Neuroprotective strategies involving ROS in Alzheimer’s disease

Abstract

Alzheimer’s disease (AD) is a neurodegenerative disorder in which oxidative stress is a key hallmark. It occurs early in disease pathogenesis and can exacerbate its progression. Several causes of oxidative stress have been determined over the years. First, mitochondria play an important role in the generation and accumulation of free radicals. In addition to mitochondria, inflammation can also induce oxidative damage, especially via microglia, and microglia are also important for Aβ clearance. In AD, both mitochondrial function and inflammatory response are affected, leading to increased ROS formation and oxidative damage to lipid, proteins and nucleic acids. Some other sources have also been identified.

From these findings, various neuroprotective strategies against ROS-mediated damages have been elaborated in AD research. This review recapitulates some of the major strategies used to prevent oxidative stress and disease progression. Outcomes from in vitro and in vivo studies using models of AD are encouraging. However, only a few clinical trials have provided positive results in terms of slowing down cognitive decline.

Nonetheless, there is still hope for improved compounds that would better target pathways implicated in ROS production. In fact, facilitating the endogenous antioxidant system by modulating transcription has great promise for AD therapy.

Introduction

Alzheimer’s disease (AD) is characterized clinically by progressive cognitive decline and neuropathologically by the presence of amyloid plaques and neurofibrillary tangles. In this neurodegenerative disease, ageing is the most critical risk factor. In addition, oxidative stress is another key feature [1]. Numerous studies have reported the presence of elevated DNA [2, 3], RNA [4, 5], lipid [6, 7] and protein oxidation [8] in brains of subjects with AD and mild cognitive impairment (MCI) [9], suggesting that oxidative stress is an early event in AD pathogenesis [10]. Remarkably, these oxidative stress hallmarks were also observed in transgenic mouse models of AD, in which markers of lipid and protein oxidation are increased, which may precede amyloid deposition [11]. As in human disease, oxidative stress occurs at early stages, prior to the appearance of amyloid plaques [12, 13] and neurofibrillary tangles [11].

Generation and accumulation of reactive oxygen species (ROS) and reactive nitrogen species (RNS) are detrimental to cells in vitro and in vivo [14], and promote cell death [15]. Therefore, it has been crucial to investigate potential causes of oxidative stress in AD research [16]. There is a large body of evidence demonstrating involvement of mitochondria [17, 18], redox-active metals [19, 20], inflammation via activated microglia [21, 22], and other ROS-mediated pathways (figure 1). These important findings have led to the development of therapeutic strategies to counteract and prevent oxidative damage [23]. Here, we will review some of the major neuroprotective strategies involving ROS in AD, focusing on strategies targeting mitochondria and other potent antioxidant-related pathways.

Figure 1. Scheme of the generation and role of free radicals in AD.

In cells, free radicals can be generated by 2 major sources: mitochondria and NAPDH oxidase. Several key players, such as metals and/or Aβ can exacerbate their production. Once accumulating inside the cells, free radicals can cause lipid, protein, DNA and RNA damage that can exacerbate AD pathogenesis.

Neuroprotective strategies targeting mitochondria

Mitochondria are major sources of ROS in the central nervous system (figure 2). They contain redox carriers that can transfer single electrons to oxygen, thus generating the ROS superoxide (O₂⁻). Enzymes of the tricarboxylic acid cycle (TCA), of the electron transport chain (complex I, II and III) (figure 2A), and monoamine oxidases are among the mitochondrial redox carriers generating superoxide (figure 2B). Mitochondria also contain other enzymes able to detoxify ROS. Indeed, superoxide is depleted following a dismutation reaction by superoxide dismutase (SOD) and transformed into hydrogen peroxide (H₂O₂) (figure 2C). SOD enzymes work in conjunction with catalases and glutathione peroxidases to remove H₂O₂ within mitochondria. In addition, in the cell, O₂⁻ and H₂O₂ can react with other molecules such as redox-active metals (Fenton’s reaction involving iron) and nitric oxide, leading to the formation of hydroxyl radicals and peroxynitrites respectively (figure 2C). In normal conditions, these chemical events require an accurate balance between ROS production and removal. With ageing and/or AD, this balance is markedly disrupted leading to ROS accumulation and oxidative damage.

Figure 2. Mitochondria and ROS.

