Is the Brain Undernourished in Alzheimer’s Disease?

Cerebrospinal fluid (CSF) amino acid (AA) levels and CSF/plasma AA ratios in Alzheimer Disease (AD) in relation to nutritional state are not known. Methods: In 30 fasting patients with AD (46% males, 74.4 ± 8.2 years; 3.4 ± 3.2 years from diagnosis) and nine control (CTRL) matched subjects, CSF and venous blood samples were drawn for AA measurements. Patients were stratified according to nutritional state (Mini Nutritional Assessment, MNA, scores). Results: Total CSF/plasma AA ratios were lower in the AD subpopulations than in NON-AD (p < 0.003 to 0.017. In combined malnourished (16.7%; MNA < 17) and at risk for malnutrition (36.6%, MNA 17–24) groups (CG), compared to CTRL, all essential amino acids (EAAs) and 30% of non-EAAs were lower (p < 0.018 to 0.0001), whereas in normo-nourished ADs (46.7%, MNA > 24) the CSF levels of 10% of EAAs and 25% of NON-EAAs were decreased (p < 0.05 to 0.00021). CG compared to normo-nourished ADs, had lower CSF aspartic acid, glutamic acid and Branched-Chain AA levels (all, p < 0.05 to 0.003). CSF/plasma AA ratios were <1 in NON-AD but even lower in the AD population. Conclusions: Compared to CTRL, ADs had decreased CSF AA Levels and CSF/plasma AA ratios, the degree of which depended on nutritional state.


Introduction
There is currently no treatment able to stop the progression of Alzheimer's disease dementia [1]. It is desirable that a future improved understanding of brain metabolic alterations in Alzheimer's disease (AD) might provide the possibility of non-pharmacological interventions to prevent or slowdown the progression of this disorder.
Lack of energy and oxidative stress, indeed, initiate and worsen over time the synaptic pathology in AD [12,14,26,27], as adequate energy availability is essential for normal  Tables 2 and 3 show the CFS AA levels and plasma AA levels in ADs and in CTRL, respectively. Compared with CTRL, levels of AAs in CSF and plasma in ADs were reduced by 75% and 31.6%, respectively. However, in the CSF compartment, the numbers and types of AAs reduced varied in relation to patient nutritional state ( Table 4). Given that Group 1 (malnutrition) and Group 2 (at risk) had similar numbers and types of altered CSF AA (with the exception of alanine, present only in group 2, and aspartic acid, present only in group 1), these groups were pooled together to form a combined group (CG) (named the "nutritionally-deteriorated" group). In CG, the alterations in numbers and types of AAs were wider than in Group 3 (normal nutrition). For clarity, these results were synthesized in Tables 5 and 6.      Table 6. Differences in the types of altered CSF amino acids in relation to nutrition state in AD patients. Compared to CTRL, CG had lower plasma levels of glutamate, alanine, isoleucine, leucine, total AAs (TAAs), and Essential AAs (EAAs), and a lower EAA/TAA ratio (Table 7). Compared to CTRL, group 3 (normonutrition) had lower plasma levels of alanine and higher plasma BCAA/EAA and BCAA/TAA ratios. Plasma histidine levels were higher in each AD subgroup compared to CTRL.

