In this study we have found substantial differences between THC and CBD treatment, supporting the previously reported disruption of memory formation by THC . We did not find a suggestive association between CBD-mediated CB1 activity, learning performance in the water maze, and adult hippocampal neurogenesis. THC impaired cognitive and enhanced locomotor function but had no effect on neurogenesis, when given chronically. The learning phenotype in the Morris water maze did not correlate with the neurogenesis phenotype. Taken together, both THC and CBD effects on this type of hippocampus-dependent function cannot be linked to adult neurogenesis in a straightforward way. This discrepancy between functional and cellular hippocampal features had not yet been shown for THC or CBD, but the phenomenon of divergence between learning paradigms and neurogenesis is known from other studies (reviewed in ). When neurogenesis was blocked by focal x-radiation the mice that had been exposed to an enriched environment still performed better in the Morris water maze than the mice housed in standard cages. Since enriched environment enhances the survival of newly generated neurons, the investigators claimed separate effects of the enriched environment on neurogenesis and on spatial learning . Other groups showed that hippocampal irradiation immediately before the test had no effect, while irradiation days before the test impaired long-term memory in the water maze, indicative of a critical time window .
We showed that CBD increased neurogenesis at the survival stage 4 weeks after BrdU injection. Similar to the neuronal survival effect of CBD it has been reported, that the synthetic non-selective cannabinoid agonist WIN-55,212-2 restored the physiologically decreased levels of adult neurogenesis in aged rats . In the current study we have studied young adult mice, but it might be worthwhile to investigate in further studies the CBD effect in aged animals. CBD is known to be beneficial in schizophrenia or schizophrenic-like behaviour , where patients show a decrease in hippocampal volume and neurogenesis might be impaired [17, 46]. CBD has some antipsychotic properties . Moreover, smoking some strains of cannabis containing relatively more CBD, in addition to THC, appears to be more protective against the psychotic symptoms induced by THC alone .
A study by Boucher et al. has shown that THC impaired spatial memory and reversal learning, even in animals that received a THC pretreatment, indicating that although tolerance to the effects of THC on neuronal activity in the prefrontal cortex was reported, cannabinoid-induced memory impairment in these animals persisted . Although we could only test our animals at one time point, including the information of tolerance resistance to THC from the reference mentioned above makes us confident that no acute or tolerance effects were present during our testing phase.
Taken together, the findings suggest diverse effects of the cannabinoid system on memory and cellular plasticity. These effects cannot be plainly categorized into impairing or enhancing effects of cannabinoid activation or deactivation . The same might be true for the finding that THC increased the performance in the rotarod test. CB1 activation in the cerebellum by intra-cerebellar THC injection led to locomotor deficits . Moreover, stimulation of cerebellar CB1 receptors with the agonists CP55,940 and HU-210 impaired rotarod performance . THC injected intraperitoneally on the other hand failed to provoke motor coordination disturbances in wild type B6/CBA mice . The route of administration seems to be a key difference between this one and the other studies. CB1 receptors were activated in all relevant brain regions and the local concentration of THC in a given brain structure was lower than when administered directly into the cerebellum . In our study the mice took up the THC via the food, which led to improved rotarod performance. In the light of therapeutically targeting locomotor dysfunction with cannabinoids this finding might be notable. The therapeutic potential of the cannabinoids was also investigated in neurological diseases such as multiple sclerosis, Gilles de La Tourette syndrome, Parkinson and Huntington disease that all include locomotor disabilities . Although the efficacy was not always clearly established, the undesirable effects observed were generally mild and well tolerated . The drugs used to treat symptoms of multiple sclerosis (Sativex, contains THC and CBD) failed to change the neuropathological hallmarks of the disease. Patients reported only minor changes in memory loss, while improvements in locomotor and spasticity and neuropathic pain were dominant . Experiments with THC and CBD in different concentrations might help to unravel the complex pattern of such treatments and should, as our results suggest, include measures of adult neurogenesis.
The neurogenic effect of CBD was not found in CB1-/- animals. Although CBD has low affinity to CB1 and its effects are often mediated via non-CB receptors (e.g. the vanilloid receptor) at least three other studies support that cannabidiol effects were CB1 receptor-dependent [19, 56, 57]. It might not be the only or usual mode of action, but with regard to enhanced adult hippocampal neurogenesis, CBD at least partially acts through the CB1 receptor. This result prompted us to investigate CB1 dependent regulation of neurogenesis utilizing CB1 receptor knock out animals as well as the CB1 receptor antagonist AM251 in Nestin-GFP-reporter mice.
