The endocannabinoid system

The ECS is composed of endocannabinoids, the machinery for their synthesis/degradation, and the receptors, namely CB1 and CB2 receptors, through which endocannabinoids exert their functions (Freund et al., 2003; Marsicano & Lutz, 2006; Pacher et al., 2006; Piomelli, 2003). Endocannabinoids, which are lipid messengers synthesized “on demand” from membrane precursors, comprise numerous members. Among these, AEA and 2-arachidonoylglycerol (2-AG) are the best-studied molecules. In the brain, activation of voltage-gated Ca²⁺ channels with/without the concomitant activation of Gq/11-coupled receptors (such as the group 1 metabotropic glutamate receptors or the muscarinic 1/3 receptors) leads to AEA and/or 2-AG synthesis (Alger, 2002; Chevaleyre, Takahashi, & Castillo, 2006; Ohno-Shosaku, Tanimura, Hashimotodani, & Kano, 2011; Piomelli, 2003). Following their transport into the cells, the degradation of AEA and 2-AG is thought to occur respectively at the postsynapse and at the presynapse (Figure 3.1A). The effects of endocannabinoids are mainly mediated by their retrograde action at CB1 receptors located on presynaptic neuronal terminals and on glial cells. Besides CB1 receptors, endocannabinoids may also act on CB2 receptors, which are mainly located on microglial cells in the CNS. As the activation of CB1 receptors by endocannabinoids leads to an inhibition of cyclic AMP levels, an inhibition of voltage-gated Ca²⁺ channels, and an activation of K⁺ channels, the result of the retrograde action of endocannabinoids on neurons is a net decrease in presynaptic excitability and thus neurotransmitter release (Figure 3.1B). Stimulation of CB1 receptors is thus a means to control for excessive presynaptic activity (Alger, 2002; Chevaleyre et al., 2006; Ohno-Shosaku et al., 2011). At GABAergic and glutamatergic synapses, such an inhibitory influence of endocannabinoids on transmitter release will bear consequences on synaptic plasticity, whether this plasticity is short lasting (as illustrated by the so-called depolarization-induced suppression of inhibition or suppression of excitation) or long lasting (Alger, 2002; Chevaleyre et al., 2006; Ohno-Shosaku et al., 2011). Recent results have indicated that 2-AG, rather than AEA, is the lipid-derived molecule involved in the inhibitory effects of endocannabinoids on short-lasting synaptic plasticity (Gao et al., 2010; Tanimura et al., 2010). In addition to the brain, the ECS is also present in the periphery, as illustrated by the presence of CB1 receptors in the heart, the liver, the muscles, the gastrointestinal tract, the endocrine pancreas, and the adipose tissue (Pacher et al., 2006; Pagotto, Marsicano, Cota, Lutz, & Pasquali, 2006). However, as opposed to the brain ECS (see below), there is still no information as to the specific relationships that may link physical exercise and peripheral CB1 receptors.

As indicated above, the ECS regulates many brain functions. Such a property is accounted for by the key inhibitory role of CB1 receptors on the release of different neurotransmitters (including the main inhibitory and excitatory neurotransmitters, namely GABA and glutamate), and by the location of this receptor in numerous brain regions (Herkenham et al., 1990; Marsicano & Lutz, 1999). This is illustrated by the observation of CB1 gene/protein expression in the cerebral cortex, the amygdala, the hippocampus, the ventral and dorsal striata, the basal ganglia, the nucleus accumbens, the midbrain (including the dorsal raphe nuclei and the ventral tegmental area), and the hypothalamus. Therein, CB1 receptors are involved in the control of essential brain functions, of which several are documented to be engaged or affected by physical exercise. For example, motor activity, food intake, energy expenditure, anxiety, learning/memory, and hedonia/motivation processes are all found to be controlled by CB1 receptors (El Manira & Kyriakatos, 2010; Lafenêtre, Chaouloff, & Marsicano, 2007; Maldonado, Valverde, & Berrendero, 2006; Pagotto et al., 2006).