Group 3 Normonurished
In the CSF compartment (Table 8), compared to CTRL, CG and group 3 had in common significant reductions in levels of asparagine, serine, glutamine, glycine, arginine, tryptophane, phenylalanine, Trp ratio, TAAs, indicating that these altered AAs were independent of nutritional state. In addition, the CG had reduced levels of threonine, alanine, valine, methionine, isoleucine, leucine, lysine, EAAs, and BCAAs. Thus, deterioration of nutritional state in ADs was associated with wider reductions in AAs. In the CG, leucine, isoleucine, threonine, TAAs, and EAAs were reduced both in plasma and CSF. Table 9 shows the CSF/plasma AA ratios in CTRL and ADs as an entire population. In CTRL the ratios were <1, indicating that the AA levels in CSF were lower than those in plasma, with the exception of glutamine and to a lesser extent BCAA/EAA whose CSF/plasma ratios were >1. Compared to CTRL, ADs had significantly lower CSF/plasma AA ratios, including glutamine and BCAA/EAA.     Table 10 shows that CSF/plasma AA ratios were similar among the normonourished and nutritionally deteriorated patients. To sum up, compared to CTRL, ADs had important reductions in numbers and types of CSF AAs. Moreover, patients' stratification according to nutritional state indicated the existence of two CSF AA profiles: one in which reduced AAs were independent of nutritional state, the other with a further reduction in AAs associated with a deteriorated nutritional state. Moreover, in CTRL subjects the CSF/plasma AA ratios were <1 but >1 for glutamine and the BCAA/EAA ratio. In ADs, CSF/plasma ratios were even lower than in CTRL, including glutamine and the BCAA/EAA ratio. Table 11 shows the correlations between plasma insulin levels, CSF AA levels, and markers of AD pathology.
Correlations between CSF AAs and plasma AAs were evidenced both in ADs (Table 12) and CTRL (Table 13). In AD, an inverse correlation in glutamine levels was observed between plasma and CSF, whereas all BCAAs, the BCAA/TAA ratio ( Figure 2) EAA/Total AA, and BCAA/EAA were positively associated.  Insulinemia was inversely correlated with CSF arginine (r = −0.52; p = 0.004) ( Figure  1), whereas no other relationship was found with other CSF AAs, nor with Aβ, tau, or phosphor tau levels. Correlations between CSF AAs and plasma AAs were evidenced both in ADs (Table  12) and CTRL (Table 13). In AD, an inverse correlation in glutamine levels was observed between plasma and CSF, whereas all BCAAs, the BCAA/TAA ratio ( Figure 2) EAA/Total AA, and BCAA/EAA were positively associated.   In CTRL, the study found positive links for serine (r = + 0.96; p = 0.0007), BCAA/TAA ratio (r = + 0.90; p = 0.0005), and valine (r = + 0.74; p = 0.046).

Discussion
The study confirms the initial hypothesis that ADs, compared to controls, may have reduced CSF AA levels and lower CSF/plasma AA ratios and that the amount of CSF AA reduction is in part dependent on patients' nutritional state.
The patients' stratification by nutritional state identified two patterns of reduced CSF AA levels, one that was independent of nutritional state (found therefore in all ADs), and  In CTRL, the study found positive links for serine (r = + 0.96; p = 0.0007), BCAA/TAA ratio (r = + 0.90; p = 0.0005), and valine (r = + 0.74; p = 0.046).

Discussion
The study confirms the initial hypothesis that ADs, compared to controls, may have reduced CSF AA levels and lower CSF/plasma AA ratios and that the amount of CSF AA reduction is in part dependent on patients' nutritional state.
The patients' stratification by nutritional state identified two patterns of reduced CSF AA levels, one that was independent of nutritional state (found therefore in all ADs), and another that was found in the nutritionally deteriorated group only (i.e., group 1 + group 2). The latter group of ADs therefore exhibited both nutritionally independent and nutritionally dependent CSF AA reductions.
The lower CSF/plasma ratios in ADs, both as an entire group and after stratification by nutritional state, suggest that AA reductions in CSF may be due not only to insufficient supply of AAs to the brain from plasma but also to brain overutilization. The following factors may sustain the hypothesis of brain AA overutilization in ADs. First, as shown in the current study, 75% of the AAs in CSF but only 31.6% of the AAs in plasma were reduced. Second, the CSF levels of several AAs, including two EAAs, were lower in normonourished ADs despite their normal plasma levels. Third, as previously documented, AA/protein dysmetabolism is more severe in the AD brain than in extracerebral areas [52]. Another investigation reported the presence of increased urea levels in CSF [52][53][54].
For more clarity, brain AA alterations will be discussed below as derived from two main potential mechanisms, one of which may be refereed to as "general" and the other "specific", while being aware that these two are intimately and inextricably interconnected.