(A) Scheme of the mitochondrial electron transport chain. Electrons are transferred from complex I (C-I) to complex IV (C-IV), including coenzyme Q10 (Q) and cytochrome c (Cyt c). (B) Scheme of mitochondria-mediated ROS from the complex I (C-I), complex II (C-II), and III (C-III) and from the tricarboxylic acid cycle (TCA). (C) Chemical reactions for the generation of reactive oxygen species (ROS) such as superoxide (O₂⁻), hydrogen peroxide (H₂O₂), reactive nitrogen species (RNS) such as peroxynitrite, and the chemistry of the Fenton reaction, which generates OH⁻ radicals.

Increased numbers of mutations of mitochondrial DNA have been found in AD [24], as have increased concentrations of 8-hydroxy-2-deoxyguanosine, a marker of oxidative damage to DNA [25]. These deletions or point mutations, which may result from oxidative stress, can cause mitochondrial dysfunction and trigger apoptotic cell death [26]. In addition to DNA damage, several mitochondrial key enzymes involved in ROS detoxification are also affected. In human AD brains, levels of the alpha-ketoglutarate dehydrogenase complex (KGDHC) [27, 28], pyruvate dehydrogenase complex (PDHC) [29], and cytochrome oxidase (COX) [30, 31] are markedly reduced. Studies on animal models of AD have also implicated mitochondria in disease pathogenesis.

Our group demonstrated that partial genetic deletion of dihydrolipoyl succinyltransferase (one of the KGDHC subunits) increased amyloid pathology, oxidative stress and enhanced memory deficit in transgenic AD female mice [32]. Deficiency of manganese superoxide dismutase (MnSOD) also increased amyloid deposition [33], tau phosphorylation [34] and behavioral deficits [35]. Conversely, overexpression of MnSOD reduced amyloid deposition, oxidative stress, and synaptic and memory deficits in two different transgenic mouse models of AD [36, 37]. It also improved cerebral blood flow and axonal transport in transgenic AD mice [38]. These results indicate that detoxification enzymes localized within mitochondria are essential to prevent free radical accumulation and oxidative stress in AD. Neuroprotection in transgenic AD mice was also reported after genetic deletion of cyclophilin D, a component of the mitochondrial permeability transition pore [39–41]. Fission and fusion of mitochondria are also impaired in AD [42]. Importantly, it has been shown that mitochondrial dysfunction is preferentially located in affected AD brain regions, suggesting that Aβ and mitochondria are linked [43]. As potential mechanisms, recent work showed that both β-amyloid and tau mutations result in mitochondrial dysfunctions, and that there are synergistic effects on mitochondrial dysfunction mediated by the two proteins together [44, 45]. Also, oxidative stress can increase β-secretase expression and tau phosphorylation, through c-Jun amino-terminal kinase/p38 mitogen-activated protein kinase [46] and glycogen synthase kinase 3 [47] respectively. Therefore, several groups have focused their efforts on developing neuroprotective strategies targeting mitochondria.

Some of the major mitochondrial targets used as therapeutics against ROS-mediated damage were members of the quinone family. First, co-enzyme Q10 (CoQ10), also called ubiquinone (figure 3A; table 1), has demonstrated antioxidant and neuroprotective properties both in vitro and in vivo [48], therefore holding great promise in the treatment of neurodegenerative disorders [49]. CoQ10 is localized within mitochondria. It is part of the electron transport chain and acts as electron carrier from complex I and complex II during oxidative phosphorylation and transfers electrons to complex III (figure 2A). Interestingly, administration of CoQ10 in transgenic AD mice reduced amyloid plaque pathology [50, 51]. Unfortunately, CoQ10 has not yet been tested in AD patients. However, another quinone (synthetic analog of CoQ10), idebenone (figure 3B; table 1), has been investigated in clinical trials, for its ability to inhibit lipid peroxidation [52]. Although several smaller studies reported beneficial effects on memory and attention after several months of treatment [53–55], a larger study reported no effect in slowing disease progression [56]. A possible limitation to CoQ10 efficacy is that maintaining the coenzyme Q in its reduced antioxidant form (termed ubiquinol) requires an intact electron transport chain, which is impaired in AD.

Figure 3. Chemical structures of antioxidants.

Table 1.