Figure 3.1   The endocannabinoid system. (a) Synthesis and degradation of the principal endocannabinoids, anandamide (AEA) and 2-arachidonoylglycerol (2-AG), in neurons. (b) Mechanisms underlying the retrograde action of endocannabinoids on transmitter release. DAGL: diacylglycerol lipase; FAAH: fatty acid amide hydrolase; MAGL: monoacylglycerol lipase; NAP-PLD: N-acylphosphatidylethanolamine-hydrolyzing phospholipase D; NAT: N-acyltransferase; PLC: phospholipase C; Tr: endocannabinoid transporter.

Wheel running activity and the endocannabinoid system

Except for the seminal study by Sparling et al. (2003) and for two recent studies confirming that exercise increases circulating AEA levels, albeit in an intensity-dependent manner (Heyman et al., 2012; Raichlen, Foster, Seillier, Giuffrida & Gerdeman, in press), all studies aimed at examining the relationships between physical exercise and the ECS have been conducted in laboratory animals provided with running wheels. The first issue that has been the focus of investigation relates to the impacts of wheel running on the synthesis/release of endocannabinoids. In rats housed with unlimited access to running wheels for 8 days, hippocampal, but not frontocortical, AEA was found to be increased in the morning of the ninth day (Hill et al., 2010). On the other hand, analyses of the other major endocannabinoid, namely 2-AG, revealed a lack of influence of wheel running in either brain region. It is worthy of mention that in this study control animals were housed with PVC tubing as a means to enrich the environment, thus leaving open the possibility that the results would have been different if controls were housed with locked wheels (see below). We have also recently addressed the question of wheel running effects on brain, but also blood, endocannabinoid levels. Besides the animal species (mice were used in the present study), our study differed from the former in several ways. Thus, our mice were allowed to run for 3h/day during 8 days and endocannabinoid levels were estimated after 30 minutes of running. These estimations were compared to those conducted in controls that were housed with locked wheels for a similar duration. As shown in Figure 3.2A, neither circulating levels of AEA nor brain concentrations of this endocannabinoid displayed changes with running activity. The former observation contrasts with that obtained in the blood samples from exercising human subjects, this contrast being likely accounted for by species differences in the physiological and psychological salience of each exercise paradigm. The analysis of 2-AG levels led to a similar picture to that observed for AEA levels, except for a significant decrease in the hippocampus of mice allowed to run that was associated with a similar trend in the frontal cortex (Figure 3.2B). Although the real significance of these observations is unknown, it is interesting to note that mice exposed for 7 days to a daily episode of social stress displayed an opposite pattern to that measured in exercising mice. Thus, 2-AG levels were found to be increased in the hippocampus and the frontal cortex when measured 40 minutes after the last stress episode (Dubreucq et al., 2012). Interestingly, such a brain region-dependent pattern of reactivity of 2-AG has been observed in rats submitted repeatedly to another type of stress, namely restraint (Patel & Hillard, 2008). These results raise the hypothesis that the respective patterns of reactivity of 2-AG after running and stress are indicative of the psychological salience of each experimental condition. Other issues that merit consideration are the possibilities that wheel running triggers major changes on rat and mouse endocannabinoid levels in discrete brain regions and/or the need for different timings than those used in our study.