Some Potential General Mechanisms for Reduced CSF AA
In addition to the need for energy production [31,55], insulin resistance, the synthesis of neuropathology markers of AD, and the mutual influence between insulin resistance and biosynthesis of AD markers are among the mechanisms able to prime/enhance brain AA overutilization.
The development of insulin resistance [2,3,56] might be favored and/or heightened by gut dysbiosis [57,58] that, via intestinal translocations of bacteria, bacterial amyloids and toxins, may increase cerebral inflammation brought about by Aβ [59]. Insulin resistance/hyperinsulinemia further reduces the residual utilization of glucose in mitochondria by blocking the mitochondrial pyruvate dehydrogenase enzyme complex [60].
Aβ itself can cause increased brain AA utilization, as demonstrated in both AD patients [61] and in AD model mice [62], via activation of mTOR signaling. On one hand, mTOR increases tau protein synthesis, the main component of the neurofibrillary tangles, [63] and, on the other hand, it stimulates mitochondrial activity [64]. In turn, activated mTOR can increase AA consumption for the synthesis of Aβ and tau protein [61]. Supporting the anabolic relationship between Aβ and mTOR, the inhibition of mTOR in AD model mice can reduce Aβ and tau protein levels, leading to the recovery of cognitive function [65,66]. Of note, BBB alteration itself contributes to increased brain AA utilization by upregulating Amyloid Precursor Protein (APP) expression and Aβ deposition [67].
The interrelationship between altered brain insulin signaling and the biosynthesis of AD markers may be another mechanism leading to brain AA overutilization. Hyperinsulinemia indeed increases tau phosphorylation [68,69], favors the formation of senile plaques [70] and inhibits degradation of extracellular neuronal Aβ by blocking the insulindegrading enzyme [71]. In turn, both Aβ and tau worsen brain insulin signaling [72], in particular in synaptic areas [73]. A puzzling problem might be the coexistence of insulin resistance/hyperinsulinemia with reduced insulin receptor density in brain structures (by up to 80% in severe AD) [74,75]. Probably, chronic hyperinsulinemia/cerebral insulin resistance may downregulate and decrease the affinity of brain insulin receptors [74,76]. In the current study, we did not find significant correlations between insulinemia/IR and CSF Aβ or tau levels. A possible explanation is that this association might be more easily found within the brain structures.
It is likely that low CSF AA levels were not due to the blood-brain barrier (BBB)restricted entry of AAs into the brain from plasma, or due to Choroid Plexuses (CPs) removing AAs from CSF and/or reducing AA secretion into CSF, which are both mechanisms that, under physiological conditions, determine brain AA availability [41]. BBB and CP activities, indeed, are altered in AD. Early BBB breakdown and dysfunction occur before the development of dementia and brain atrophy [77][78][79][80], as also confirmed in post-mortem studies [81]. In AD, CPs undergo epithelial atrophy, basement membrane thickening, and stroma fibrosis [39]. Therefore, it is most likely that the reductions in CSF AA levels may be due to increased brain uptake [41] and consumption of AAs.
In AD, elevated urea cycle activity and elevated urea levels, both in the TCA cycle [52] and in CSF (not measured in the present study) [52][53][54], support brain AA overutilization. An excess of brain AA utilization has a negative impact on brain function as, in addition to increased urea levels, causes hyperammonia, given that neurons and glia lack all the enzymes of urea cycle [82]. Hyperammonia worsens brain dysfunction, as it is implicated in several metabolic processes including, to name only a few, oxidative damage [83,84] impaired mitochondrial activities [85,86] (also in synaptic regions [87]), and decreased astrocyte glutamine formation via reducing glutamine synthetase enzyme to scavenge ammonia [88]. Urea itself, if elevated, is neurotoxic [89]. Within the brain, urea is also produced from transaminase activities in neurons and astrocytes, involving alanine, aspartic acid, glutamate [52].
Some potential mechanisms underlying reduced CSF AA levels are summarized in Figure 3. In summary, increased AA utilization in the AD brain might over time reduce the balance between positive effects (energy production) and negative ones (hyperammonia, high urea levels, increased synthesis of AD markers), resulting in abnormal neuron and glia death [31].

Some Potential, Specific Mechanisms for Reduced CSF AA Levels Independently of Patients' Nutritional Status
The reductions in CSF asparagine, serine, glutamine, glycine, arginine, tryptophan, phenylalanine and Trp ratio were independent of patient nutritional state, suggesting that they were linked to AD per se; the brain may also use these AAs for specific metabolic tasks.
Low CSF serine and glycine confirm the findings from a post-mortem study [52] documenting the decrease in these AAs in the hippocampus, entorhinal cortex, middle-temporal gyrus, sensory cortex, motor cortex, cingulate gyrus, and cerebellum. Among the factors responsible for their reduction, increased utilization in the glycolytic pathway to generate energy, to increase/maintain glutathione system [90] and for neurological functions might play a major role. In the brain, serine can be converted into the stereoisomer D-serine, which is a potent activator of the NMDA glutamate receptor, which in turn is involved in the formation of new synapses and essential for learning and memory . Some general potential mechanisms underlying brain amino acid overutilization and reduced CSF amino acid levels. [83,84] impaired mitochondrial activities [85,86] (also in synaptic regions [87]), and decreased astrocyte glutamine formation via reducing glutamine synthetase enzyme to scavenge ammonia [88]. Urea itself, if elevated, is neurotoxic [89]. Within the brain, urea is also produced from transaminase activities in neurons and astrocytes, involving alanine, aspartic acid, glutamate [52].
Some potential mechanisms underlying reduced CSF AA levels are summarized in Figure 3.