Antioxidants and mechanisms of action

Name	Description	Main target	Mechanisms of action	Results Mouse	Results Human
Coenzyme Q10	-Ubiquinone family -Component of mitochondrial electron transport chain	Mitochondria	-Antioxidant within mitochondria -Electron carrier from complex I and II to complex III	+	nt
Idebenone	-Quinone family -Synthetic analog of Coenzyme Q10	Mitochondria	-Inhibition of lipid peroxidation	+	−
MitoQ	-Mitoquinone mesylate -CoQ10 + triphenylphosphonium ion -Adsorption in inner mitochondrial membrane	Mitochondria	-Same action as CoQ10		nt
Latrepirdine	-or Dimebon -Non-selective antihistaminic drug	Neurotransmission Mitochondria	-Blockage of neurotransmitter receptors: dopamine, serotonine, glutamate -Blockage of L-type calcium channel -Inhibition of Aβ-induced toxicity -Increase of succinate dehydrogenase activity, mitochondrial membrane potential, ATP	+	+/−
Acetyl-L-carnitine	-Acetylated form of L-carnitine	Mitochondria Nrf2	-Facilitation of glucose and lipid metabolism -Increase of α-secretase activity -Increase of mitochondrial enzyme levels -Facilitation of Nrf2 activation -Synergistic effect with R-α lipoic acid	+	+
R-α lipoic acid	-Organosulfur compound -Derived from octanoic acid	Mitochondria Nrf2	-Cofactor in aerobic metabolism for pyruvate dehydrogenase complex -Facilitation of Nrf2 activation -Synergistic effect with acetyl-L-carnitine	+	nt
Vitamin E	-α-tocopherol -lipid soluble antioxidant		-inhibition of lipid peroxidation	+	+/−
Pramipexole	-Non-ergoline dopaminergic agonist	Neurotransmission Mitochondria	-Agonist of D2, D3 and D4 dopaminergic receptors -Inhibition of cell death via mitochondrial pathway -Facilitation of mitochondrial membrane potential	+ (PD,ALS)	nt
SS-31	-Szeto-schiller small peptide -Adsorption in inner mitochondrial membrane	Mitochondria	-Scavenger of H2O2 -Inhibition of linoleic acid and low-density lipoprotein oxidation -Act via CD36 pathway -Increase neurite outgrowth -Inhibition of cyclophilin D	+ (PD,ALS)	nt
Clioquinol	-Hydroxyquonoline family -Antifungal and antiprotozoal drug	Metals	-Chelator of copper, zinc and iron -Inhibition of metal-induced free radical generation -Inhibition of Aβ aggregation	+	+
PBT2	-Second generation metal binding compound	Metals	-Chelator of copper, zinc and iron -Inhibition of metal-induced free radical generation -Inhibition of Aβ aggregation -Facilitation of Aβ clearance by increasing Aβ degrading components	+	+
Curcumin	-Curcumoid -Ginger family -Polyphenol	Inflammation Mitochondria	-Inhibition of lipid peroxidation -Inhibition of oxidative DNA damage -Increase of glutathione S-transferase and cytochrome P450 -Ability to bind Aβ -Inhibition of Aβ aggregation -Inhibition of tau phosphorylation	+	−
Ibuprofen	-Non-steroidal anti-inflammatory drug	Inflammation	-Inhibition of COX2 -Act via NOX pathway	+	−
Tarenflurbil	-R-flurbiprofen	Inflammation	-Inhibition of Aβ42 production -Inhibition of γ-secretase activity	+	−
Naproxen	-Non-steroidal anti-inflammatory drug	Inflammation	-Inhibition of COX2	+	−
Celecoxib	-Sulfa non-steroidal anti-inflammatory drug	Inflammation	-Inhibition of COX2	+	−
Apocynin	-Acetovanillone	NADPH oxidase	-Inhibition of NOX assembly -Interaction with SOD -ROS scavenger	+ (ALS)	nt
N6-(1-iminoethyl)-L-lysine	-Analog of L-arginine	iNOS	-Inhibition of iNOS	+	nt
Bezafibrate	-Fibrate drug -Agonist of PPAR	PPAR Mitochondria	-Activation of PGC-1α -Act via lipid metabolism	+ (PD)	nt
Pioglitazone	-Thiazolidinedione	PPAR Mitochondria	-Activation of PPAR-γ -Act via glucose and lipid metabolism -Ability to bind to mitochondrial protein MitoNEET	+	nt
Rosiglitazone	-Thiazolidinedione -antidiabetic drug	PPAR Mitochondria	-Activation of PPAR-γ -Act via glucose and lipid metabolism -Ability to bind to mitochondrial protein MitoNEET	+	+/−
Polyphenols	-Tannins and phenylpropanoids	Mitochondria Inflammation	-Antioxidant	+	nt
Triterpenoids	-CDDO -Nrf2/ARE activator	Nrf2/ARE	-Activation of ARE genes, including mitochondrial antioxidant genes -Inhibition of anti-inflammatory genes, including COX2 and iNOS	+	nt
Dimethyl fumarate	-Ester -α,β-unsaturated electrophilic compound -Nrf2/ARE activator	Nrf2/ARE	-React with glutathione by Michael addition	nt	+ (MS)

