CBD and THC have been the focus of many studies by the scientific community and clinicians alike [35]. The predominant cannabinoids in hemp, however, are THCA and CBDA [36]. CBDA is known to cross the blood–brain barrier and to penetrate the brain [37]. In support of its therapeutic benefits, CBDA application in animal models has been shown to have anti-hyperalgesia [19,20], anti-seizure [21], anti-inflammatory [22], and anti-emetic properties [38]. CBDA has also been shown to rescue object and spatial cognitive function and memory deficits in Aβ₄₂-treated mice [26]. CBDA is, however, the chemical precursor to CBD [21,28,29], which has also been shown to have many therapeutic properties. The possible effects mediated by this metabolite therefore cannot be dismissed. Long-term treatment with CBD has also been shown to prevent the development of social recognition memory deficits and improve object recognition in APP/PS1 mice [39,40]. CBDA is known to have increased oral absorption compared to CBD, with a serum concentration 30 times higher than that of CBD following oral doses of 2 and 4 m/kg [41].

Here, we have shown for the first time that the deficit in LTP produced by acute application of Aβ₄₂ (amyloid-derived diffusible ligands) [11,42,43] can be reduced by prior treatment with acute application of CBDA. We have also shown that acute CBDA did not significantly alter the level of LTP or PTP, similar to our previous observations with CBD [11]. We used only one concentration of CBDA (10 μM) and therefore cannot speculate as to the potential dose dependency of the effect. In a previous study using the same model, we found that the protective effects of CBD at the same concentration were reversed by the PPARγ antagonist GW9662 [11]. Here, we also demonstrated a similar reduction in the acute protective effects of CBDA against acute Aβ when PPARγ was inhibited. As CBDA has also been shown to activate PPARγ [44], the acute protective effects mediated here are likely mediated via this signalling cascade. The PPARγ nuclear receptor in the CNS is expressed in low levels, increasing under inflammatory conditions. PPARγ activation causes inhibition of NFkB to decrease expression of inflammatory genes [45]. In addition, CBDA has been shown to inhibit the expression and activity of cyclooxygenase 2 (COX-2) [25,46], further enhancing its anti-inflammatory effects. Most importantly, CBDA has been shown to be a powerful antioxidant comparable to vitamin E [24].

While we demonstrated the effects of CBDA in vitro against Aβ₄₂, the next step was to examine possible effects of CBDA in both control WT mice and in a transgenic model of AD (APPswe/PS1dE9; APP/PS1). There were no data available on the chronic effects of CBDA on synaptic transmission in either WT or APP/PS1 mice. In agreement with our previous report, LTP in the CA1 region in slices from APP/PS1 mice was depressed compared to levels in slices from vehicle-treated non-transgenic littermates [13]. While there was a slight trend towards increased LTP levels in slices from non-transgenic WT littermates with increasing doses of CBDA, these changes were not significant compared to those recorded in slices from vehicle-treated mice. This suggests that CBDA does not alter synaptic LTP and is not detrimental to plasticity in WT mice. Chronic treatment of APP/PS1 mice with CBDA at all concentrations, however, reversed the deficit we observed in LTP recorded in vehicle-treated APP/PS1 mice. This could be due to the alterations we observed in proteins involved in synaptogenesis and synaptic function (see the proteomics discussion).

While our focus was on the effects of CBDA on LTP, the possible overall change in excitability was also of interest. CBD and CBDA have been proposed to act as antiseizure agents [21] and could possibly have altered neuronal excitability. Short-term changes in excitability were monitored using PPF at a range of inter-pulse intervals (10–150 ms). We observed a deficit in PPF in slices from vehicle-treated APP/PS1 mice in contrast to other studies that have shown little change in paired-pulse ratio in the hippocampal CA1 region of APP/PS1 mice [47]. We found that CBDA could reverse this deficit. It has been proposed that in aging synapses there is an increase in resting presynaptic calcium following Schaffer collateral stimulation, accompanied by a decrease in PPF [48]. Cannabinoids have been reported to inhibit T-type calcium channels, with CBDA having a higher efficacy than CBD, which could influence neurotransmitter release [49]. The reduction in PPF in slices from APP/PS1 mice may suggest higher levels of residual presynaptic calcium related to increased tonic release. Alternatively, the previously reported decrease in L-type channel density in APP/PS1 mice may be involved [50]. The effects we observed are highly unlikely to be due to residual CBDA present in slices, as the experiments were conducted 48 to 72 h following the last IP injection. The half-life of CBDA has been reported to be less than 4 h in mice [37].