Some Potential, Specific Mechanisms for Reduced CSF AA Levels Independently of Patients' Nutritional Status
The reductions in CSF asparagine, serine, glutamine, glycine, arginine, tryptophan, phenylalanine and Trp ratio were independent of patient nutritional state, suggesting that they were linked to AD per se; the brain may also use these AAs for specific metabolic tasks.
Low CSF serine and glycine confirm the findings from a post-mortem study [52] doc-umenting the decrease in these AAs in the hippocampus, entorhinal cortex, middletem-poral gyrus, sensory cortex, motor cortex, cingulate gyrus, and cerebellum. Among the factors responsible for their reduction, increased utilization in the glycolytic pathway to generate energy, to increase/maintain glutathione system [90] and for neurological func-tions might play a major role. In the brain, serine can be converted into the stereoisomer D-serine, which is a potent activator of the NMDA glutamate receptor, which in turn is involved in the formation of new synapses and essential for learning and memory increase [83,84] impaired mitochondrial activities [85,86] (also in synaptic regions [87]), and decreased astrocyte glutamine formation via reducing glutamine synthetase enzyme to scavenge ammonia [88]. Urea itself, if elevated, is neurotoxic [89]. Within the brain, urea is also produced from transaminase activities in neurons and astrocytes, involving alanine, aspartic acid, glutamate [52].
Some potential mechanisms underlying reduced CSF AA levels are summarized in Figure 3.

Some Potential, Specific Mechanisms for Reduced CSF AA Levels Independently of Patients' Nutritional Status
The reductions in CSF asparagine, serine, glutamine, glycine, arginine, tryptophan, phenylalanine and Trp ratio were independent of patient nutritional state, suggesting that they were linked to AD per se; the brain may also use these AAs for specific metabolic tasks.
Low CSF serine and glycine confirm the findings from a post-mortem study [52] doc-umenting the decrease in these AAs in the hippocampus, entorhinal cortex, middletem-poral gyrus, sensory cortex, motor cortex, cingulate gyrus, and cerebellum. Among the factors responsible for their reduction, increased utilization in the glycolytic pathway to generate energy, to increase/maintain glutathione system [90] and for neurological func-tions might play a major role. In the brain, serine can be converted into the stereoisomer D-serine, which is a potent activator of the NMDA glutamate receptor, which in turn is involved in the formation of new synapses and essential for learning and memory decrease. BBB: Blood-Brain-Barrier.
In summary, increased AA utilization in the AD brain might over time reduce the balance between positive effects (energy production) and negative ones (hyperammonia, high urea levels, increased synthesis of AD markers), resulting in abnormal neuron and glia death [31].

