Previous research has shown that inhibiting AEA catabolism mitigates neuroinflammation in response to various proinflammatory stimuli [24, 25, 26, 27, 28] and prevents neurodegeneration in an experimental animal model of frontotemporal dementia [29]. Within the context of amyloidosis, genetic ablation of FAAH significantly reduced Aβ production and accumulation, reactive astrogliosis, and neurodegeneration, while concurrently improving synaptic and cognitive function in 5XFAD transgenic animals [12, 13]. These collective findings implicate that FAAH could be viewed as a promising therapeutic target for inflammatory and neurodegenerative processes linked to AD. Nevertheless, the molecular mechanisms underlying the beneficial effects of AEA degradation inhibition on amyloid‐related neuroinflammation and neurodegeneration remain largely elusive.

The results of this study provide compelling evidence that inhibition of FAAH exerts protective effects against cognitive decline and amyloid pathology in mouse models of AD through multiple mechanisms converging on the modulation of Aβ production.

A key finding here described was that inhibiting FAAH activity‐both genetically and pharmacologically‐led to a strong downregulation of BACE1 in the hippocampus of APP/PS1 and Tg2576 mice. BACE1 is a critical enzyme in the amyloidogenic pathway, as it catalyzes the initial cleavage of APP to generate the pathogenic Aβ peptides (Fig. 9). The suppression of BACE1 expression was associated with increased DNA methylation of its promoter region, suggesting an epigenetic mechanism by which elevated AEA levels, resulting from FAAH inhibition, may modulate gene expression.

In addition to regulating BACE1, FAAH inhibition also shifted APP processing towards the non‐amyloidogenic pathway. Indeed, URB597 treatment significantly increased the expression of ADAM9, an α‐secretase that cleaves APP within the Aβ domain, precluding its amyloidogenic processing. Furthermore, URB597 enhanced the levels of PSEN2, a component of the γ‐secretase complex that participates in the non‐amyloidogenic cleavage of APP. These changes in APP‐cleaving enzyme expression were reflected in the increased production of the soluble APP‐α fragment, the product of the non‐amyloidogenic pathway, and reduced levels of soluble APP‐β, the amyloidogenic counterpart.

In vitro experiments using primary neurons isolated from Tg2576 mice further confirmed the BACE1‐suppressing effects of URB597. This CB₁‐dependent downregulation of BACE1 was accompanied by a concomitant decrease in Aβ₄₂ levels, the most pathogenic Aβ species.

Collectively, these findings suggest that FAAH inhibition modulates multiple aspects of APP processing, favoring the non‐amyloidogenic pathway and reducing the production of Aβ peptides. The epigenetic regulation of BACE1 expression and the CB₁‐mediated suppression of this enzyme appear to be a couple of key mechanisms underlying the anti‐amyloidogenic effects of FAAH inhibition. Notably, our previous research has demonstrated that AEA can indeed regulate gene expression through epigenetic mechanisms, specifically inducing promoter DNA hypermethylation in keratinocytes via CB₁ receptor activation, leading to transcriptional downregulation of differentiation‐related genes [30].

Our results indicated no significant alterations in the hippocampal expression of NAPE‐PLD between AD‐like mice and wild‐type counterparts. Similarly, FAAH expression levels showed no statistically significant changes, except for a minor, yet non‐significant, upregulation observed in Tg2576 mice compared to wild‐type controls. These findings suggest that the overall expression of these enzymes in the hippocampus does not substantially differ between AD‐like models and wild‐type animals. However, it is worth noting that when we examined FAAH expression specifically in isolated neurons from Tg2576 mice, we observed a 2.8‐fold increase in FAAH mRNA levels compared to neurons isolated from WT mice. This localized upregulation in FAAH expression may indicate a neuron‐specific dysregulation of the endocannabinoid system in the Tg2576 model, which might not be detectable when analyzing the hippocampus as a whole. This finding aligns with previous reports suggesting that alterations in endocannabinoid signaling may be restricted to specific cellular populations or brain regions in neurodegenerative conditions [8, 10, 31, 32].