In a study of hippocampal autapses, CBDA was found to have little effect on neuronal cannabinoid receptors. CBDA did, however, increase calcium in 2% of dorsal root ganglia suggesting it may act via a TRPV1 activation [51]. If CBDA was metabolised to CBD in vivo, then it could influence depolarisation-induced suppression of excitation [52], consistent with CBD acting as a negative allosteric modulator (NAM) at CB1 receptors [53]. In a study on the binding properties of CBDA, it was shown to improve the formation of heteromers of CB1R and CB2R, while CBD decreased affinity. In addition, CBDA affinity was in the micromolar range in radioligand-based assays using either CB1R or CB2R-containing membranes, higher than CBD. It was also suggested that CBDA may act as an inverse agonist or NAM at both CB1R and CB2R [54]. If CBDA was decarboxylated to CBD [55], the latter could reduce the uptake of anandamide to influence the ECB system via inhibition of fatty acid amide hydrolase, but this would require a high concentration of CBD since the IC₅₀ in rat tissue is 15.2 μM [56]. While inhibition of anandamide breakdown has been shown to restore LTP in a model of osteoarthritis via activation of CB1R [57], others have shown that elevation of endogenous anandamide impaired LTP [58]. The alterations in PPF and LTP we observed were likely due to the major alterations in proteins associated with mitochondrial function and synaptogenesis, many of which were rescued by CBDA treatment (see the proteomics discussion). We investigated the possibility that CBDA treatment could reduce the level of either soluble or aggregated Aβ. The levels of soluble Aβ₄₀ were similar in CBDA-treated and vehicle-treated APP/PS1 mice (see Supplementary Materials). Reported levels of Aβ₄₀ and Aβ₄₂ vary across the literature; a previous report shows lower levels of soluble Aβ₄₀ and Aβ₄₂ at 8–9 months in APPswe/PS1dE9 brain extract [59], while another study reported a much higher value for aggregated Aβ₄₂ at 5 and 12 months [60]. Our data demonstrating a decrease in the level of aggregated Aβ₄₂ were significant in mice treated with CBDA (30 mg/kg); however, the level of Aβ₄₂ that we measured was lower than predicted. Our data demonstrating a decrease in the level of aggregated Aβ₄₂ suggest that CBDA may be acting in a similar manner to THC. The psychoactive cannabinoid agonist THC has been shown to directly interact with Aβ peptide, inhibiting aggregation [61]. Alternatively, assuming decarboxylation of CBDA to CBD, the latter has been shown to enhance microglial phagocytosis of beta amyloid via TRPV2 [62]. This may influence levels of aggregated peptide.

Mass spectroscopy/proteomic analysis on cortical tissue was conducted to determine potential metabolic and signalling pathways that were significantly different between vehicle-treated wild-type and APP/PS1 mice. We were also interested to determine whether CBDA treatment (10 mg/kg) altered any of these pathways in the APP/PS1 cortex of APP/PS1 mice.

Comparing the canonical pathways that were significantly different between APP/PS1 mice and WT controls highlighted mitochondrial dysfunction and processes, including protein sorting and signalling, post-translational protein phosphorylation, integrin signalling, and synaptogenesis.