In addition to endocannabinoid levels, several studies have explored the effects of wheel running on CB1 receptors. These analyses have focused on the protein itself (number and affinity of the receptor protein) and/or on its functional characteristics. In the above-mentioned study aimed at exploring the impact of wheel running on rat brain endocannabinoid levels, CB1 receptor binding capacities in controls and exercising animals were also investigated. It was found that the number of binding sites for a CB1 receptor agonist increased in the hippocampus, but not in the frontal cortex (Hill et al., 2010). However, such an increase was counterbalanced by a reduction in receptor affinity for the ligand (as revealed by an increased dissociation constant). Besides, one study reported that 10 days of wheel running increased the gene expression of the CB1 receptor in the hippocampus of female mice (Wolf et al., 2010). Concerning CB1 receptor function, 8 days of wheel running hypersensitized the activation of hippocampal CB1 receptor-associated G proteins by a CB1 receptor agonist, thus suggesting that wheel running bears stimulatory consequences on hippocampal CB1 receptor function (Hill et al., 2010). Indeed, such a conclusion may extend to another brain region, namely the striatum. Thus, the ability of a CB1 receptor agonist to decrease striatal GABA neurotransmission, as assessed by the reduction in the frequency of spontaneous and miniature inhibitory postsynaptic currents (IPSCs), was reinforced after 15 days (but not 7 days) of wheel running activity (de Chiara et al., 2010). Such a sensitizing effect of wheel running was selective for GABAergic neurons as the CB1 receptor-mediated decrease in striatal glutamatergic transmission was of similar amplitude in control mice and in running mice. It is important to note here that providing control mice with a sweetened drinking solution led also to increased sensitivity of CB1 receptors on striatal GABAergic neurons (de Chiara et al., 2010). On the basis of this parallel, it has been proposed that it is the hedonic impact of running that may account for the above-mentioned electro-physiological results. It should be noted, however, that in this study the controls were housed in standard cages, leaving unaddressed the possibility that in addition to the hedonic impact of running, the enrichment of the environment contributed to the increased sensitivity of striatal CB1 receptors. This issue is particularly important in view of the promising observation that the desensitization of the inhibitory control exerted by CB1 receptors on striatal GABAergic transmission that is observed in animals exposed to three daily stress episodes was reported to be reversed by 1 day of housing with running wheels (Rossi et al., 2008).

Figure 3.2   Endocannabinoids and wheel running. (a) Plasma and brain tissue anandamide (AEA), and (b) 2-arachidonoylglycerol (2-AG) levels 30 minutes after the initiation of wheel running in animals housed for 8 days with free wheels or with locked wheels. Values are the mean ± S.E.M. of 3 (blood) and 8 (brain regions) samples/group. *P < 0.05 for the difference between sedentary and running subjects.