Some Potential, Specific Mechanisms for Reduced CSF AA Levels Independently of Patients' Nutritional Status
The reductions in CSF asparagine, serine, glutamine, glycine, arginine, tryptophan, phenylalanine and Trp ratio were independent of patient nutritional state, suggesting that they were linked to AD per se; the brain may also use these AAs for specific metabolic tasks.
Low CSF serine and glycine confirm the findings from a post-mortem study [52] documenting the decrease in these AAs in the hippocampus, entorhinal cortex, middletemporal gyrus, sensory cortex, motor cortex, cingulate gyrus, and cerebellum. Among the factors responsible for their reduction, increased utilization in the glycolytic pathway to generate energy, to increase/maintain glutathione system [90] and for neurological functions might play a major role. In the brain, serine can be converted into the stereoisomer D-serine, which is a potent activator of the NMDA glutamate receptor, which in turn is involved in the formation of new synapses and essential for learning and memory processes [46,91,92]. The importance of serine for brain activity may be inferred by the very strong correlation found in this study between plasma and CSF serine levels. Impaired serine production in astrocytes contributes to cognitive dysfunction in AD [92].
Glycine is normally produced by the brain [40]. Its reduction [36] may be due to both reduced formation from its immediate precursor, serine, and increased utilization in the glutathione system pathways. Low brain glycine may contribute to synaptic and brain toxicity, as this AA functions in the brain as a inhibitory neurotransmitter, similarly to GABA, which may be reduced in AD CSF [42].
Low asparagine could be due to reduced conversion of oxalacetate [93] from an altered TCA cycle. Its reduction may negatively impact cell proliferation and the urea cycle, given that this AA is a powerful stimulator of ornithine decarboxylase and cell proliferation [94].
Low CSF glutamine levels were in line with some studies [95,96] and contrasted with a recent investigation [97]. Glutamine is normally produced by brain [40] and its reduction is not due to lack of glutamate (this being similar between AD and CTRL) but instead may derive from both increased utilization for energy production [96] in the mitochondrial TCA cycle [98] and reduced synthesis due to impaired activity of the oxidized glutamine synthase enzyme (GS) [9,99,100]. Of note, the compromising of GS activity occurs early in AD [101]. Low brain glutamine in astrocytes reduces the capacity for detoxifying ammonia and leads to reduced Aβ autophagy capacity [101][102][103]. Under normal conditions, glutamine is neuroprotective [101]. Supporting this, CSF glutamine and glutamate are inversely correlated with the amyloid tau index [97], and GS is negatively associated with t-tau and p-tau [104]. Glutamine CSF/plasma ratio was >1 in CTRL, indicating the importance of brain production of this amino acid under normal conditions. This suggests that glutamine is exported from the brain rather than imported into the brain from plasma.
Low CSF arginine [42] may derive from its overconsumption in increased urea cycle activity, as well as in the enzymatic activities of nitric oxide synthase (NOS) and arginase where arginine serves as substrate. NOS in AD is upregulated [105] and elevated production of NO is responsible for increased nitrosative stress [106], causing mitochondrial dysfunction and the degeneration of both synapses and neurons [107]. Another arginineconsuming reaction is the production of the inhibitory neurotransmitter GABA through its conversion to ornithine [108]. Low arginine leads to disruption of the urea cycle, and consequently to hyper-ammonia, endothelial dysfunction and reduced synthesis of creatine and creatine kinase BB, which results in worsening neuronal energy deficits.
In the case of AD, the only CSF amino acid that significantly correlated with insulinemia was arginine. This correlation may be explained by the hyperinsulinemia-induced upregulation of brain arginase activity, as observed in obesity [109].
Increased arginase activity consumes arginine and reduces NO synthesis, as arginase competes with NOS for the same arginine substrate. In brief, whatever the prevalent mechanisms, the upregulation of the two enzymes (NOS and arginase) competing for the same substrate (arginine) greatly contributes to reducing brain arginine availability.
Two Essential Amino Acids (EAAs), tryptophan and phenylalanine, proved to be decreased in AD CSF. Being precursors of the neurotransmitter serotonin and catecholamines dopamine, norepinephrine, epinephrine, and tyramine, their reduction suggests that brain deficits in neurochemical transmission contribute to synapse dysfunction and the loss of synapses, neurons and neuronal networks [50].
The alterations in chemical neurotransmitters occur early in AD, before the pathological markers of AD or the loss of neurons [50].
The decrease in tryptophan may be attributed to activation of the kynurenine pathway [110], which is upregulated in AD [111,112]. The diversion of tryptophan towards the kynurenine pathways reduces serotonin formation, contributing to AD pathogenesis. Another potential mechanism for reduced tryptophan may be its increased utilization for brain protein synthesis, given the important role played by this amino acid in protein synthesis [50]. The loss of serotoninergic neurons correlates with AD severity and learning and memory impairments [50].
Previous investigations reported that phenylalanine was increased in the brain [52] and decreased in serum [113] of AD patients. These findings are the opposite of the results of the current investigation. An explanation for these contradictory results may be that the previous studies were carried out on two different AD populations and using two different methodologies, one of which involved examination of post-mortem patient brains [52]. The measurements of phenylalanine levels in plasma and CSF in the present study were concomitantly carried out on a unique AD group. This explanation is also valid for tryptophan, which, in the above studies, was decreased in serum and increased in the brain.
Low phenylalanine levels may result from its use for protein synthesis [50], increased breakdown in the TCA cycle for energy production, and residual catecholamine formation [114]. The depletion of catecholamine neurotransmitters impairs behavioral and cognitive functioning, in particular under stressful conditions and during aging [1]. Impaired dopamine neurotransmission causes low performance in the spatial working memory and spatial recognition memory [115,116]. Moreover, reduced brain catecholamine availability increases patient perception of fatigue [117].
Low Trp ratio suggests a greater prevalence of the deficit in serotoninergic transmission compared to the catecholaminergic deficit.

