**5. The role of brain lipids in preventing and treating Alzheimer's disease**

#### **5.1. Therapeutic approaches based on inhibitors of cholesterol biosynthesis**

Over the last 2 decades, the relationship between cholesterol levels and the risk of developing AD has become more evident, in turn encouraging the use of statins to treat or prevent AD (see Section 3.1.). Statins are a group of drugs used to treat hypercholesterolemia as they inhibit HMG-CoA reductase, the principal enzyme involved in cholesterol synthesis. In animal models of AD, simvastatin administration to guinea pigs decreased brain and CSF Aβ levels, an effect that is reversed by discontinuing the treatment [128]. By contrast, simvastatin failed to modify brain levels of Aβ in other studies but it improved the cognitive capacity of trans‐ genic AD mice [129]. Thus, it appears that simvastatin can possibly prevent cognitive decline in AD mice without affecting amyloidogenic APP processing, in turn suggesting that the amyloid pathology may be a consequence more than the primary causal agent of AD, possibly due to changes in membrane lipids. In another study, lovastatin and pravastatin reduced the amount of Aβ in the brains of AD mice, while simultaneously increasing the levels of sAPPα [130]. Therefore, the results of preclinical research into these drugs are encouraging, although the outcome of human studies has been inconsistent, in part due to the differences in study design and data analysis [131].

While several observational studies in human subjects support the hypothesis that statins may prevent AD development, other studies argue against such effects [132]. Nevertheless, some clinical trials are investigating the use of statins in AD, such as simvastatin or atorvastatin. The first trial to analyze the effect of simvastatin on cognitive scores and APP processing was completed in 2003. This clinical study was performed over 12 weeks on patients with AD, and it reported changes in APP metabolites in the CSF: sAPPα and sAPPβ levels were significantly reduced but not those of Aβ or tau. Remarkably, a significant cognitive improvement in response to simvastatin treatment was found in patients with AD [133]. Unexpectedly, subsequent results based on a 12 month treatment failed to show such cognitive improvements in the same patients, even though cholesterol metabolism was altered in the brain [134]. Unfortunately, a later larger trial performed on 406 mild-to-moderate AD patients also failed to identify clinical benefits of simvastatin (the multicenter CLASP trial). This CLASP trial (clinicaltrials.gov ID: #NCT00053599) evaluated the safety and efficacy of an 18 month treatment with simvastatin to prevent AD progression. Once again, simvastatin treatment lowered lipid levels but it did not slow the progressive AD-related decline in cognitive performance [135]. Despite the apparent lack of clinical improvement on cognition in patients with AD, the University of Wisconsin (Madison, USA) evaluated simvastatin in cognitively normal people at risk of developing FAD. This study (ESPRIT study: clinicaltrials.gov ID: #NCT00486044), compared the changes in CSF Aβ and cognitive scores following simvastatin or placebo administration, as well as markers of cholesterol metabolism and inflammation. Again, no specific effect of simvastatin was observed on CSF Aβ or tau levels but a improve‐ ment in terms of cognitive performance was reported [136]. As a result, a follow-up study attempted to evaluate similar outcome measures after a longer course of simvastatin (the SHARP study; clinicaltrials.gov ID: #NCT00939822). Additional clinical trials with a more precise methodological design are also being developed to define the clinical efficacy of simvastatin. For instance, the SIMaMCI study (clinicaltrials.gov ID: #NCT00842920) on 445 subjects assesses the time until participants suffer phenoconversion to dementia, with conversion being defined as an increase in the Clinical Dementia Rating (CDR) score above 0.5. The trial also focuses on the change in cognitive scores from a healthy state to MCI and dementia.