Tg2576 and APP/PS1 mouse models both express mutant forms of human APP with the Swedish mutations and exhibit age‐dependent accumulation of Aβ peptides, amyloid plaques, neuroinflammation, oxidative stress, and cognitive deficits [33, 34, 35]. However, there are significant differences between these models in terms of amyloid pathology and the onset of cognitive deficits: Tg2576 mice overexpress the human APP gene with the Swedish mutation [17, 36] and develop amyloid plaques and cognitive deficits over a longer time scale, making them suitable for studying the physiological aspects and the cognitive evolution of AD pathogenesis [17, 36, 37, 38]. In line with several findings describing a very late onset of plaque deposition in the Tg2576 AD‐like mouse model [39, 40, 41], we detected amyloid plaque formation in 11‐month‐old Tg2576 mice only in the midbrain region. It is worth noting that FAAH pharmacological inhibition demonstrated efficacy also against amyloid burden, reducing both density and size of Aβ plaques. Aβ deposition in the midbrain has been, for instance, described in 9‐month‐old 5xFAD mice [42], and recent analysis of neuropathological data, including a large amount of quantitative neuroimaging studies, corroborates the view that the temporal evolution of Aβ deposition can follow a gradient oriented to cerebral areas showing high metabolic request, such as association cortices and the midbrain [43]. Interestingly, the midbrain area is also the focus of a line of investigation ascribing to early degeneration of midbrain dopamine (DA) neurons an important role in AD pathogenesis [44], and a recent study reported that evoked DA release from midbrain neurons reduced Aβ aggregation via a neprilysin‐dependent mechanism [45].

In addition to the human APP, APP/PS1 mice also express the mutant presenilin‐1 (PS1) gene. The presence of mutant PS1 accelerates APP metabolism (Fig. 9), leading to faster and higher Aβ peptide production and more extensive plaque deposition compared to Tg2576 mice [46]. This makes APP/PS1 mice particularly useful for studying the biochemical aspects of APP metabolism and plaque deposition.

Notably, genetic deletion of FAAH slowed down the development of recognition deficits in APP/PS1 mice. While only 6‐month‐old APP/PS1 animals, but not older ones, could recognize the previously seen partner or the previous location of an object, 12‐month‐old double mutant APP/PS1‐FAAH‐KO mice retained these abilities, like wild‐type controls. This suggests that FAAH deletion can delay cognitive decline in this AD model. Previous studies have shown that APP/PS1 animals exhibit spatial learning deficits at 8–10 months of age [47, 48], and that 3‐month‐old mice can show memory impairment in partner recognition tests [49], similarly to the results here described. However, our study highlights the importance of using different age groups when targeting such a complex pathological process, such as the development of AD‐like pathology in the mouse brain. Indeed, similar studies using only one age group have often shown conflicting results. For example, genetic deletion of FAAH in 5xFAD mice ameliorated synaptic dysfunction in 5‐month‐old mice [13], whereas FAAH deletion did not protect 6‐month‐old PS19 mice from neuroinflammation or cognitive decline [14]. Moreover, we observed that FAAH‐ΚΟ mice had some recognition deficits. Although an early study reported enhanced acquisition in an aversive (but not appetitive) learning model [50], subsequent studies suggested that elevated AEA levels impair long‐term potentiation (LTP) and memory [51] and leads to increased pro‐inflammatory glial activity [52]. We hypothesize that the absence of FAAH may disrupt brain function in wild‐type animals, but in the context of the ongoing AD‐like pathology in the brains of APP/PS1 animals, its net effect appears to be positive.

Chronic inhibition of FAAH significantly improved cognitive performance in Tg2576 mice. Behavioral assessment evaluating various hippocampal‐dependent functions revealed that FAAH inhibition, via URB597 intranasal delivery, ameliorated memory deficits in both short‐term and long‐term paradigms. Notably, these improvements were more pronounced with prolonged treatment durations, indicating a time‐dependent efficacy. These findings align with previous studies highlighting the ability of URB597 to modify distinct forms of synaptic plasticity associated with hippocampal functions [11, 53, 54]. It is plausible that chronic enhancement of the AEA tone led to a significant reduction in memory deficits by normalizing basal synaptic transmission and LTP in the hippocampus of amyloidogenic mice. This hypothesis is supported by Ruiz‐Pérez et al. [13], who reported significant improvements in synaptic plasticity in another murine model of AD with genetic FAAH deletion. Collectively, these results suggest that FAAH inhibition consistently provides beneficial effects to synaptic function across various AD models.

Although our findings broadly corroborate the therapeutic potential of FAAH inhibition, some discrepancies with previous studies merit further examination. For example, the extent of cognitive improvement observed in our study utilizing URB597 was notably greater than that reported by others [12]. Variations in experimental design, including the mouse model (Tg2576 versus 5xFAD), route of administration (intranasal versus intraperitoneal), duration (5 months versus 12 days), and dosage (5 versus 3 mg·kg⁻¹) of treatment, may explain these differences. Conversely, our results in APP/PS1‐FAAH‐KO mice appear to align with findings in 5xFAD mice lacking FAAH, particularly regarding the recovery of cognitive impairment [12, 13].