Some Potential, Specific Mechanisms Underlying Reduced CSF Levels in Nutritionally-Deteriorated AD
In nutritionally deteriorated ADs, CSF levels of six EAAs (the three BCAAs plus threonine, methionine, lysine), of alanine, and of TAAs and EAAs as a group were reduced. Leucine, isoleucine, alanine, TAAs, and EAAs were altered both in plasma and CSF. Thus, the brain availability of these AAs, in particular the essential leucine and isoleucine, was in part conditioned by their limited supply in plasma [118,119], and hence by abnormalities of metabolism in extracerebral areas [46]. By contrast, brain reductions in threonine, lysine, and valine would be due to their brain overutilization (plasma methionine was not available; thus, the level of this AA can be discussed only in terms of brain content).
Chronic protein breakdown [46] and/or insufficient dietary BCAA intake may explain the lack of leucine and isoleucine in both plasma and CSF. The importance of circulating BCAA for brain BCAA availability is highlighted by the correlations between plasma and CSF BCAA found in the current investigation, both in CTRL and in ADs. The strong correlations of the BCAA/TAAs ratio and valine between plasma and CSF both in controls and in ADs indicate that the higher the plasma levels of BCAAs among the circulating AAs, the higher their contribution to AA levels in the CSF compartment.
The differences in degree in the above correlations between ADs and controls for BCAA/TAAs ratio and valine might confirm that, in AD, an important role may be played by brain overconsumption of BCAA for both protein synthesis and (in particular for valine) for TCA cycle activity for energy generation.
Brain BCAA overutilization could occur to facilitate the synthesis of phosphorylated tau protein, the main component of neurofibrillary tangles, due to increases in the process of mTOR autophosphorylation [61]. Increased mTOR activation may be induced by Aβ levels [63] and by BCAA themselves, mainly leucine [120], which are potent activators of mTOR signaling for protein synthesis.
The reductions in leucine and valine in CSF confirm the findings from a previous study [36], which also documented a correlation between plasma and CSF valine levels and memory and cognitive functions.
Based on the background of existing literature, it is unclear whether BCAA may potentially exert neuroprotective or toxic effects.
Some experimental investigations reported unfavorable effects on the development and progression of AD [121,122]. In the AD mouse model, tau hyperphosphorylation was decreased by a protein-restricted diet [123]. In contrast, clinical studies reported favorable effects on the retrieval of neurocognitive and neuropsychological functions after BCAA supplementation in patients with cirrhosis and chronic hepatic encephalopathy [124,125], with latent portosystemic encephalopathy [126], with phenylketonuria [127,128], and patients with severe traumatic brain injury undergoing rehabilitation [129] or in a minimally conscious state following traumatic brain injury [130]. Neurocognitive retrieval after BCAA supplementation was also found in an animal model of traumatic brain injury [131]. Cognitive performance was improved after BCAA supplementation in healthy exercising subjects [132,133]. In a large epidemiological study, low plasma BCAA levels were found to be independent predictors of incident dementia [134].
We postulate that reduced brain BCAA availability may accelerate the progression of AD for at least five main reasons. First, low brain BCAA might impair the turnover of glutamate, the main excitatory neurotransmitter in the brain [135,136] as nitrogen from BCAA is used for the formation of glutamate in astroglia [137,138]. The nitrogen from BCAA is particularly important, as the content of aspartate, the other nitrogen donor, is limited in astroglia [139].
Second, low brain BCAA levels may increase neurotoxicity and neurodegeneration by limiting the glutamate-glutamine cycle, leading to accumulations of both glutamate and ammonia [125]. Third, BCAA modulate the balance between excitatory and inhibitory activities by regulating the pools of glutamate, glutamine, and GABA [140]. Fourth, low brain BCAA may reduce protein synthesis for brain tissue repair, sprouting, and circuitry remodeling [141]. Lastly, low BCAA, in particular leucine, may limit the recovery of cognition by reducing insulin production [120].
Of note, in CTRL subjects, the BCAA/EAA CSF/plasma ratio was >1, supporting, under normal conditions, the importance for the brain of maintaining high levels of BCAA among the EAAs. The decreased CSF levels of EAAs lysine and threonine, despite their normal plasma concentrations, may be secondary to increased brain oxidation and/or increased incorporation into brain proteins. These two amino acids are the most essential among EAAs, since their oxidation is irreversible, as they do not undergo transamination activities [142,143]. Consequently, low brain availability of lysine and threonine creates increased risk for brain synthesis activities. Moreover, lysine has potent antiproteolytic activity [144] even under conditions of a low protein diet [145], and indirectly modulates GABAergic transmission by means of its catabolic product, pipecolic acid [113].
Neural plasticity is regulated by protein phosphorylation occurring in serine, threonine, and tyrosine residues (protein serine-threonine kinase) [146]. The neuronal proteins regulated by protein serine-threonine kinase include enzymes for neurotransmitter biosynthesis and degradation, neurotransmitter receptors and transporters, ion channels, and proteins involved in neuronal growth and motility, to name only a few. These physiological serine-threonine kinase activities lead us to suspect that a lack of brain threonine and serine may have a very negative impact on residual brain function in AD.
Reduced alanine may be due to an altered glycolysis pathway, leading to decreased availability of pyruvate to be converted into alanine via the transamination process. Altered brain methionine metabolism may explain the reduced CSF levels of methionine found in the current study and may also explain the increased homocysteine (a by-product of methionine) levels in AD serum and brains [147] and the association of elevated serum homocysteine levels with a decreased methionine cycle in AD [95].
As, in the current study, plasma methionine could not be detected, we cannot exclude a reduced contribution of plasma methionine to decreased CSF levels. Another mechanism leading to reduced CSF methionine may be its incorporation into proteins, since it is essential for the initiation of protein and nucleic acid synthesis and its active form, S-AdenosylMethionine (SAME), is the most important methyl donor in the body. Methylation processes via SAME stabilize and protect proteins, including myelin, and are involved in the formation of creatine, carnitine, melatonine, and polyamines and in the metabolism of serotonin. Another metabolic condition leading to low methionine may be excessive amino acid conversion to cysteine (and taurine), an important precursor of the antioxidant glutathione, which may undergo a chronic decrease in AD due to high brain ROS formation [148]. Table 14 shows some potential negative effects on residual brain activities and functions in the study ADs resulting from reduced brain AA availability. Table 14. Some potential impacts on brain metabolic activities due to reduced brain AAs.