The tools to study the role of the endocannabinoid system in exercising subjects

The study of the functional role of any transmitter system involves classically the characterization of the responses elicited by direct agonists for the receptors of that transmitter. This strategy is often completed by the use of indirect receptor agonists, i.e., drugs acting respectively on the membrane transporters for that transmitter or on the synthesis/release of that transmitter. Such studies bring information on the potential impact of that transmitter system on a vast array of functions but do not inform on the tonic and permissive role that transmitter system may exert during particular situations. Pharmacological and genetic assays based respectively on the administration of antagonists or on the use of constitutive/conditional mutants for these targets are nowadays the routine procedures to gather that information. Studies on the relationships between the ECS and exercise have all focused on the tonic role the ECS may be endowed with during exercise, doing so by means of CB1 receptor antagonists, such as SR141716 (Rinaldi-Carmona et al., 1994) and AM251 (Gatley, Gifford, Volkow, Lan, & Makriyannis, 1996), or using constitutive mutants for the CB1 receptor wherein this receptor is absent from the whole body (Ledent et al., 1999; Marsicano et al., 2002; Zimmer, Zimmer, Hohmann, Herkenham, & Bonner, 1999). Because CB1 receptors are located on different neuronal (and glial) populations and expressed in numerous brain regions (see above), it is useful, however, to further define the population of CB1 receptors through which the ECS may play a role. This can be achieved using conditional mutants for CB1 receptors, and we will illustrate below the potential use of this strategy. The generation of these conditional mutations is based on the so-called “Cre/lox P” technique (Morozov, Kellendonk, Simpson, & Tronche, 2003). This technique is based on the ability of a recombinase protein (Cre) to excise any sequence of DNA flanked by short sequences called “loxP” sites. Given their small dimensions (34 base pairs), the introduction of loxP sites into the genome of mice flanking a given gene to generate “floxed” mutant mice does not change its expression. Therefore, “floxed” mice can be considered phenotypically wild-type mice. However, the presence of the Cre recombinase in specific cell types or tissues will lead to the specific excision of the “floxed” gene, generating conditional (i.e., spatially and/or temporally restricted) mutant mice. Thus, crossing mice bearing a Cre recombinase specific for a given cell population with mice hosting a CB1 gene sequence flanked by loxP sites (“floxed CB1”) generates two types of subjects in the progeny: (a) those bearing just the floxed CB1 gene sequence (phenotypically wild-type animals) and (b) those lacking the CB1 sequence in the cell populations expressing the Cre recombinase (conditional CB1 mutant animals). As an illustration, crossing mice that either carry or do not carry the Cre recombinase Nex with mice hosting a floxed CB1 gene generates wild-type controls (termed below Glu-CB1^+/+) and mice lacking CB1 receptors from glutamatergic neurons (termed below Glu-CB1^−/−; Monory et al., 2006). Using other mutant mice expressing a Cre recombinase specific to a cell population (Monory et al., 2007) allows further dissecting the functional role of CB1 receptors with respect to the cell population where these receptors are expressed. For example, the respective roles of CB1 receptors located on GABAergic neurons (the brain cells in which CB1 receptors are the most abundantly expressed; Marsicano & Lutz, 1999; Monory et al., 2006) and on glutamatergic cells (where CB1 are expressed to a low level; Marsicano & Lutz, 1999; Monory et al., 2006) have been recently documented with respect to the control of fasting-induced food intake (Bellochio et al., 2010) and long-term memory deficits triggered by THC (Puighermanal et al., 2009). As indicated above, investigations on the relationships between the ECS and exercise have used CB1 receptor antagonists and constitutive/conditional CB1 receptor mutants (see below). Undoubtedly, the use of conditional mutants for the CB1 receptor will expand in the near future to further define the role of the ECS during exercise. This is also true for other promising tools that have been generated recently, among which (a) selective inhibitors of the enzymes degrading the endocannabinoids or of the transporter that allows the cellular uptake of these lipids (Long et al., 2009; Piomelli, 2003; Schlosburg et al., 2010), and (b) mouse mutants for FAAH (Cravatt et al., 2001), DAGL (Gao et al., 2010; Tanimura et al., 2010), and MAGL (Chanda et al., 2010).