Other clinical studies have assessed atorvastatin, lovastatin, and pravastatin in AD. The only clinical trial showing cognitive improvement associated with atorvastatin administration was a phase-2 pilot study comparing a 1-year course of atorvastatin to a placebo in patients with mild-to-moderate-AD who were also taking a cholinesterase inhibitor and vitamin E (clini‐ caltrials.gov: #NCT00024531). This study reported trends towards benefits on cognition and function [137, 138], leading to a larger phase-3 randomized trial involving 640 patients to confirm the potential clinical benefits of atorvastatin in patients with mild-to-moderate AD also treated with donepezil (the LEADe study; clinicaltrial.gov ID: #NCT00151502). Unfortu‐ nately, no clinical benefit was observed after 18 months of treatment [139, 140], and this was considered the definitive trial on atorvastatin regarding symptomatic AD treatment. It is worth noting that APP metabolites were not assessed in these studies and that decreased circulating cholesterol, as well as improved neurovascular response and cerebral blood flow were found in atorvastatin-treated patients with AD (clinicaltrials.gov: #NCT00751907) [141]. Lovastatin has been less frequently studied in randomized AD trials, and it was shown to be efficient in reducing serum Aβ levels in patients AD, although no cognitive evaluations were performed (clinicaltrial.gov: #NCT00046358) [142]. In the case of pravastatin, APP processing was not analyzed and the cognitive evaluation of treated patients revealed no significant improvement relative to the placebo group (clinicaltrial.gov: # NCT00303277) [143].

an effect that is reversed by discontinuing the treatment [128]. By contrast, simvastatin failed to modify brain levels of Aβ in other studies but it improved the cognitive capacity of trans‐ genic AD mice [129]. Thus, it appears that simvastatin can possibly prevent cognitive decline in AD mice without affecting amyloidogenic APP processing, in turn suggesting that the amyloid pathology may be a consequence more than the primary causal agent of AD, possibly due to changes in membrane lipids. In another study, lovastatin and pravastatin reduced the amount of Aβ in the brains of AD mice, while simultaneously increasing the levels of sAPPα [130]. Therefore, the results of preclinical research into these drugs are encouraging, although the outcome of human studies has been inconsistent, in part due to the differences in study

While several observational studies in human subjects support the hypothesis that statins may prevent AD development, other studies argue against such effects [132]. Nevertheless, some clinical trials are investigating the use of statins in AD, such as simvastatin or atorvastatin. The first trial to analyze the effect of simvastatin on cognitive scores and APP processing was completed in 2003. This clinical study was performed over 12 weeks on patients with AD, and it reported changes in APP metabolites in the CSF: sAPPα and sAPPβ levels were significantly reduced but not those of Aβ or tau. Remarkably, a significant cognitive improvement in response to simvastatin treatment was found in patients with AD [133]. Unexpectedly, subsequent results based on a 12 month treatment failed to show such cognitive improvements in the same patients, even though cholesterol metabolism was altered in the brain [134]. Unfortunately, a later larger trial performed on 406 mild-to-moderate AD patients also failed to identify clinical benefits of simvastatin (the multicenter CLASP trial). This CLASP trial (clinicaltrials.gov ID: #NCT00053599) evaluated the safety and efficacy of an 18 month treatment with simvastatin to prevent AD progression. Once again, simvastatin treatment lowered lipid levels but it did not slow the progressive AD-related decline in cognitive performance [135]. Despite the apparent lack of clinical improvement on cognition in patients with AD, the University of Wisconsin (Madison, USA) evaluated simvastatin in cognitively normal people at risk of developing FAD. This study (ESPRIT study: clinicaltrials.gov ID: #NCT00486044), compared the changes in CSF Aβ and cognitive scores following simvastatin or placebo administration, as well as markers of cholesterol metabolism and inflammation. Again, no specific effect of simvastatin was observed on CSF Aβ or tau levels but a improve‐ ment in terms of cognitive performance was reported [136]. As a result, a follow-up study attempted to evaluate similar outcome measures after a longer course of simvastatin (the SHARP study; clinicaltrials.gov ID: #NCT00939822). Additional clinical trials with a more precise methodological design are also being developed to define the clinical efficacy of simvastatin. For instance, the SIMaMCI study (clinicaltrials.gov ID: #NCT00842920) on 445 subjects assesses the time until participants suffer phenoconversion to dementia, with conversion being defined as an increase in the Clinical Dementia Rating (CDR) score above 0.5. The trial also focuses on the change in cognitive scores from a healthy state to MCI and

Other clinical studies have assessed atorvastatin, lovastatin, and pravastatin in AD. The only clinical trial showing cognitive improvement associated with atorvastatin administration was

design and data analysis [131].