Reduced CSF AAs
Effects on Brain Metabolism Effects on Brain Activities

Were the Brains of AD Patients Undernourished? Suggestions for Clinical Practice
Under physiologic conditions, the brains of normal adults would seem to be spared from malnutrition because of very low cell proliferation and cell replacement in the brain [149], low cellular protein turnover [150], and the supply of amino acids to the brain from extracerebral tissues, such as skeletal muscle and liver [151].
However, the findings from more recent studies seem to question the assumption that the brain could be protected from malnutrition. First, the brain protein synthesis rate in normal adults was shown to be higher than that of skeletal muscle tissue [152]: brain protein synthesis was 0.17 ± 0.01%/h in the neocortex and 0.13 ± 0.01%/h in the hippocampus, but only 0.05 ± 0.01%/h in skeletal muscle.
The high rates of brain protein synthesis were documented in other previous studies [153,154]. Second, under protein deficiency conditions, the brain utilizes circulating amino acids from protein breakdown occurring in extracerebral body areas, mainly the skeletal muscle [155]. Of interest, sarcopenia in AD patients has been reported to be associated with brain atrophy and altered cognitive performance [55] and, in elderly subjects, with cognitive impairment. Third, the current study is highly suggestive of a lack of EAAs (tryptophan, phenylalanine) in both normonurished ADs and nutritionally deteriorated ones. In addition, the lack of NON EAAs, mainly in nutritionally deteriorated ADs, contributes to reducing the availability of nitrogen for protein synthesis [156].
Thus, we postulate that the brain of ADs, in particular ADs with MNA < 24, might be considered an "undernourished brain", caused by an unbalanced ratio of brain AA utilization to AA supply from plasma. Theoretically, brain AA overconsumption could lead or contribute over time to impairment of circulating EAAs, particularly in patients with poor or marginal EAA intakes and in sarcopenic patients.
A problem arising from the current study is whether it is possible to reconcile two conflicting factors: the brain's needs for EAAs/AAs to sustain metabolic processes such as repairing, circuit remodeling, and growth of the brain's residual functioning areas and, at the same time, the need to avoid enhanced EAA utilization, which results in the synthesis of AD neuropathology markers. It has been hypothesized that a nutritional strategy could involve providing patients with a diet low in protein but with high levels of BCAAs, aromatic amino acids, glutamine, histidine, and threonine [157,158]. Another hypothesized nutritional treatment could be omitting leucine but augmenting the supply of valine and isoleucine [31], which are less potent stimulators of mTOR than leucine [120]. An additional reported nutritional treatment could be a reduction in methionine intake, methionine being responsible for the increase in Aβ and phosphorylated tau levels in the brain [157]. In mouse AD models, methionine restriction was shown to be neuroprotective by diminishing the levels of Aβ formed [159].
Until investigations address these issues, a reasonable approach in clinical practice for patients with AD at present might be the following: 1.
Routine determinations of AA levels, both in plasma and CSF, in ADs.

