antagonist. It is known that PPARα is a target of the eCB AEA (except for CB1R); therefore, the blockade of the FAAH enzyme by URB597, which leads to an increase in the level of AEA, had the same effect as the administration of the PPARα agonist. On the other hand, no effects on consolidation or memory retrieval were observed after the administration of WY14643 [
99]. These results demonstrated novel mechanisms for enhancing learning through PPARα activation: either directly through the injection of a PPARα agonist, or indirectly through the administration of an FAAH inhibitor.
Busquets-Garcia et al. studied the role of the endocannabinoids AEA and 2-AG, as well as rapamycin, in modulating the specific types of memory (contextual hippocampus-dependent memory and memory on object recognition in the V-maze) [
100]. In these experiments, two inhibitors of eCB catabolism, which increase the levels of AEA and 2-AG, as well as THC and rapamycin were injected to the mice of different groups. The authors showed that an increase in the 2-AG level did not affect memory consolidation and mTOR signaling in the hippocampus; at the same time, the modulation of AEA and the administration of THC induced the disturbance of these processes, which was removed by rimonabant (i.e., through CB1R) [
100]. Besides, the pharmacologically elevated AEA level (above normal concentrations
) impairs LTP in hippocampal slices, as well as learning, and memory in behaving mice (spontaneous alternation and spatial recognition memory in the Y maze); however, any significant effect on CB1R protein levels was not revealed. As the authors beleaved, the elevated AEA level inhibits CaMKIV and CREB phosphorylation via the activation of CB1Rs [
56].
Thus, a diversity in the effects of increased content of the two eCBs was found: 2-AG did not change the memory, and AEA caused its deficiency.
However, Campolongo and co-authors [
83] convincingly showed that the consolidation of
aversive hippocampus-dependent memory is facilitated by increasing the level of AEA by the administration of URB597 through the activation of CB1 and CB2 receptors. A year later, these authors found that memory consolidation in the avoidance task was also facilitated by an increase in the concentration of 2-AG caused by the systemic administration of JZL184 immediately after training. In this case, the memory consolidation was facilitated by the activation of CB2 receptors and the prevention of the mTOR signaling activation in the hippocampus through the CB2R-dependent mechanism [
101]. Thus, these two works [
83,
101] showed the role of both CB1 and CB2 receptors in the consolidation of memory in the model of memorizing negative experiences that require the activation of inhibitory mechanisms. It should be also emphasized that conflicting results regarding the effect of URB597-mediated increase in the AEA level on memory consolidation were obtained: no effect [
99], deterioration [
100], and improvement [
83]. Some studies also revealed different effects of the JZL184-mediated increase in the 2-AG content on memory consolidation: no change [
100] or facilitation [
101]. It is important to note that these studies did not always control the modulation of the levels of other biologically active lipids; differences in their concentration may be the reason for the observed inconsistencies in the results. Thus, it was reported that the facilitation of learning in the task of passive avoidance under the influence of URB597 was mediated mainly by another biologically active lipid, oleoylethanolamide (OEA), which affects PPARα, but not CBR [
99]. OEA was shown to play an important role in the regulation of the activity of the basolateral amygdala [
102]; besides, the level of AEA in this area was modulated by emotional stimuli [
103].
Summing up the effects of eCBs on cognitive functions in animals, one can conclude that the use of the most adequate experimental approaches, for example, mediated learning or the application of olfactory signals that are of the greatest importance for rodents, allowed one to demonstrate the positive influence of ECS activation on both learning and plastic processes in the hippocampus. These approaches break the popular opinion about predominantly negative influence of eCBs on cognition. In addition, taking into account the presence of CB1R on astroglial cells, experiments revealed their significant role in memory and the development of LTP in the hippocampus. Using a test based on the involvement of the hippocampus in the control of behavior, it was convincingly shown that the consolidation of hippocampus-dependent memory is facilitated by an increase in the level of AEA, through the competitive activation of CB1 and CB2 receptors, and in the level of 2AG, mainly via the activation of CB2 receptors.