In female CB1-/- mice on a C57BL/6 background we found increased proliferation 24 h after BrdU injection and decreased net-neurogenesis (7 days and 4 weeks after BrdU injection). Jin and colleagues reported impaired progenitor cell proliferation in CB1-/- mice but found contradicting increases with pharmacological CB1 antagonists SR141716A and AM251 [30, 58]. We did not observed such discrepancy. The time point of analysis in the Jin et al. study was 3 days after BrdU injection. Thus, they might not have detected the most acute effects. Using the same compound AM251 on wild type mice, we got different results at 7 days after BrdU injection. Although we observed an increase of BrdU-labeled cells at 1 h and 24 h, which would be in line with the findings by Jin et al., we found a decrease in BrdU-labeled cells at 48 h and 7 days. When phenotyped, DCX-positive cells accounted for the increase in proliferation at early time points, but at late time points fewer DCX- and more Nestin-positive cells were present indicating that maturation was impaired at the DCX-stage. This supports our data, that CB1 stimulation or blockage had different effects on neuronal progenitor proliferation and differentiation or maturation. The same pro-proliferative effect of AM251 at 24 h after BrdU injection have also been observed by Hill et al. in rats .
One notable difference between the study of Jin et al. and other studies (including ours) was their use of a CB1-/- strain bred onto the CD1 background . We have previously shown that CD1 show a very unusual pattern of baseline adult neurogenesis. Despite lower levels of proliferation compared to C57BL/6 they actually achieve high levels of net neurogenesis since survival exceeds any other strain investigated so far . Another difference between the studies might have been the use of male vs. female mice since a recent report demonstrated differences in CB1 receptor abundance in the hippocampus between female and male mice . Unfortunately Jin et al. did not report the gender examined in their studies. On the other hand, receptor abundance per se does not allow strong conclusions about receptor activity.
In addition, we show that the time point of measurement is critical when assessing the effects of the antagonists. Our findings imply, that CB1 receptor activity would increase proliferation of type-1/2a, reduce proliferation of type-2b/3 but accelerate maturation from these cells and lead to a net reduction of adult neurogenesis. Consequently, the increase observed in the Jin et al. study after 3 days of AM251 in parallel to 3 days of BrdU is likely to actually reflect a mix of increases and decreases, which can only be untangled with a different BrdU injection protocol and a distinction of the different precursor cell types.
CB1 receptors are expressed in the course of neuronal development but they are present on all precursor cells, beginning with the radial glia-like type-1 cells . CB1 expression appears to increase with differentiation, an observation that has also been made in embryonic cortical development . Together with our previous data on wild type mice  these data indicate that CB1 is expressed by cells that are primarily affected by activity-dependent regulation of adult hippocampal neurogenesis. We could consequently show that this type of regulation is impaired, if the CB1 receptor is absent. Keeney at al. have shown that the CB1 antagonist Rimonabant (SR141716) decreased running activity in C57Bl/6 female mice when injected for 9 consecutive days at the peak of running . The situation in the knock out animal in our study is different, since CB1 is absent constitutively and not only at the peak of running like in the Keeney et al. study. It is also notable that SR141716 has different effects on neurogenesis than the absence of CB1 . We measured running performance as the distance run per day for 10 consecutive days. As long as running the same distance is indicative of a similar stimulus for neurogenesis, the conditions should have been the same for CB1-/- and wild type mice. In the hippocampus of wild type mice we found an upregulation of CB1 receptor mRNA in the ENR and RUN mice along with an increase in Nestin mRNA only in the RUN paradigm. This is in line with studies reporting an increase in density of CB1 receptors in the hippocampus after voluntary wheel running. When AM251 was administered, activity-induced neurogenesis was impaired . This result also supports our findings that activity-induced neurogenesis is absent in CB1-/- mice. In contrast, a study using male mice that ran over a period of 6 weeks, CB1-/- animals covered less distance but showed greater numbers of DCX-expressing cells in the dentate gyrus indicating that a running-phenotype can be discovered after a prolonged running period . Another study reported that CB1 receptor sensitivity in the striatum increased after voluntary wheel running . In the light of therapeutic interventions targeting the cannabinoid system, increasing the receptor by simple running might be of interest.
The putative contribution of new neurons to hippocampal function has recently become increasingly clearer. Neurogenesis and specific aspects of learning (temporal separation, contextual integration, flexibility of relearning, and integration of novelty) [34, 68, 69] have been linked and a role in affective behavior has been described [70, 71]. Jiang and colleagues have suggested that the CB1-mediated effects of HU210 on adult neurogenesis might have anxiolytic and anti-depressant-like consequences . It might thus be that the cannabinoid-dependent regulation of adult neurogenesis is more relevant for the emotional than for the cognitive aspects of hippocampal function.