The endocannabinoid system and running performance

In 2004, it was reported that an acute pretreatment with the CB1 receptor antagonist AM251 before the active phase of the diurnal cycle increased in a dose-dependent manner the wheel running distance covered overnight by lean mice while decreasing that distance for the highest doses in obese mice (Zhou & Shearman, 2004). However, several years later, a study using the same antagonist reported that this drug, even if provided repeatedly, did not influence wheel running in rats (Hill et al., 2010). It should be noted here, however, that the latter study relied on the administration of the CB1 receptor antagonist at the beginning of the inactive phase of the diurnal cycle, i.e., 10 hours before the normal onset of the running episode. This opens the possibility that the antagonist was devoid of activity when the animals started their running activities. In a third work, it was documented that the administration of the CB1 receptor antagonist SR141716 2 hours after the onset of wheel running activity decreased the running distances in a dose-dependent manner (Keeney et al., 2008). This decrease, which was observed both in a control mouse line and in a mouse line selected for high wheel running activity, was accounted for by sex-dependent decreases in running duration and/or speed (Keeney et al., 2008). These three pharmacological studies thus lead to different conclusions, that heterogeneity being accounted for by their respective experimental settings (acute/repeated treatment, delay between treatment and running activity). Besides these experimental differences, one additional variable that differed between the studies was the duration of time provided to the animals to habituate to the wheels before pharmacological treatments. In the first and third studies, mice were provided with a 5- to 7-day training period beforehand, whereas in the second study rats received their daily antagonist administration on the first day of housing with wheels. In our hands, pretreatment with the CB1 receptor antagonist SR 141716 30 minutes before a 3-hour daily wheel running episode that started at the onset of the dark phase of the diurnal cycle actually decreased running activity in 7-day trained mice, a finding extended to mice given the opportunity to run for the first time (Dubreucq et al., 2013). Taken together with the data from the study from Keeney et al. (2008), our data strongly suggest that the ECS acts as a tonic stimulatory control of running activity, including when running activity has already been engaged. The use of CB1 mutant mice provides definitive evidence for this suggestion. Thus, in a study in which mice were given unrestricted access to running wheels, constitutive CB1 mutant mice displayed a 30–40% decrease in wheel running activity compared to their wild-type littermates (Dubreucq, Koehl, Abrous, Marsicano, & Chaouloff, 2010). A detailed analysis of the behaviors of CB1 mutant mice revealed that the negative consequence of CB1 receptor deletion on running activity was accounted for by decreased time spent running and decreased maximal velocity. Furthermore, wheel running differences between CB1 receptor mutants and wild-type littermates were already observed when mice began to engage in their running activity, i.e., during the dark phase of the diurnal cycle (Chaouloff, Dubreucq, Bellocchio, & Marsicano, 2011). In the present context, the following observations are noteworthy. First, this inhibitory consequence of CB1 receptor deletion on running performance is unlikely to be accounted for by an intrinsic impact on locomotor activity as CB1 receptor knock-out mice do not differ from their wild-type littermates when housed for 1 week in cages provided with horizontal and vertical actimeters (Chaouloff et al., 2011). Second, the inhibitory impact of CB1 receptor deletion on wheel running activity was also observed in mice provided with a restricted (3-h) daily access to the running wheels (Dubreucq et al., 2013).

Whenever a reduction in wheel running activity in SR141716-pretreated/treated animals is observed (see above), the amplitude of that reduction ranges between 30 and 60% of the activity measured in vehicle-injected animals. As already mentioned above, CB1 knock-out mice display a 30 to 40% decrease in wheel running activity, compared to wild-type animals. Taken together, these observations suggest that the ECS exerts only a partial tonic stimulatory control of wheel running activity. Where and how such a control is exerted is a question we have recently tried to address although (1) wheel running is a complex behavior (Sherwin, 1998), and (2) the ECS controls many functions in the CNS as well as in the periphery. Thus, in the CNS, the ECS exerts a tight stimulatory control of motoric behavior through the CB1 receptor-dependent modulation of the excitability of neuronal networks in the neocortex, the striatum, the cerebellum, and the spinal cord (El Manira & Kyriakatos, 2010). Accordingly, pharmacological antagonism or genetic silencing of CB1 receptors in either of these networks could have inhibitory consequences on running activity. The same holds true for the impact that the ECS exerts over metabolism through actions on food intake and energy expenditure (Pacher et al., 2006; Pagotto et al., 2006). Another hypothesis, which does not exclude the former ones, lies in the observation that rodents will lever-press to get access to running wheels (Belke & Wagner, 2005) and display conditioned place preference for brief and sustained wheel access (Greenwood et al., 2011; Lett, Grant, & Koh, 2001). These results are in keeping with the proposal that although wheel running is an extremely complex behavior, it can be considered a natural reward with high motivation roots (Sherwin, 1998). Because the ECS, through CB1 receptors, modulates in a positive manner rewarding and motivational processes (Maldonado et al., 2006), we recently suggested that the ECS might control running activity by acting on the naturally rewarding properties of running. Indeed, our most recent experiments using conditional CB1 mutant lines wherein CB1 receptors are deleted from selective neuronal or glial populations indicated that (1) the tonic control exerted by CB1 receptors on running performance is fully accounted for by CB1 receptors located on GABAergic neurons, (2) these receptors are located on GABAergic nerve terminals lying in the ventral tegmental area (VTA), a key area involved in reward-based motivation processes, and (3) the presence of CB1 receptors on GABAergic neurons prevents the negative after-effects of exercise on VTA dopaminergic activity (Dubreucq et al., 2013).

