Alzheimer's disease (AD) is a consequence of several detrimental processes, such as protein aggregation, oxidative stress, mitochondrial malfunction, and neuroinflammation, finally resulting in the loss of neuronal functions (Blass and Gibson, 1991; Sullivan and Brown, 2005; Yao et al., 2009; Querfurth and Laferla, 2010; Serviddio et al., 2011; Cassano et al., 2016, 2019; Carapelle et al., 2020). Starting from this evidence, multitarget drugs have been increasingly sought after over the last decades (Barone et al., 2019; Cassano et al., 2019). In this regard, palmitoylethanolamide (PEA) seems to exert neuroprotective effects by modulating more therapeutic targets at the same time (Valenza et al., 2021). In fact, it has been demonstrated that PEA, the naturally occurring amide of ethanolamine and palmitic acid, is an endogenous lipid that exerts its beneficial effects through the peroxisome proliferator-activated receptors (PPARs) involvement, transient receptor potential vanilloid type 1 channel, orphan G-protein-coupled receptor 55, and the so-called entourage effect on the endocannabinoid system (Devchand et al., 1996; Delerive et al., 2001; LoVerme et al., 2005, 2006; Scuderi et al., 2011; Tomasini et al., 2015; Bronzuoli et al., 2019; Beggiato et al., 2020a,b). In addition to its known anti-inflammatory activity, PEA protects cultured mouse cerebellar granule cells from glutamate toxicity and reduces histamine-induced cell death in hippocampal cultures (Skaper et al., 1996).

Furthermore, it has been demonstrated that PEA exerts in vitro and in vivo a combination of neuroprotective and anti-inflammatory effects in amyloid-β (Aβ)-induced toxicity by interacting at the PPAR-α nuclear site (D'Agostino et al., 2012; Scuderi et al., 2012, 2014). The neuroprotective action of PEA has been established also in transgenic animal models of AD (D'Agostino et al., 2012; Bronzuoli et al., 2018; Scuderi et al., 2018). In fact, we have previously demonstrated that ultramicronized PEA (um-PEA), a formulation endowed with better bioavailability, can rescue learning and memory impairments in a triple transgenic mouse model of AD (3 × Tg-AD) by exerting anti-inflammatory properties, dampening reactive astrogliosis, and promoting the glial neurosupportive function. Moreover, um-PEA strongly suppresses Aβ(1–42) expression and reduces the abnormal phosphorylation of tau in 3 × Tg-AD mice (Scuderi et al., 2018).

Mitochondrial dysfunctions are related to inflammation and other energy-dependent disturbances, where the generation of reactive oxygen species (ROS) exceeds the physiological antioxidant activity, causing cellular oxidative damage (Lin and Beal, 2006; Reddy, 2009; Romano et al., 2017). Mitochondrial dysfunction can lead to glutamatergic neurotransmission alterations (Vos et al., 2010) and excitotoxicity (Schinder et al., 1996). In fact, excessive glutamate concentration can cause excitotoxicity, leading to calcium influx, mitochondrial dysfunction, and subsequent cell death (Mahmoud et al., 2019). Therefore, considering the tight link between mitochondria and glutamate, it could be crucial to investigate their alterations in the context of AD pathology.

In the last decade, neuroimaging has been used to complement clinical assessments in the early detection of AD, showing that morphological change (such as measurement of the hippocampal volume) might be an important hallmark for the diagnosis of AD. More recently, the development of magnetic resonance spectroscopy (MRS) in the field of AD revealed that tremendous metabolic changes occur during the progression of AD correlating with cognitive abnormalities (Meyerhoff et al., 1994; Shonk et al., 1995; Frederick et al., 1997; Rose et al., 1999; Kantarci et al., 2000; Huang et al., 2001). Nevertheless, few studies with relatively small sample sizes investigated MR spectral profile alteration as a biomarker for treatment response in AD, and some of them correlated this alteration with psychiatric symptoms in patients with AD (Sweet et al., 2002; Bartha et al., 2008). Although metabolite abnormalities in AD were demonstrated in different samples and pathologically confirmed cases, the pathological significance of these changes is not fully understood. Therefore, in our study, we exploited the availability of the MRS to investigate whether it can provide complementary predictive information regarding (i) progression of the AD-like pathology and (ii) um-PEA treatment effects.