2.
Quantification of dietary protein/EAA intakes since the first diagnosis and frequent monitoring over time.

3.
Improving physical activity and exercise, aiming at the preservation of/increase in skeletal muscle mass, strength, and function, since muscle tissue is the main body store of AAs.

4.
In nutritionally deteriorated patients (MNA < 24 scores), who represented more than half of the ADs considered in this study, a supplementation of EAAs to restore their plasma levels may be needed as EAAs are also vital for maintenance of the structures and functions of extracerebral organs and tissues. Of note, supplementation of the diet with L-serine prevents synaptic loss and behavioral deficits in AD mice [92]. Future studies will address the question of whether, in addition to EAA supplementation, supplying lactate, pyruvate, and ketone bodies may reduce the brain's need for AAs to generate energy.

5.
All these clinical suggestions should be added to an antinflammatory and antioxidant diet [160], along with nutritional interventions aimed at reducing abnormalities in gut microbiota and slowing down the progression of the disease [161].
A well-planned investigation could examine whether, in ADs with normal nutrition (MNA > 24), the prevention of nutritional deterioration and correction of brain tryptophan and phenylalanine levels should be pursued. In these patients, supplementing these two aromatic amino acids might prove useful for improving their content in the brain.

Limitations of Study
The study has several limitations that require further research to be resolved. A limitation of the present investigation is the small number of subjects studied, particularly controls.
Urea levels were not determined, either in AD or CTRL CSF. Elevated CSF urea in AD would have corroborated the existence of excessive brain AA consumption.
Patients' intakes of dietary proteins were not quantified. Information regarding amount and type of ingested amino acids, in particular EAAs, would have been very useful for understanding the contribution of poor diet to reduced plasma and CSF EAAs in patients with nutrition deterioration, as well as to reduced CSF tryptophan and phenylalanine in normonurished AD.
Analysis of body composition with measurements of skeletal muscle tissue, the most important AA repository in the body, would have strengthened the discussion.
As a strength of the study, it is important to underline that the adoption of a simple nutritional test (MNA) may be useful in clinical practice for suggesting the existence of altered AA contents both in plasma [46] and CSF, in the absence of a direct measurement of EAAs in both areas.

Conclusions
The study documents that AD patients, compared to CTRL, had lower CSF AA levels and lower CSF/plasma AA ratios. The reductions in CSF AAs were in part linked to the disease per se, being independent of patients' nutritional status, and in part linked to deterioration in nutrition.
Future well-planned studies are needed to understand whether and how it is possible to safely improve CSF AA levels in advanced AD patients. Informed Consent Statement: All subjects or their legal representatives gave their informed consent for inclusion before they participated in the study. Participants did not receive financial compensation. Data Availability Statement: Data available on request due to restrictions, e.g., privacy or ethical. The data presented in this study are available on request from the corresponding author. The data are not publicly available due to the rules of Mondino Foundation but might be available only on request.

Conflicts of Interest:
The authors have no potential conflict of interest to declare. Roberto Aquilani is Scientific Consultant of Professional Dietetics, Milano, Italy.