Legends:

“nt” = not tested

“+” = Beneficial effects on AD progression

“−“ = No beneficial effects on AD progression

“PD” = Parkinson’s disease

“ALS” = Amyotrophic Lateral Sclerosis

“MS” = Multiple Sclerosis

Another ubiquinone derivative, mitoquinone mesylate or mitoQ (figure 3C; table 1), has been used to prevent oxidative damage in AD [57]. MitoQ consists of CoQ10 linked to a triphenylphosphonium ion which has a positive charge. Therefore, it accumulates in mitochondria which have a strongly negative membrane potential (about −120mV). More precisely, mitoQ is adsorbed in the inner mitochondrial membrane facing the matrix. This ROS-enriched region provides a real potency to MitoQ. In addition, MitoQ can function independently of the electron transport chain. In non-neuronal cell cultures and isolated mitochondria, it reduced oxidative stress and cell death [58]. In order to test for its tolerance and potential side effects, wild-type mice were treated with mitoQ for 7 months. Data showed that it was well tolerated and without any adverse effect; in particular, mitoQ was not pro-oxidant [59]. Like CoQ10, mitoQ has not yet been tested in human clinical trials for AD. A clinical trial of mitoQ conducted in patients with Parkinson’s disease (PD) did not show any improvement [60].

Other mitochondrial antioxidants have been selected to determine their potential neuroprotective effects. Latrepirdine or Dimebon (figure 3D; table 1), first used as a nonselective antihistamine, showed promise in vitro for preventing ROS-mediated damage in neurodegenerative diseases [61, 62]. Latrepirdine has several mechanisms of action. First, it can act on a number of neurotransmitter receptors, such as serotoninergic, α-adrenergic [63] and glutamatergic receptors (NMDA and AMPA) [64]. Latrepirdine also blocked Aβ-induced toxicity and L-type calcium channel in cultured neurons [65]. In neuroblasma cells, administration of dimebon enhanced mitochondrial function under normal conditions by increasing succinate dehydrogenase activity, mitochondrial membrane potential, and ATP levels, which may exert indirect effects on ROS generation through mitochondria. It also protected against cell death under stress conditions [66]. Given acutely, dimebon increased Aβ secretion in neuroblastoma N2a cells and levels of Aβ₄₀ in the interstitial fluid of transgenic AD mice [67]. Several clinical trials have been conducted in AD patients. In a phase 2 trial, Dimebon was well tolerated and improved cognition, activities of daily living, behavior, and overall function in MCI and AD patients compared to placebo [68]. However, more recently, the phase 3 CONNECTION trial in AD patients revealed no benefit in any parameter [69, 70].

Acetyl-L-carnitine (ALCAR) (figure 3E; table 1) [71, 72] and R-alpha lipoic acid (LA) (figure 3F; table 1) [73] are also candidates as mitochondrial antioxidants. During exercise and in order to facilitate fatty acid utilization, L-carnitine and acetyl-CoA are converted into ALCAR within mitochondria by carnitine-O-acetyltransferase. Once transported outside mitochondria, the conversion is reversed. LA is an organosulfur compound derived from octanoic acid. LA is primarily a cofactor in aerobic metabolism for PDHC. In cells, LA is reduced to dihydrolipoic acid, its bioactive form providing its full antioxidant properties [74]. In combination with LA, ALCAR decreased ROS-mediated damage, mitochondrial dysfunction due to aging in rats, and improved cognitive and motor functions [75, 76]. Interestingly, in cell models of AD, administration of ALCAR increased alpha-secretase activity and physiological amyloid precursor protein (APP) metabolism, which can enhance the release of non-amyloidogenic fragments of APP [77]. Indeed, ALCAR increased levels of sAPPα and CTF-83, and decreased levels of CTF-99 APP fragments [77]. In addition, ALCAR and LA combined treatment reduced oxidative damage and improved cognitive behavior in normal mice maintained on vitamin-free, iron-enriched, oxidative-challenge diet [78]. The combination also improved mitochondrial structure and memory deficits in apoE4 mice [79]. As an additional possible mechanism of neuroprotection, both ALCAR [80] and LA [81] can restore levels of mitochondrial antioxidant enzymes, and increase nuclear translocation of the nuclear factor erythroid-related factor 2 (Nrf2) that can upregulate transcription of antioxidant genes. In a clinical trial done for one year, AD patients receiving ALCAR (without LA) had slower deterioration of cognition compared to placebo [82]. Thal and colleagues in 1996 found that ALCAR was effective only in early onset of AD compared to placebo [83]. In another study done by the same group, ALCAR did not show any benefits in early onset AD patients [84]. However, more recently, a meta-analysis of ALCAR treatment trials in AD patients slowed clinical scales and psychometric tests of MCI and AD patients [85], giving hope for the use of this drug.