In this article, we used the 3 × Tg-AD mice harboring three mutant human genes (betaAPPSwe, PS1M146V, and tauP301L) to directly test the hypothesis that (i) progressive accumulation of Aβ is closely related to mitochondrial, metabolic, and glutamatergic alterations, and (ii) chronic um-PEA administration may ameliorate such alterations. To test these hypotheses, we used an integrated approach, based on ex vivo studies of several mitochondrial bioenergetic parameters, in vivo studies of magnetic resonance imaging (MRI) and spectroscopy (MRS) to evaluate brain metabolites, as well as neurochemical analysis of extracellular levels of glutamate by microdialysis in 3 × Tg-AD vs. non-Tg control mice tested at two different stages (mild and severe) of AD-like pathology and subcutaneously treated for 3 months with um-PEA.

The most novel findings of our study are the demonstration that chronic um-PEA treatment can ameliorate the complex-I respiration rate, the F_oF₁-ATPase (complex V) activity, as well as ATP content in the cortical mitochondria from 3 × Tg-AD mice. Otherwise, mitochondrial bioenergetics, as well as the release of glutamate after depolarization were not ameliorated by um-PEA treatment in the HIPP of both young and adult 3 × Tg-AD mice. More interestingly, progressive age- and pathology-related changes were observed in the cortical and hippocampal metabolism that closely mimic the alterations observed in the human AD brain; these metabolic alterations were not affected by chronic um-PEA treatment.

We revealed for the first time that (i) AD-like pathology differently affects mitochondrial bioenergetics in the FC and HIPP and (ii) um-PEA treatment does not promote significant ameliorations simultaneously in those brain regions of 3 × Tg-AD mice. This is in line with the temporal- and regional-specific development of AD neuropathology in the brain of 3 × Tg-AD mice, which closely mimics their development in the human AD brain (Oddo et al., 2003a,b; Hirata-Fukae et al., 2008).

Compared with many studies, we adopted a longitudinal study design with a sufficiently long observation period that allowed us to assess early alterations and changes over time in both non-transgenic and 3 × Tg-AD mice. This study further extends the knowledge of the molecular modification during both normal aging and the progression of AD-like pathology in our murine model of AD (Oddo et al., 2003a,b). Moreover, we analyzed the modulatory effect of PEA on pathways and factors consistently associated with AD-like pathology and symptoms. In fact, we selected two different subsets of 3 × Tg-AD mice, which allowed us to investigate the therapeutic potential of um-PEA in restraining the development of AD-like pathology at mild and severe stages, focusing on the mitochondrial bioenergetics, cerebral metabolism, and glutamatergic transmission. Regarding mitochondria and glutamate, we have previously demonstrated that deficits of glutamatergic transmission and mitochondrial dysfunction coexist in the FC and HIPP of 18-month-old 3 × Tg-AD mice, which show a substantial number of amyloid plaques and tau pathology; the HIPP was the most affected area at that age (Cassano et al., 2012).