Mitochondrial Dysfunction: Mitochondria are the primary location for ATP production via the electron transport chain, acetyl coenzyme A (acetyl-CoA) oxidation, and oxidative phosphorylation. Cellular processes, including cell adhesion, synaptogenesis, protein sorting, and neurotransmitter release, require energy expenditure [63] depending on the metabolism of ATP and adenosine diphosphate (ADP). Considering mitochondrial dysfunction, of the top 25 proteins suggested to be involved in the IPA pathway analysis, 11 were significantly differentially expressed in the APP/PS1 cortex, with the others acting as connecting elements between those that were differentially expressed. Amongst those that were upregulated were APP, Cox5a, Cox6a1, Cycss, Ndufs4, and PDHX, while those that were downregulated included Sdhc, Prkaa1, Prkaa2, Pik3c3, mTOR, Gstt1, and Dnajc11 in APP/PS1 mice compared to wild-type controls (shown in the heat map in the Supplementary Materials. Considering how these proteins can alter function, these involve changes along the mitochondrial electron transport chain, with increases in Cox5a (cytochrome c oxidase subunit 5A, complex IV) and Ndufs4 (NADH oxidoreductase subunit 4) potentially increasing levels of reactive oxygen species, leading to oxidative stress and cellular damage. In addition, Cycs (a mobile electron transport carrier), NDUFS4, and PDHX (pyruvate dehydrogenase) can all increase ATP production, which can be used both as an energy source and as an agonist at purinergic receptors such as the pro-inflammatory P2X7 receptor. In the case of PDHX, its upregulation leads to increased levels of acetyl-CoA, which is a key substrate of the Krebs cycle to increase ATP production. Downregulation of Sdhc (succinate dehydrogenase) may impair complex II function, leading to reduced electron transport from succinate to ubiquinone and to diminished mitochondrial electron transport chain activity. Decreased production of fumarate from succinate could also impair the TCA cycle and thereby ATP production. There are also changes in the ATP sensor AMPK (AMP-activated protein kinase) with downregulation of Prkaa1/2 protein encoding the alpha one subunit of AMPK, which regulates cellular glucose and lipid metabolism. Upon AMPK activation, it phosphorylates and inactivates acetyl-CoA carboxylase (ACC) and beta-hydroxy beta-methylglutaryl-CoA reductase (Hmgcr), key enzymes involved in regulating de novo biosynthesis of fatty acids and cholesterol. Studies suggest that this catalytic subunit may control whole-body insulin sensitivity and is necessary for maintaining insulin resistance [64]. Interestingly, the altered expression of some of these mitochondrial proteins in APP/PS1 were reversed by CBDA, with the mitochondrial dysfunction pathway also showing less activation following CBDA treatment in the APP/PS1 mice. Mitochondrial dysfunction has been noted previously in APP/PS1 mice [65,66].

Beyond its partial reversal of protein expression involved in mitochondrial dysfunction, treatment with CBDA appears to alter protein expression in complexes I, II, and IV of the electron transport chain, which may increase activity within the electron transport chain in CBDA-treated APP/PS1 mice. For example, Cox5a (cytochrome C oxidase subunit 5A), a component of cytochrome C oxidase, is a member of complex IV, drives oxidative phosphorylation, and may increase ROS production. Its expression was increased in APP/PS1 mice, which was reversed by CBDA. Interestingly, Cox5a, which was rescued by CBDA in our study, has been reported to impact LTP. Transgenic mice with systemic Cox5a overexpression demonstrated improvement in hippocampal synaptic plasticity [67] Conversely, it has been suggested that NMDAR-dependent LTP induction causes a rapid burst of dendritic mitochondria fission, which, when blocked, prevents LTP [68]. This suggests that some of the mitochondrial proteins rescued by CBDA could impact LTP by altering mitochondrial fission through their regulation of DRP1 (dynamin related peptide 1), a cytosolic dynamin GTPase crucial for mitochondrial fission [69].