148 Update on Dementia

dementia.

The substantial variability in outcome from these human studies makes it difficult to ascertain whether statins might have a beneficial role in preventing or treating AD. One possible reason to explain such inconsistency relates to the ability of statins to cross the BBB and enter the brain. In this respect, the chemical structure of statins can vary greatly, which justifies why some of them cross the BBB better than others. Accordingly, simvastatin and lovastatin appear to cross the BBB via passive diffusion, whereas pravastatin depends on an active transport system. Although this could justify the lack of clinical effect of pravastatin in clinical trials, it is also true that pravastatin reduced Aβ load in AD mice, suggesting that pravastatin does reach the brain and exert its pharmacological effects [130, 144]. In this sense, clinical studies have investigated different statins with substantial variation in BBB permeability, making it difficult to reconcile the conflicting findings in the literature.

Another confounding factor would be the AD patient's ApoE genotype which may affect the effectiveness of statins in AD prevention and treatment. In fact, individuals with the ApoE4 allele may experience less benefit from statin treatment in terms of cholesterol levels than others with the E2 or E3 alleles [145]. Therefore, although some trials in humans have taken the ApoE genotype into account, not all do. In addition, statins have a number of pleitropic effects on physiology and metabolism besides lowering cholesterol levels. For instance, statins can alter the expression of genes related to cell growth, signaling, trafficking, and apoptosis, which in turn can potentially affect the results of trials. In this sense, inhibition of HMG-CoA reductase activity can lead to decreased isoprenylation of proteins which in turn may cause a variety of downstream effects [146]. Thus, low isoprenoid levels may inhibit the secretory APP pathway leading to intracellular accumulation of APP metabolites that bias their analysis in the CSF or plasma [147].

In summary, cholesterol-lowering drugs such as statins have potential therapeutic effects for the treatment of AD. Based on preclinical studies in animal models and clinical trials in humans, statins represent a valuable group of compounds with promising therapeutic effects in AD. However, individual statins show different outcomes in terms of APP metabolism and cognitive improvement. In part, these disparities may be explained by the variability in BBB permeability and the different biochemical effects of these drugs observed to date.

#### **5.2. Therapeutic approaches based on PUFAs**

Neuroprotective effects of long-chain ω-3 PUFAs (see Section 3.2.) encouraged a number of clinical trials to assess the effects of ω-3 fatty acid administration to patients with AD over a defined time period, particularly focusing on the cognitive benefits of DHA and EPA. Inter‐ estingly, decreases in plasma DHA are associated with cognitive decline in healthy elderly adults and DHA administration to these patients improved the physiological memory loss and cognitive decline that frequently appears in the elderly [148] (clinicaltrials.gov ID: #NCT0027813). However, DHA administration to patients with AD did not significantly improve cognitive scores [80] (clinicaltrials.gov ID: #NCT00440050). Another randomized study involving administration of a commercially available fish oil as source of DHA and EPA only improved cognition in a small subgroup of patients with very mild cognitive dysfunction, with no clear beneficial effects in most patients [149] (clinicaltrials.gov ID: #NCT00211159). Finally, the most recent trial was carried out on a small group of patients with mild-tomoderate AD who were administered fish oil containing DHA and EPA. In this pilot study, significant recovery of cognitive capacity was evident in the patients treated with fish oil (with or without lipoic acid supplementation) [150] (clinicaltrials.gov ID: #NCT00090402). Together, these studies indicate that DHA supplementation may represent a plausible therapeutic approach for the treatment of the physiological age-related cognitive decline, although it is unclear what type of ω-3 PUFAs could be used to treat AD. Some of these discrepancies in the different randomized studies may reflect the source of the ω-3 PUFAs administered to the patients. As yet there is no consensus with regards the defined sources of ω-3 PUFAs or a standard ratio or dose of DHA and EPA: Quinn et al. [80] evaluated 2 g/day DHA, Freund-Levi et al. [149] evaluated the effect of fish oil administration with a DHA and EPA content of 1.7 and 0.6 g/day, respectively (EPAX 1500 TG; Pronova Biocare, Norway), and Shinto et al. [150] evaluated a fish oil daily dose containing 675 mg DHA and 975 mg EPA, the latter trial being the only efficacious treatment against AD in humans and having a different DHA:EPA ratio with respect to the former.