Despite accumulating evidence supporting the link between inflammation, mitochondria, and metabolism, several important questions remain unanswered. Furthermore, there are no studies addressing the effects of PEA on mitochondria bioenergetics in in vivo experimental models of AD, and only one study, to our knowledge, has assessed in vivo the effects of PEA on mitochondrial dysfunction observed in central nervous system pathological conditions (Cristiano et al., 2018). In this regard, Cristiano and colleagues found that PEA treatment improves hippocampal mitochondrial function and reduces oxidative stress in a genetic murine model (BTBR T + tf/J), which exhibits a behavioral phenotype of autism spectrum disorder. In particular, 10-day treatment with um-PEA was able to (i) restore hippocampal mitochondrial state 3 respiration using succinate as substrate, (ii) increase superoxide dismutase (SOD) activity, (iii) counteract ROS, and (iv) decline the energy efficiency, as evidenced by the decreased degree of coupling (Cristiano et al., 2018). Together, these findings demonstrate that PEA counteracts mitochondrial dysfunction and helps rescue brain energy metabolism during pathological states, balancing ROS production/antioxidant defenses, and limiting oxidative stress. Moreover, PEA-treated BTRB mice showed a low activation of pro-inflammatory cytokines at hippocampal and serum levels (Cristiano et al., 2018). Likewise, we have previously observed that 3-month treatment of um-PEA almost completely abolishes the increase in inflammatory markers observed in 6-month-old 3 × Tg-AD mice and suppresses the expression of proinflammatory mediators, including interleukin (IL)-1β, IL-16, IL-5, monocyte chemotactic protein 5 (MCP-5), and macrophage colony-stimulating factor (M-CSF), but not inducible nitric oxide synthase (iNOS) and tumor necrosis factor (TNF), while enhancing the anti-inflammatory IL-10 (Scuderi et al., 2018).

In this study, 6-month-old 3 × Tg-AD mice with a mild AD-like pathology showed a significant reduction of ATP content in the FC that was counteracted by um-PEA treatment. Otherwise, at 12 months of age, together with severe hallmarks of AD-like pathology, mitochondrial alterations were more evident and regional-specific. In fact, cortical mitochondria of 3 × Tg-AD mice exhibited a reduced respiratory capacity, which might partially block electron flow within the respiratory chain and consequently increase oxidative stress; this effect was counteracted by um-PEA. The reduced respiratory activity was not observed after incubation with a complex II-linked substrate. The same pattern of alterations was observed in the cortical mitochondria of 18-month-old 3 × Tg-AD mice, which present diffuse extracellular Aβ deposits and extensive human tau immunoreactivity (Cassano et al., 2012). Different from the latter age, at 12 months the reduced oxygen consumption observed in the transgenic cortical preparations was accompanied by alterations of F_oF₁-ATPase activity and ATP content that were counteracted by um-PEA treatment.

Regarding the HIPP, 12-month-old 3 × Tg-AD mice showed the same mitochondrial dysfunctions previously observed with mice at 18 months of age (Cassano et al., 2012). In fact, here we confirmed a significant increase in state 3 and state 4 respiration rates with complex I-linked substrates and a significant impairment of ATP homeostasis. Different from FC, the pharmacological treatment with um-PEA did not ameliorate the hippocampal mitochondrial bioenergetics. Taken together, these findings suggest that um-PEA counteracts mitochondrial dysfunction, helps rescue brain energy metabolism during pathological states only in the FC, and such discrepancy might be due to the difference in Aβ/tau-linked alterations observed between these brain regions. In fact, it has been documented that Aβ and tau pathology differently impact brain regions, with HIPP showing more severe alterations compared to FC (Oddo et al., 2003a,b; Cassano et al., 2012; Bellanti et al., 2017). However, other mechanisms could underlie the regional-specific effect of um-PEA on mitochondrial bioenergetics and other experiments will be necessary to explain this phenomenon.

Brain structures undergo drastic modifications during aging, becoming progressively less interconnected and undergoing several metabolic and structural changes. A clinical scanner by MRS may serve to identify patients with AD before clinical symptom onset to help distinguish AD from other neurodegenerative disorders, as well as evaluate treatment effects.

In this regard, clinical studies demonstrated that NAA, a marker of neuronal integrity, can be detected by MRS and used to differentiate normal aging and pathological dementia (Ackl et al., 2005; Jessen et al., 2005; den Heijer et al., 2006). Besides decreased levels of NAA, several clinical studies demonstrated that mINS, a marker of glial activation, was increased in patients with AD (Glanville et al., 1989; Miller et al., 1993; Bitsch et al., 1999; Jessen et al., 2000; Kantarci et al., 2000; Huang et al., 2001).