Protein Sorting: Protein sorting as a process involves both (i) signal-based sorting, where amino acids act as sorting signals, facilitating either gated transport or translocation of protein targets, and (ii) vesicle-based trafficking at the Golgi apparatus (GA), plasma membrane, or endosomes. There are clear changes in protein sorting (on a pathway level but also on an individual protein level) in the cortex of the APP/PS1 mice. Amongst those significantly downregulated were adaptor-related protein complex 1 subunit gamma 1 (Ap1g1), ARF guanine nucleotide exchange factor 2 (Arfgef2), coatomer protein subunit beta (Copb1), and transmembrane p24 trafficking protein 2 and 9 (Tmed2 and Tmed9) in APP/PS1 mice compared to wild-type controls (see the heat map in the Supplementary Materials). Downregulation of Ap1gp1 impacts protein trafficking within the trans-Golgi network and endosomes, while variants of Ap1g1 can cause neurodevelopmental abnormalities [70]. Arfgef2 plays an important role in intracellular vesicular trafficking, with mutations being associated with neural migration disorders [71]. The coatomer protein (Copb1) is required for budding from the GA interacting with GTPases encoded by the GTP-binding protein Sar1A. Both Copb1 and Sar1 are critical to retrograde transport between the endoplasmic reticulum (ER) and the GA. Interestingly, polymorphisms in coatomer associated proteins have been shown to be associated with increased Alzheimer’s disease risk [72]. The transmembrane domains (Tmed) Tmed2, and Tmed9 are glycoproteins that are present on the ER and involved in protein trafficking to the Golgi apparatus. Notably, Tmed2 plays a role in vesicular transport and regulates both interferon and Toll-like receptor signalling. Both Tmed9 and Vps29 have a role in autophagy, while the vacuolar protein sorting (Vps) proteins interacting with the retromer protein complex are involved in retrograde transport from the endosome to the trans-Golgi network (reduced in the heat map). The regulation of interferon and Toll-like receptor signalling via action on protein sorting processes in APP/PS1 may play a role in neuroinflammation.

CBDA treatment caused a significant increase in and reversal of expression of Tmed2, Tmed7, Copb1, Ap1g1, and Arfgef2. Consequently, this reversal should improve communication and protein trafficking to help restore function in the APP/PS1 cortex towards the control levels observed in WT mice. It should also restore Toll-like receptor trafficking to regulate inflammation and neurodegeneration the APP/PS1 cortex [73].

Synaptogenesis: The ability to form new synapses is key to the maintenance of neuronal health. A number of synaptogenesis-associated proteins were altered in expression in the APP/PS1 mouse cortex compared to wild-type controls. Amongst those significantly downregulated included metabotropic glutamate receptor 2 (Grm2), metabotropic glutamate receptor 5 (Grm5), kalirin RhoGEF kinase (Kalrn), serine/threonine-protein kinase mTOR (mTOR), Ras-related protein Rab-5A (Rab5a), neuronal proto-oncogene tyrosine-protein kinase Src (SRC), Ras-related protein R-Ras2 (Rras2), phosphatidylinositol 3-kinase catalytic subunit type 3 (Pik3C3), Cacna1b (calcium voltage-gated channel subunit alpha1 B), and synaptotagmin-7 (Syt7) (see the heat map in the Supplementary Materials. Considering the function of these proteins, the decreased expression of Grm2 and Grm5 will directly affect neurotransmission and synaptogenesis. Kalrn activates rho GTPases, impacting neuronal shape, growth, and plasticity via the actin cytoskeleton, while mTOR (mammalian target of rapamycin) activation is reported to be associated with increased synaptogenesis [74]. Downregulation of mTOR, in contrast, may reinforce the pathway analysis, showing decreased synaptogenesis in APP/PS1 mice, an effect reversed by CBDA. Syt7 regulates vesicular docking at the plasma membrane [75] and is crucial for normal synaptic transmission. The decreased expression of Cacna1B in the APP/PS1 cortex is also likely to impact neurotransmitter release. compromising synaptic transmission and consequently, reducing plasticity.