It is likely that differences in the source of ω-3 PUFAs together with variable DHA:EPA ratios might explain the variation in the results observed when treating AD patients with long-chain ω-3 PUFAs. Moreover, the presence of mercury in some fish oil supplements may provoke some neurological problems that could counteract the beneficial effects of DHA and related compounds. In this context, ω-3 PUFAs also exert their physiological function through the production of hydroxylated bioactive derivatives, such as NPD1 (see Section 3.2.). In fact, it has been demonstrated that NPD1 levels are dramatically reduced in the AD brain, even more so than DHA [68]. These data suggest that abnormally low levels of DHA in AD would be accompanied by impaired conversion of this fatty acid into NPD1 and other RVs. In fact, reduced levels of 15-LOX, the key enzyme involved in the generation of the D-series RVs and protectins, were observed in the brain of patients with AD, in turn demonstrating that lipid second messenger generation from DHA is impaired in AD [68]. Assuming that the conversion of DHA into hydroxylated derivatives is needed to mediate DHA-related physiological activity, such 15-LOX modifications could at least partially explain why DHA administration did not improve cognition in patients with AD. In this context, it is noteworthy that some cognitive improvement was observed when fish oil alone was used as the source of ω-3 PUFAs, suggesting that these oils might contain other PUFAs that impart neuroprotection independ‐ ently of DHA and EPA (hydroxylated PUFAs such as RVs or other PUFA derivatives) [81]. This hypothesis is supported by the high efficacy of HDHA(see below DHALifort) on cognitive score and by the aforementioned epidemiological meta-analysis showing an inverse correla‐ tion between AD incidence and fish oil intake but not with DHA/EPA (ω-3 PUFA) intake (see Section 3.2) [78].

in AD. However, individual statins show different outcomes in terms of APP metabolism and cognitive improvement. In part, these disparities may be explained by the variability in BBB

Neuroprotective effects of long-chain ω-3 PUFAs (see Section 3.2.) encouraged a number of clinical trials to assess the effects of ω-3 fatty acid administration to patients with AD over a defined time period, particularly focusing on the cognitive benefits of DHA and EPA. Inter‐ estingly, decreases in plasma DHA are associated with cognitive decline in healthy elderly adults and DHA administration to these patients improved the physiological memory loss and cognitive decline that frequently appears in the elderly [148] (clinicaltrials.gov ID: #NCT0027813). However, DHA administration to patients with AD did not significantly improve cognitive scores [80] (clinicaltrials.gov ID: #NCT00440050). Another randomized study involving administration of a commercially available fish oil as source of DHA and EPA only improved cognition in a small subgroup of patients with very mild cognitive dysfunction, with no clear beneficial effects in most patients [149] (clinicaltrials.gov ID: #NCT00211159). Finally, the most recent trial was carried out on a small group of patients with mild-tomoderate AD who were administered fish oil containing DHA and EPA. In this pilot study, significant recovery of cognitive capacity was evident in the patients treated with fish oil (with or without lipoic acid supplementation) [150] (clinicaltrials.gov ID: #NCT00090402). Together, these studies indicate that DHA supplementation may represent a plausible therapeutic approach for the treatment of the physiological age-related cognitive decline, although it is unclear what type of ω-3 PUFAs could be used to treat AD. Some of these discrepancies in the different randomized studies may reflect the source of the ω-3 PUFAs administered to the patients. As yet there is no consensus with regards the defined sources of ω-3 PUFAs or a standard ratio or dose of DHA and EPA: Quinn et al. [80] evaluated 2 g/day DHA, Freund-Levi et al. [149] evaluated the effect of fish oil administration with a DHA and EPA content of 1.7 and 0.6 g/day, respectively (EPAX 1500 TG; Pronova Biocare, Norway), and Shinto et al. [150] evaluated a fish oil daily dose containing 675 mg DHA and 975 mg EPA, the latter trial being the only efficacious treatment against AD in humans and having a different DHA:EPA