Herein, we performed for the first time a comprehensive evaluation of six brain metabolites in the areas most involved in AD-like pathology. Interestingly, HIPP of 3 × Tg-AD mice was characterized by lower concentrations of NAA and increased mINS, as seen in human patients with AD (Glanville et al., 1989; Klunk et al., 1992; Miller et al., 1993; Bitsch et al., 1999; Jessen et al., 2000; Kantarci et al., 2000, 2007; Huang et al., 2001; Cheng et al., 2002; Schott et al., 2010). In particular, NAA reduction was observed at both ages (6 and 12 months), whereas mINS increased gradually with mouse age. Similar results were observed with APP/PS1 double transgenic mice, which is characterized by the early appearance of Aβ plaques, neuronal degeneration, and synaptic loss in the brain (Marjanska et al., 2005; Chen et al., 2009). The authors found that NAA was significantly reduced both in the FC and HIPP of 5-month-old APP/PS1 transgenic mice when pathology showed the formation of sparse Aβ plaques in these areas, and the number of neurons decreased. As the age of the mice increased, the NAA decrease was continuously accompanied by the increase in both mINS and the number of Aβ plaques (Marjanska et al., 2005; Chen et al., 2009). In our 3 × Tg-AD mice, we found an increase in mINS by age only in the HIPP, but not in the FC, where a significantly higher concentration was observed only at 6 months of age compared to the non-Tg group.

Other results were obtained in the FC with different transgenic AD mice. In fact, no increase in mINS in the FC was observed in the transgenic mouse line PS2APP [PS2N141I × APP(swe)], which develops an age-related cognitive decline associated with severe amyloidosis (Von et al., 2005). Therefore, further studies need to be performed to justify the discrepancy existing in the literature, which may be attributed to the differences in the experimental parameters and animal models of AD-like pathology.

Our longitudinal study revealed a significant decrease in tCr only in the HIPP of 12-month-old 3 × Tg-AD mice. This metabolic profile, together with lower NAA, is a clear indicator of the age-related hypometabolism in the HIPP of 3 × Tg-AD mice, further suggesting that HIPP is the most affected area at that age (Cassano et al., 2012; Scuderi et al., 2018).

A decrease in NAA/Cr ratio accompanied by an increased mINS/Cr ratio occurs in patients with AD; the latter temporal progression of metabolite abnormalities in patients with AD correlates with the progression of the Aβ/tau-linked alterations (Glanville et al., 1989; Klunk et al., 1992; Miller et al., 1993; Bitsch et al., 1999; Jessen et al., 2000; Kantarci et al., 2000, 2007; Huang et al., 2001; Cheng et al., 2002; Schott et al., 2010). Interestingly, we found a similar pattern of changes in the metabolite ratio both in the FC and HIPP, as seen in human patients with AD.

The role of Cho metabolite in AD is more controversial since several clinical studies demonstrated an increase (Meyerhoff et al., 1994; Pfefferbaum et al., 1999; Kantarci et al., 2007), others no change (Moats et al., 1994; Schuff et al., 1997; Rose et al., 1999; Krishnan et al., 2003), whereas a decrease was reported in one study (Chantal et al., 2002). In our study, we observed a significant decrease of tCho only in the HIPP of both 6- and 12-month-old mice. The Cho cytosolic concentration derives from the breakdown products of phosphatidylcholine (Schuff et al., 1997). Therefore, it has been proposed that catabolism of the phospholipid membrane bilayer allows AD subjects to produce choline to compensate for declining acetylcholine (Wurtman et al., 1985).

