In this work we describe several beneficial effects of the prolonged oral treatment with two cannabinoid agonists with different pharmacological profiles. Both cannabinoids were effective at decreasing the inflammatory parameters and Aβ levels. However, it was the CB2 selective agonist JWH that was able to prevent cognitive deficits and glucose metabolism reduction.
Our results differ from a recent work  reporting variable effects on water maze performance and fear conditioning, no changes in Aβ levels or plaque burden following subchronic treatment with the CB1/CB2 mixed agonist HU-210. Methodological differences may account for the results, since they used male and female double APP23/PS1 transgenic mice, at 4 (young) or 6-9 weeks of age, and a very low dose of the cannabinoid agonist (10 or 50 μg/kg) injected twice daily .
CB1 activation induces psychoactive effects. In fact mixed CB1/CB2 and CB1 selective agonists after acute administration decrease motor activity and impair learning and memory [37–39], although at higher doses than the one we used in this study (0.2 mg/kg/day). However, chronic administration of those drugs induces tolerance to their acute effects in different behavioural tests [40, 41]. Therefore, the fact that WIN did not affect learning in the novel object recognition test in wild type mice might suggest tolerance after prolonged administration. Similarly WIN was without effect on recognition memory of Tg APP mice. In contrast, CB2 selective agonists, such as JWH, are devoid of psychoactivity after acute administration and does not alter motor activity [, Martín-Moreno et al., in preparation] after systemic administration or object recognition memory following its intrahippocampal injection . After prolonged administration to wild type mice JWH neither altered learning nor memory in the present work. Interestingly this compound effectively counteracted the cognitive impairment of Tg APP mice.
The brain is the organ with the highest glucose consumption, which is believed to be coupled to neural activity. The reduction in brain glucose uptake has been repeatedly demonstrated in AD patients [31, 44], in particular in regions involved in memory, and it is highly correlated with cognitive deficits. Prolonged oral WIN administration reduced glucose uptake, as measured by PET 18F-DG, in wild type mice in cortical regions and hippocampus. This result may be deleterious, in spite of the observation of no memory impairment in the cognitive test selected. In fact, previous autoradiographic work has reported either normal 3H-DG uptake in hippocampus or a decrease, depending on the dose of WIN  acutely administered. Decreased 18F-DG uptake in Tg APP in the present work is in essential agreement with previous reports, either by autoradiographic techniques  or PET . Notably JWH administration for 4 months did not alter glucose uptake in wild type mice, while it completely abrogated the reduction observed in Tg APP. Brain activity in general, and cognition in particular rely on glucose metabolism , therefore the effects of the CB2 agonist on both glucose utilization and recognition memory are of therapeutic interest.
Inflammation has several drawbacks including learning and memory impairment , in particular during ageing [49, 50]. The compounds under study behaved as anti-inflammatory agents, in agreement with previous reports [11, 12, 20, 25, 49–51]. Microglial activation, but not astrogliosis, was observed in 11 month old Tg APP mice. Previous works have shown prominent reactive astrogliosis and increased GFAP expression in transgenic mice, that appears to be age-dependent and related with disease progression [26, 28, 52], albeit being restricted to plaques. The lack of astrogliosis and changes in GFAP expression may be explained by the absence of plaques in the mice model at this age. This result is in agreement with the detection of GFAP mRNA, as assessed by non-radioactive in situ hybridization, in reactive astrocytes in close proximity with Aβ plaques at 14 months of age, but not before, when plaques were absent . Indeed we found no plaques with glial associated cells, although there was a significant enhancement in microglial cell density in Tg APP mice. Continuous JWH treatment for 4 months normalized this parameter, but WIN was ineffective. In different contexts (eg lesions) both astrocytes and microglia could be engaged in inflammation. However in light of those results we can ascribe the increase in inflammatory mediators to microglial activation, given that there was no overt astrogliosis. COX-2 protein levels and TNF-α mRNA expression is increased in AD and its transgenic model [54, 55]. Both cannabinoids significantly decreased COX-2 and TNF-α, as expected, since both compounds share CB2 receptor activation, as shown by the down-regulation of its expression in Tg APP mice. It should be noted that although CB2 receptors could be expressed by some neurons  they are mainly expressed by microglial cells [6, 9–11], and are involved in the modulation of several inflammatory mediators [12, 19–21, 25]. We did expect an increase in CB2 receptor expression in the transgenic model given that Aβ addition to microglial cultures enhance it . However, in Tg APP mice we did not observe CB2 co-localization with Iba-1, which contrasts with the microglial co-expression in AD brain [13, 33]. We neither found an increase by Western blotting, in agreement with our previous results in AD patients or Aβ-injected rats . This supports the notion that at this age there is an ongoing glial activation of low magnitude in Tg APP and that cannabinoids down-modulate this response.