Vitamin E (α-tocopherol) has also been used as an antioxidant in AD therapy (figure 3G; table 1). It is a lipid-soluble antioxidant that protects cell membranes from oxidation [86]. Indeed, by reacting with lipid radicals generated from lipid peroxidation, vitamin E inhibits formation of free radical intermediates, thus prevent complete oxidation. Its administration reduced lipid peroxidation in both young and aged transgenic AD mice, but reduced amyloid deposition only when the drug was administered at early ages [87]. Vitamin E slowed down the disease progression in AD patients as measured by an increase in the clinical dementia rating scale or time to institutionalization [88]. In a subsequent trial however, administration of vitamin E to patients with MCI did not prevent the patients from developing AD [89]. The ability of vitamin E to modify the course of AD is therefore controversial, but the results in MCI were disappointing.

Pramipexole (figure 3H; table 1), a dopaminergic agonist, in vitro reduced Aβ-induced caspase activation leading to cell death [90]. Independently of the dopamine, its effects also include reduction of ROS generation from mitochondria, and it has been shown to localize to mitochondria where it may exert its antioxidant effects [91]. By preventing mitochondrial-related cell death, pramipexole can maintain the mitochondrial membrane potential and therefore sustain mitochondrial function. In vivo, this drug was neuroprotective in models of PD by improving motor performances, reducing induced microglial activation and proteasomal inhibition, and by enhancing brain-derived neurotrophic factors and autophagy [92], giving hope for future trials in AD and other neurodegenerative diseases [93]. This drug was also protective in animal models of ALS and it was well tolerated in a phase 2 trial. A phase 3 trial in ALS patients was recently announced as a collaboration between Knopp Pharmaceuticals and Biogen.

We have also tried the effects of the Szeto-Schiller peptides (SS-31) which selectively localize to the inner mitochondrial membrane and produce antioxidant effects within mitochondria. Indeed, in vitro these peptides were able to scavenge H₂O₂, inhibit oxidation of linoleic acid and low-density lipoprotein (LDL), and diminish mitochondrial swelling [94]. As a mechanism of action, SS31 can target the CD36 pathway where it alters ligand levels, and ligand-receptor interactions. In ischemia-reperfusion, SS31 reduced CD36 expression, and ligand levels by inhibiting LDL peroxidation [95]. These small peptides are neuroprotective against MPTP [96] and in a transgenic mouse model of ALS [97]. SS-31 also protected against amyloid toxicity in vitro and in vivo by increasing neurite outgrowth, rescuing mitochondrial structure and function and decreasing cyclophilin D expression [98].

Antioxidant neuroprotective strategies targeting mitochondria have produced positive outcomes in vitro and in vivo. Most of the interventions produce clear antioxidant and protective effects. Unfortunately, although a few initial trials in MCI and AD patients suggested slowing of disease progression, such results have generally not been confirmed. Therefore, there is a crucial need for improved compounds with increased absorption and solubility, and ability to cross the blood brain barrier and reach mitochondria.

Neuroprotective strategies targeting other ROS-mediated pathways

As mentioned in the introduction, other sources of oxidative stress and free radicals have been identified and have served to elaborate new therapeutic strategies against ROS-mediated damage in AD. In this section, we will review some of the promising antioxidant agents and pathways implicated in ROS production.

In addition to mitochondria, Aβ itself can generate free radicals in the presence of metal ions [99], and methionine 35 is critical for these reactions [100, 101]. Free radicals are generated early in the Aβ aggregation process, when oligomers and protofibrils are formed [102]. One strategy to reduce Aβ-induced free radical generation and metal catalyzed Aβ aggregation would be to chelate the copper and zinc which bind to Aβ, increasing its aggregation. Clioquinol (figure 3I; table 1), a member of the hydroxyquinoline family used as antifungal and antiprotozoal drug, has been considered a potent chelator of copper, zinc and iron. Trials of clioquinol and second generation metal binding compound PBT2 showed improvement in both transgenic AD mice [103], as well as AD patients [104, 105], possibly by inhibiting metal-induced free radical production and by disaggregating metal-induced Aβ assemblies. Interestingly, PBT2 also increased activities of the matrix metalloproteases such as neprilysin, insulin degrading enzyme and tissue plasminogen activator, which lead to increased Aβ clearance [106].