Although Glu is the principal excitatory neurotransmitter involved in learning and memory, it has been less well investigated by MRS/MRI. In this regard, few clinical studies revealed a reduction of Glu levels in different brain regions of patients with AD (Antuono et al., 2001; Hattori et al., 2002; Rupsingh et al., 2011) that correlates with the severe reduction of the cerebral metabolic rate for glucose (Mosconi et al., 2008; Bedse et al., 2015; Barone et al., 2016). In fact, glutamate neurotransmission is carried out by a glial-neuronal process that includes the oxidation of glucose and the ATP-dependent glutamine-glutamate cycle. A total of 80–90% of total cerebral glucose usage is attributable to the energy requirements of glutamatergic neurotransmission (Magistretti et al., 1999; Gatta et al., 2016; Pardeshi et al., 2017; Tramutola et al., 2018; Sharma et al., 2019; Bukke et al., 2020a,b). The neurochemical changes obtained from in vivo human MRS studies agree with the results observed in preclinical studies. In fact, the analysis of cortex and HIPP in APP-PS1 mice showed an age-dependent reduction of Glu levels, which correlates with increasing brain amyloidosis (Marjanska et al., 2005). Similar findings were observed in the cortex of APP (Dedeoglu et al., 2004) and PS2APP mice (Von et al., 2005). This is in line with the Glu spectroscopy findings between human AD subjects and the three transgenic mouse models of AD.

In our study, we demonstrated a temporal- and regional-specific alteration of Glu in 3 × Tg-AD mice with the HIPP the only affected area. In particular, at 12 months of age, we found a significant reduction (−38%) of Glu in the HIPP of vehicle-treated 3 × Tg-AD vs. vehicle-treated non-Tg mice. These results, observed in the vehicle-treated 3 × Tg-AD mice, together with lower hippocampal NAA (−45%), tCr (−24%), and tCho (−31%), further support the evidence of an age-related hypometabolism in the HIPP. MRS analysis might be of interest also to depict how neural correlates do change along with pharmacological treatment. In this regard, in our experimental condition, we did not observe any gross treatment-dependent differences between genotypes.

To investigate whether the decrease in hippocampal glutamate levels in the metabolic pool was paralleled by alterations in the glutamatergic nerve terminals, the pattern of glutamate release in basal condition, and in response to K⁺-stimulation was explored by microdialysis sampling in the vHIPP of 6- and 12-month-old mice. We have previously demonstrated that the basal release of glutamate was significantly increased (+80%) at 6 months of age whereas a trend toward an increase (+43%) was observed in 12-month-old 3 × Tg-AD compared to non-Tg mice (Scuderi et al., 2018). Here, we demonstrated that at both ages the depolarization-evoked glutamate release resulted completely disrupted in 3 × Tg-AD compared to non-Tg mice, thus suggesting that the deficit of glutamate metabolic pool observed by MRS might sustain, at least in part (at 12 months of age), the decrease of extracellular glutamate concentration in mutant mice.

Our results not only elucidate the temporal correlation among mitochondrion, metabolic alterations, and glutamatergic dysfunction but also, in future studies, allow the assessment of putative therapeutic strategies. However, this study presents some limitations. For instance, additional measurements would allow performing a tight correlation between molecular markers of bioenergetics dysfunction, for instance, hydrogen peroxide production, proton leak or membrane potential, and metabolic alterations. Finally, this study sheds additional light on the effects of PEA treatment, displaying selective bioenergetic effects. At present, the use of a PEA treatment may represent a possible novel frontier for the future clinical treatment of AD.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The animal study was reviewed and approved by Ethics Committee of the University of Foggia.

TC, CS, LS, and GS designed the research. FB, VB, MA, RV, GP, RC, and SB performed the research. FB, RC, GV, GS, and TC analyzed data and interpreted the results. FB and TC wrote the manuscript. All authors contributed to the article and approved the submitted version.

This project was supported by the Italian Ministry for Education, University and Research (PRIN to LS and GV), and the Department of Clinical and Experimental Medicine (Dipartimenti di eccellenza – Legge 232/2016). This paper/manuscript/study has been published with the financial support of the Department of Medical and Surgical Sciences of the University of Foggia.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.