Aβ removal is considered a therapeutic strategy in AD, promoted either by vaccination [57, 58] or by enhancing its clearance towards the periphery [59, 60]. One of the most interesting findings of the present work is the Aβ lowering ability of both cannabinoids, which we report for the first time. Prolonged oral JWH treatment decreased Aβ1-40 levels in brain and both cannabinoids decreased the more amyloidogenic fragment, Aβ1-42. Given that the drugs did not alter Aβ release we speculated that APP cleavage was not altered, and therefore studied whether cannabinoids changed the peptide transport in vitro. Rat choroid plexus expressed CB1, in agreement with , and also CB2 receptor protein, making feasible their activation by the drugs under study. Cannabinoids favored Aβ transport, that was mainly observed at shorter times (1-3 h) compared to control experiments. This interesting effect merits further study.
At variance with those effects oral treatment with WIN, but not JWH, normalized the levels of GSK3-β in Tg APP mice. Neurofibrillary tangles (NFTs), resulting from an abnormal phosphorylation of microtubule-associated tau proteins, represent a key pathological hallmark of Alzheimer's brain. GSK3-β is the kinase mainly responsible for tau hyperphosphorylation, therefore inhibiting its activity is considered of therapeutic interest. The effect of WIN after prolonged oral administration is in accordance with reports showing a CB1 dependent increase in GSK3-β phosphorylation in cultured cells , in brain after acute cannabinoid agonist administration , and with a reduction in tau phosphorylation . Neurofibrillary tangles are intraneuronal elements and neurons are in general devoid of CB2 receptors. Therefore the effects of WIN on GSK3-β, which were not mimicked by the CB2 selective agonist JWH, might be explained by its interaction with CB1 receptors in neurons.
Given that our previous work had shown that cannabinoids were preventive against the Aβ effects, both in vitro and in vivo [12, 20] we decided to start the continuous oral treatment at 7 months of age. At this time Tg APP mice do not have plaques and show normal learning and memory compared to wild type mice. Nevertheless, the treatment ended at 11 months of age when Tg APP begin to show memory disruption. According to our results the prolonged drug treatment decreased microglial activation of Tg APP mice along several inflammatory mediators, which were increased. However, ageing alters microglial responsiveness (eg to Aβ production and deposition), which is highly dynamic and context dependent . Therefore a potential caveat of our results is that they may not be applicable to aged pathological microglia as occurs in severe AD. However, a preventive treatment at very early stages of the disease may be feasible and beneficial as has been shown with the anti-inflammatory trifusal, both in amnestic mild cognitively impaired patients  and in Tg APP mice .
Over the last decade important findings on the involvement of the endocannabinoid system in AD has been gathered. Indeed, in AD brain there is increased expression of CB2 receptors in microglia and of fatty acid amide hydrolase, the enzyme responsible for anandamide degradation, in astrocytes around plaques . However CB1 localization is markedly altered and its protein expression and functionality diminished . Furthermore, molecular rearrangements in different endocannabinoid system elements suggest that 2-AG signaling is increased, possibly contributing to synaptic failure in AD , while anandamide levels are decreased and are inversely correlated with Aβ levels . Interestingly, the CB2 receptor expression has been reported to be increased both in the brain of AD patients  and in peripheral blood, where a significant correlation was found with the dementia score . Finally, a CB2 PET radiotracer is accumulated in brains showing neuroinflammation (eg LPS injected and Tg APP/PS1 mice; ). These latter results suggest the importance of CB2 receptor as a biomarker of the neurologic disease, but also as a therapeutic target. CB2 receptor increased expression in AD appears to be a consequence of microglia activation, but more importantly they render microglia susceptible to cannabinoid modulation, decreasing the generation cytotoxic molecules and inhibiting microglial activation, while promoting its migratory activity [10, 11, 20].