CBDA treatment of APP/PS1 mice significantly reversed the expression of many of these proteins. Fundamental proteins Grm5, Kalrnm, Src, and PikK3c3 were all significantly upregulated by CBDA treatment, while Rras2, Cacna1b, and Syt7 had increased expression, but these were not significant. This demonstrates that CBDA could significantly improve synaptogenesis in the APP/PS1 cortex, changing protein expression toward the levels in wild-type controls. These changes are likely to be directly linked to improved levels of LTP and PPF in APP/PS1 following CBDA treatment.

Integrin Signalling:Alterations in integrin signalling links directly to alterations in synaptogenesis [76] via decreased synaptic protein expression, with decreases in Gria1 (AMPA receptor subunit 1), Grm2 (glutamate metabotropic receptor2), and Grm5 (glutamate metabotropic receptor 5). There is also a notable decrease in Kalrn (Kalirin), the synaptic regulator [77] that is known to be reduced in expression in the hippocampus of Alzheimer’s disease patients [78]. Integrins are important structural transmembrane receptors involved in adhesion between cells and to the extracellular matrix [79] but also play an important role in neurite growth [80]. With regard to integrin protein expression changes, amongst those that were significantly upregulated were Cttn (Src substrate cortactin) and Pak3 (p21 activated kinase, the target receptor of paxillin, a prototypical integrin), while those that were downregulated included ARF GTPase-activating protein (Git1), Ras-related protein 2A (Rab2A), Ras-related protein R-Ras2 (Rras2), and tetraspanin-7 (Tspan7) in APP/PS1 mice compared to wild-type controls. The decrease in Git1 in APP/PS1 may result in decreased spine density and spine development [81], while the decrease in the endocytotic membrane protein Tspan7 is likely to interrupt the integrin beta1/Fak/Src signalling pathway [79], with Src also being reduced. Decreased Tspan7 is involved in spine maturation and AMPA receptor trafficking [82], and it also interacts with the metalloprotease ADAM10, which can regulate both presenilin and amyloid precursor protein [83]. Decreased Pik3c3 in APP/PS1 cortical tissue is likely linked to decreased plasticity. Pik3c3 is widely expressed in neurons, producing PI3P in dendritic spines. Pik3c3 is also required for mTOR signalling [84]. Deletion of Pik3c3 leads to a loss of synapses in cortical pyramidal neurons, gliosis, and neurodegeneration [85]. The increase in Pak3 is known to be specifically involved in FAD APP-mediated neuronal apoptosis, in which the serine–threonine kinase interacts with APP [86].

While we noted a decrease in v-raf murine sarcoma viral oncogene homolog B1 (Braf) in APP/PS1, there was an increase above levels observed in the WT cortex following CBDA treatment. As Braf is known to be associated with cell division and differentiation linked to ERK/MAP kinase, this may also drive improved synaptic plasticity. Integrin signalling is key to new synapse formation [87], and it is clear from the canonical pathway analysis that this was inhibited in APP/PS1 mice but improved following CBDA treatment. This has important consequences for the capacity to signal not only at synapses but also at dendritic spines and neurites, likely to help rescue synaptic plasticity and influence paired-pulse facilitation. However, the molecular machinery to support integrin signalling was diminished in the APP/PS1 mice, as a number of key proteins were downregulated. While CBDA treatment improved the directionality of protein expression of many proteins associated with integrin signalling in the integrin, the proteins Cttn, Pak3, Src, and Pik3c3 were significantly increased.

All of the above changes in mitochondrial, synaptogenesis, protein sorting, and integrin signalling protein expression were improved by the administration of CBDA to the APP/PS1 mice. Given the consequences of decreased ATP production for energy-dependent processes such as protein sorting and synaptogenesis, it is of no surprise that mitochondrial change may underline the synaptic plasticity changes produced by CBDA in the APP/PS1 mice, but this remains to be experimentally tested.