It is likely that differences in the source of ω-3 PUFAs together with variable DHA:EPA ratios might explain the variation in the results observed when treating AD patients with long-chain ω-3 PUFAs. Moreover, the presence of mercury in some fish oil supplements may provoke some neurological problems that could counteract the beneficial effects of DHA and related compounds. In this context, ω-3 PUFAs also exert their physiological function through the production of hydroxylated bioactive derivatives, such as NPD1 (see Section 3.2.). In fact, it has been demonstrated that NPD1 levels are dramatically reduced in the AD brain, even more so than DHA [68]. These data suggest that abnormally low levels of DHA in AD would be accompanied by impaired conversion of this fatty acid into NPD1 and other RVs. In fact, reduced levels of 15-LOX, the key enzyme involved in the generation of the D-series RVs and protectins, were observed in the brain of patients with AD, in turn demonstrating that lipid

permeability and the different biochemical effects of these drugs observed to date.

**5.2. Therapeutic approaches based on PUFAs**

150 Update on Dementia

ratio with respect to the former.

DHA-derived NPD1 produces many beneficial effects in animal and cell models of AD [75]. On the one hand, NPD1 suppresses Aβ42 peptide shedding by downregulating BACE-1 activity while enhancing α-secretase activity, thereby upregulating sAPPα levels and shifting the cleavage of APP from the amyloidogenic to the non-amyloidogenic pathway. Thus, NPD1 stimulated secretion of sAPPα strengthens neurotrophic signaling and prevents Aβ oligomer neurotoxicity, which may in turn be accompanied by a number of beneficial effects, such as the prevention of neuronal and axonal injury, improved neuronal plasticity, and enhanced learning memory [151–153]. In addition, like other RVs, NPD1 also displays anti-inflammatory properties. Indeed, NPD1 administration decreases Aβ42-triggered expression of the proinflammatory COX-2 and of B-94 (a TNF-α-inducible pro-inflammatory factor), and it prevents apoptosis in cultured cells by upregulating the expression of anti-apoptotic members of the Bcl-2 protein family.

The neuroprotective properties of NPD1 have encouraged the development of new pharma‐ cological approaches based on hydroxylated derivatives of ω-3 PUFAs to treat AD. Regardless of the use of natural RVs and protectins to treat inflammatory and neurodegenerative diseases [154], synthetic ω-3-PUFA bioactive hydroxyl derivatives have also been used to treat such disorders. This kind of therapeutic approach, aimed at modulating brain lipids to treat neurological diseases, is framed within so-called membrane lipid therapy (MLT) [155–157]. In this context, a novel hydroxylated derivative named HDHA (2-hydroxy-docosahexanoic acid) has been proposed as a promising therapeutic approach to treat AD. HDHA (DHALifort; PharmaConcept, Hungary) administration influences the brain lipid composition, increasing the PE species carrying long-chain PUFAs, which are significantly reduced in patients with AD (see Section 3.3.). Upon normalization of the membrane lipid composition by HDHA treatment, the membrane structure recovers the presence of liquid-disordered prone mem‐ brane structures [158] (**Figure 4**). These lipid changes are paralleled with a reduction in Aβ accumulation and tau hyperphosphorylation, and recovery of cognitive scores in a transgenic mouse model of AD (5xFAD mice) [159, 160] (see **Figure 4**).