The endocannabinoid system and the emotional consequences of running

We have indicated above the importance of laboratory animal studies in the quest for the neurobiological mechanisms underlying the positive emotional effects of physical exercise in humans. However, this quest lies on a prerequisite, which is that these animal models of human physical activity have positive emotional effects. The literature on that particular aspect of exercise has yielded numerous, albeit contradictory, data. Indeed, part of that heterogeneity is accounted for by the respective experimental settings among which the conditions under which the sedentary/control animals are housed (see below).

Role of the endocannabinoid system in the emotional profile displayed by wheel running subjects

To our knowledge, only one study has explored the respective influences of wheel running and CB1 receptors on some aspects of emotionality (Dubreucq et al., 2010). In that study, male CB1 mutant mice were compared with their wild-type littermates for their behavioral responses in an open field and their cued fear memory after a 6-week housing with locked or free wheels. It was observed that neither the exploration of the center of the open field, an index of anxiety (see above), nor the peripheral locomotion in that test were sensitive to wheel running in the two genotypes (Dubreucq et al., 2010). This result confirms the above-mentioned difficulty of revealing the anxiolytic impact of wheel running, if any. On the other hand, examination of cued fear memory expression, which is an amygdala-dependent process (as opposed to contextual fear memory, which is hippocampus-dependent; Maren & Quirk, 2004), revealed a significant interaction between wheel running and CB1 receptor expression. Confirming previous observations, CB1 receptor deletion per se increased cued fear expression and delayed fear extinction during recall sessions, as assessed from freezing reactions to a tone previously associated with a shock during the conditioning session (Marsicano et al., 2002). Interestingly, wheel running, which was ineffective in wild-type animals, counteracted both the increased fear expression and the delayed extinction that are observed in sedentary CB1 mutant mice (Dubreucq et al., 2010). The observation that wheel running was ineffective on cued fear memory in wild-type animals contrasts with previous findings showing that wheel running increases contextual fear memory (Burghardt, Pasumarthi, Wilson, & Fadel, 2006; Greenwood, Strong, Foley, & Fleshner, 2009). This may indicate that wheel running bears differential effects on hippocampaland amygdala-dependent fear memories. In keeping with the inhibitory impact of wheel running on the hypersensitized fear expression displayed by CB1 receptor mutants, we recently wondered whether such an interaction between wheel running and CB1 receptor deletion would still be observed if CB1 receptors were selectively deleted from selective neuronal populations. As indicated above, the Cre/lox P technique has allowed the generation of CB1 receptor mutants wherein CB1 receptors are missing from selected neuronal populations, including cortical glutamatergic neurons (Bellocchio et al., 2010; Monory et al., 2006, 2007; Puighermanal et al., 2009). We thus housed male Glu-CB1^−/− and male Glu-CB1^+/+ mice for 3 weeks with either locked or free wheels, and then examined in both genotypes anxiety-related behaviors using the elevated plus-maze (see above), and cued fear memory expression and extinction during fear recall tests. In the elevated plus-maze, wheel running was found to exert hypolocomotor influences, as assessed by the number of visits to the closed (i.e., protected) arms of the maze (Figure 3.3A). On the other hand, wheel running counteracted the decrease in the percent time spent in the open (i.e., unprotected) arms of the maze that was displayed by the mutant mice (Figure 3.3B). Taken with the data mentioned above, this set of observations thus suggests that wheel running in mice does not bear intrinsic anxiolytic effects but may be able to reverse innate anxiety. Whether this counteracting effect of wheel running extends to environmental situations that further raise anxiety levels is an issue that surely deserves future investigation. When Glu-CB1^−/− and male Glu-CB1^+/+ mice were cued fear conditioned by a tone-shock pairing, fear expression responses to single tone exposure 1 to 3 days later revealed that wheel running had an influence in mutant, but not in wild-type mice (Figures 3.3C and 3.3D). Thus, wheel running decreased fear expression and accelerated fear extinction on recall sessions 2 and 3 in Glu-CB1^−/−, but not in Glu-CB1^+/+, mice. Such a differential effect of wheel running in the two genotypes closely resembles that measured in constitutive CB1 mutants, as opposed to their wild-type littermates (see above). This indicates that (1) wheel running controls in a tonic manner cued fear memory expression/extinction and (2) that such a control involves CB1 receptors located on cortical glutamatergic neurons. In turn, the former finding opens the promising hypothesis that physical exercise in humans may set back to normal levels abnormally high fear expression and/or delayed extinction (as observed, e.g., in patients suffering post-traumatic stress disorder). Our finding reinforces former evidence for a tight control of fear memory by the ECS (Marsicano et al., 2002); however, how such a control is exerted in exercising subjects remains to be explored.