In addition, the network analysis of the proteome of the APP/PS1 mouse cortex compared to that of the WT controls suggests a central node to be APP and predicted activation of Fam20c proteins and platelet alpha granule release. Increased expression of APLP2 and Scg2 converged to result in predicted activation of Fam-20 substrates. Taken together, this suggests a pro-inflammatory phenotype in the APP/PS1 mouse cortex. Interestingly, treatment with CBDA inhibited both Fam20c proteins and platelet alpha granule release without changing APP tone per se. This suggests that both Fam20 inhibition and the inhibition of platelet alpha granule release play a role in anti-inflammatory actions of CBDA in the APP/PS1 mouse cortex. Decreased expression of Eftud2 (a key spliceosomal protein) was also observed in the APP/PS1 mice, which confirms the pathway analysis of altered protein processing. Edtud2 regulates other proteins associated with ribosomal function and protein synthesis and is reported to be brain protective [88]. Its decreased expression in APP/PS1 mice may underlie in part the reduction in synaptic plasticity and neurodegenerative AD-like phenotype. The network analysis also confirmed the altered protein sorting processing seen in the pathway analysis. In particular, it highlights endosome-to-Golgi apparatus protein trafficking (retromer inhibition) in addition to Sec22b inhibition of cargo transport in vesicles that complex with SNAREs [89]. CBDA treatment appeared to change protein expression, reverting towards wild-type. All of the above changes in mitochondrial, synaptogenesis, protein sorting, and integrin-signalling protein expression were also seen on a pathway level and were improved by the administration of CBDA to the APP/PS1 mice.

Of the additional proteins upregulated in APP/PS1 mice, high-mobility group box protein one (Hmgb1) is a nuclear protein known to be released by glial cells in AD to initiate neuroinflammation. Hmgb1 functions as an archetypal alarmin and a typical damage-associated molecular pattern (DAMP) molecule signalling via RAGE and TLR4 to increase inflammation and alter cognition [90]. Treatment with CBDA reversed the increase in HMGB1, suggesting a powerful anti-inflammatory action. We also found an increase in the RNA metabolism gene Hnrnph2, in agreement with other studies of AD [91]. Also increased in APP/PS1 was HSpd1 or HSp60, a member of the heat shock protein family essential for the folding and assembly of newly imported proteins in the mitochondria. This has been shown to be upregulated in sporadic and familial AD cortices [92]. This was significantly reversed by CBDA treatment. Also of interest is dimethylargininase (Ddah2), a zinc protein involved in nitric oxide synthase regulation, which is elevated in neurons displaying cytoskeletal abnormalities and oxidative stress in Alzheimer’s disease (AD) [93]; it was also increased in APP/PS1 and decreased significantly following CBDA. Considering the endocannabinoid system, there was downregulation of FAAH (fatty acid amide hydrolase), which is known to be involved in the degradation of anandamide in APP/PS1 cortical tissue. FAAH inhibitors have been proposed as potential agents for AD therapy, as genetic inactivation of FAAH in an AD mouse model restored decreases in PPF and LTP [94]. While CBDA restored LTP in our model, it caused an increase in FAAH.

In a previous investigation of the proteomic and lipidomic profile of APPswe/PS1dE9 mice, [95] suggested alterations in both the metabolic and glycerophospholipid profiles at the age of 7 to 8 months. Oxidative stress has been shown to be involved in early memory deficits in APPswe/PS1dE9 mice, linked to decreased levels of glutathione peroxidase and superoxide dismutase [96]. Mitochondrial dynamics are known to be altered in APPswe/PS1dE9 mice, with increased levels of fission and fusion proteins commencing at 3 months, which precede the onset of memory decline [97].The CBDA-mediated improvement in mitochondrial function and synaptogenesis is likely to be key to our observations of improved synaptic plasticity in the form of LTP and short-term changes in transmitter release measured via PPF. Caution should be exercised with regard to the proteomics data interpretation, as (i) the sample size is modest and (ii) this will require independent proteomics confirmation by Western blot of the indicated protein changes and their downstream signalling cascades. It does, however, point towards key mechanisms for further validation.

In summary, we demonstrated that CBDA treatment acts to improve synaptic plasticity and regulate the expression of some essential proteins in the APP/PS1 cortex. CBDA, which has good bioavailability, may provide an alternative approach for the treatment of Alzheimer’s disease and warrants further research.