HDHA also enhances the survival of neuron-like cells exposed to different insults, such as oligomeric Aβ and NMDA-mediated neurotoxicity (*in vitro*), and it promotes hippocampal neuronal cell proliferation in 5XFAD mice *in vivo* [159, 160], suggesting that HDHA induced neuroregeneration both *in vivo* and *in vitro*, which in part may explain its efficacy against neurodegeneration and memory loss. As part of its mechanism of action, HDHA dampens the binding affinity of oligomeric and fibrillar Aβ to lipid-raft membrane domains. Moreover, it enhances the unfolded protein response (UPR) and autophagy in neuron-like cells, which in turn may promote neuronal survival [160, 161]. In this sense, although the molecular role of autophagy in AD is complex and still largely unknown, it is thought that activation of salvage autophagy would avoid the intracellular accumulation of Aβ and its precursors by reducing the neuritic pathology (see **Figure 2**) [162, 163]. Therefore, the pleitropic effects of HDHA have proven beneficial to treat AD, suggesting that its molecular target is an upstream entity such as the membrane lipid bilayer. Thus, the normalization of the PE, DHA, cholesterol, and SM content mediated by HDHA would restore membrane lipid structure, which in turn would regulate amyloidogenic secretase activity tau phosphorylation and neuronal degeneration.

**Figure 4.** Proof of concept for the use of HDHA in AD mice and the proposed molecular mechanism of action. (A) Diagrams showing representative outlines of control and AD mice (5xFAD mice) that received HDHA or the vehicle alone, in the Radial Arm Maze test (RAM). A black point at the end of one arm represents where the mice find a food pellet. (B) Quantitative analysis of test performance is addressed by quantifying working (reentry of an arm already visited) and reference (entry into an unbaited arm) memory errors. Both parameters increased significantly in AD mice while HDHA treatment prevented such behavioral impairment until cognitive scores were almost totally reverted to those of the controls. Bars represent the mean ± SEM. One-way ANOVA followed by Bonferroni's post hoc test: \*p < 0.05, difference relative to healthy controls; #: p < 0.05 difference relative to the untreated AD group. C) Postulated mechanism of action for HDHA. HDHA enriches brain membranes in PE carrying DHA and other long PUFAs. These lipid changes may influence the structure of the cell membrane by promoting the appearance of liquid-disordered prone structures and potentially preventing AD-related cell signaling by: (i) downregulating APP amyloidogenic proc‐ essing and Aβ-induced tau protein hyperphosphorylation; and (ii) decreasing neuron vulnerability to extracellular tox‐ ic agents such as oligomeric Aβ. Together, this evidence supports a neuroprotective role of HDHA that may be associated with the improved cognitive capabilities observed in AD mice. Adapted from [159, 160].

Interestingly, the cellular heat shock response (HSP) depends on the plasma membrane composition, such that increased membrane fluidity is related to enhanced expression of heatshock proteins (HSP) [164]. In this context, these proteins (particularly Hsp70, Hsp60, and Hsp27) are involved in the mechanism of action of lithium in compacting Aβ plaques, lowering the density of dystrophic neurites and preventing neuronal degeneration in a mouse model of AD [119]. Therefore, lipid derivatives like HDHA that enhance membrane fluidity might also reduce the neurite pathology and prevent neuronal loss in AD via a mechanism involving Hsp expression. Regardless of amyloid production and the neuritic pathology, inflammation is also a key player in AD. In this sense, another synthetic hydroxyl derivative of ARA, 2-HARA (2 hydroxy-arachidonic acid) is a COX-1 and COX-2 inhibitor [165]. The inhibitory effect over COX-1 has been related to alternative microglia activation, as well as reduced Aβ production and tau hyperphosphorylation in a transgenic model of AD [166]. Thus, 2-HARA may be a promising therapeutic approach to mitigate the inflammatory component of AD, driving microglia activation towards an alternative neuroprotective phenotype, and reducing ADrelated amyloid and tau pathologies. To summarize, MLT is a therapeutic concept targeting membrane lipids that could be used to treat neurological disorders such as AD. In this context, recent findings about ω-3 PUFA RV-like mediators, such as HDHA and 2-HARA, offer a wide range of possibilities to design new bioactive compounds to treat neurodegenerative diseases.