Figure 3.3   CB1 receptors and behavioral consequences of wheel running. (a) Activity and (b) anxiety-related behavior in the elevated plus-maze of Glu-CB1+/+ and Glu-CB1−/− mice housed for 3 weeks with either locked or free wheels. Cued fear memory during recall sessions in Glu-CB1+/+ and Glu-CB1−/−mice housed for 3 weeks with locked wheels (c) or with free wheels (d). Values are the mean ± S.E.M. of 13–15 mice/group. + P < 0.05 for the main influence of the housing condition in the two genotypes. * at least P < 0.05 for the difference between sedentary and running subjects.

The endocannabinoid system and the neurogenic impact of wheel running

Adult neurogenesis, i.e., the formation of new neurons, occurs in the hippocampus where it is highly sensitive to wheel running (Ernst, Olson, Pinel, Lam, & Christie, 2006; van Praag, 2009). Thus, housing with running wheels increases neurogenesis in the dentate gyrus, an effect already observed after 1 week of wheel running. However, such a stimulatory impact of exercise is not specific to running as environmental enrichments also share that capacity (Kempermann, Kuhn, & Gage, 1997). As already underlined above with regard to the emotional effects of wheel running, the neurogenic impact of an enrichment in the environment raises the question of the intrinsic consequence of the presence of the wheel. Indeed, a recent study wherein mice were housed in standard cages or in cages complemented with either locked wheels or free wheels revealed the following. Housing with wheels, whether locked or unlocked, stimulated the proliferative phase of the neurogenic process, compared with standard housing (Bednarczyk et al., 2010). However, when focusing on later stages of the neurogenic process (i.e., survival, differentiation, maturation), mice housed with free wheels displayed increased neurogenesis, as compared to the two other mouse groups. This indicates that exercise has an intrinsic stimulatory influence on the last phases of the neurogenic process only.

Several studies have analyzed whether the ECS contributes to the stimulatory effects of wheel running on neurogenesis. In the aforementioned study in which a CB1 receptor antagonist was administered daily for 8 days to male rats, it was observed that such a treatment prevented the stimulatory influence of wheel running on hippocampal cell proliferation (Hill et al., 2010). Supporting this, one study using female mice reported that wheel running elicited cell proliferation in the dentate gyrus, as compared to standard housing, a change that was absent in CB1 mutant mice (Wolf et al., 2010). On the other hand, another study using male CB1 mutant and wild-type mice reported that although CB1 receptor mutation decreased neurogenesis, in line with past results (Aguado et al., 2006; Jin et al., 2004) and running performance (see above), the percent stimulatory influence of wheel running on neurogenesis was independent from the mouse genotype (Dubreucq et al., 2010). Taken together, these results indicate the need for future experiments to delineate the role of CB1 receptors on exercise-induced neurogenesis. In keeping with the contradictory data that have emerged on that subject, these experiments will need to detail the respective roles of CB1 receptors at each stage of the neurogenic process and compare the effects of wheel running to standard housing and housing with locked wheels.