Several natural compounds with potent antioxidant properties, such as spices, green tea, resveratrol, and vitamins, have been evaluated as therapeutic agents for AD [107]. Curcumin, a polyphenol (figure 3J; table 1), can acts as a free radical scavenger and antioxidant which inhibits lipid peroxidation and oxidative DNA damage [108]. It increases the expression of glutathione S-transferase and inhibits cytochrome P450. Curcumin has been used extensively over the years both in vitro and in vivo in transgenic mouse models of AD [109]. In the triple transgenic AD mouse model overexpressing mutant P301L tau, APP and presenilin 1 mutations, curcumin treated mice fed a high fat diet showed improved behavior and reduced tau phosphorylation [109]. In other transgenic mouse models of AD, low and high doses of curcumin reduced levels of oxidized proteins, insoluble and soluble Aβ, amyloid plaques and astrocytosis [110], and restored dystrophic neurites [111]. Curcumin was also able to reduce Aβ aggregation in vitro and in vivo [112]. Interestingly, its effects on amyloid clearance may be due to its ability to bind Aβ and increase Aβ uptake from macrophages [113]. More recently, it was reported that curcumin administration decreased motor dysfunction, neuronal loss and lipid peroxidation present in the spinal cord of old transgenic AD mice [114]. Curcumin was also tested clinically in AD patients during a pilot trial of 6 months comparing 2 formulations, powder and capsule. No differences were found in the curcumin treated group while assessing cognition, levels of isoprostanes and Aβ. However, it should be noted that no cognitive decline was observed in the patients receiving the placebo, which may have biased the results [115]. Furthemore, absorption of curcumin is very poor and better formulations are being developed.

Inflammation plays a key role in AD [116, 117] and in MCI [118]. First, inflammation is involved in Aβ clearance in the brain, in which microglia participate actively by internalizing and degrading soluble [119] and aggregated forms of Aβ [120]. In the triple transgenic AD mice, deficiency in the microglial chemokine receptor Cx3cr1 prevented neuronal loss [121]. In old transgenic AD mice, microglial function is impaired, as shown by the decrease of Aβ-binding scavenger receptors (scavenger receptor A, CD36, and RAGE), and Aβ-degrading enzymes (insulysin, neprilysin, and MMP9) [122]. In the same animals, microglial levels of proinflammatory cytokines interleukin-1β and tumor necrosis factor alpha were markedly increased [122]. Inflammation is also responsible for the expression of cytokines, increasing cellular toxicity and exacerbating AD progression [123]. Therefore, several groups have tested the effects of anti-inflammatory drugs [124, 125]. In relation to oxidative stress, microglia have been identified as an important source of ROS. Indeed, activated microglial cells are able to generate free radicals, specifically superoxide via the NADPH oxidase (NOX) enzyme [126], including in AD [127]. NOX is a transmembrane protein that is activated by the presence of cytosolic elements at the plasma membrane, such as rac, p67^phox, or p47^phox proteins (figure 4). The NOX assembly can then generate superoxide by reducing O₂ via electron transfer. Previous reports showed that the NOX system may be altered in AD, as shown by increased levels of p47^phox and p67^phox in the membrane fraction of AD brains, suggesting activation of NOX. In MCI brains, NOX activity was markedly increased in comparison with control patients [128]. Consistent with these data, Park and colleagues found that deficiency of NOX2 in transgenic AD mice reduced oxidative stress and improved cerebrovascular function and memory deficits without affecting Aβ levels or amyloid plaques [129]. In addition, in transgenic AD mice treated with ibuprofen, there was a reduction of amyloid plaque burden, microglial activation, and markers of oxidative stress [130]. Importantly, fibrillar Aβ increased ROS generation in microglial cells and stimulated the translocation of Rac (another cytosolic element of the NOX assembly) from the cytosol to the membrane, suggesting that Aβ can affect NOX2-mediated pathways [131]. Conversely, in microglial cells, ibuprofen pretreatment reduced ROS production induced by fibrillar Aβ administration. Indeed, non-steroidal anti-inflammatory drugs such as Ibuprofen (figure 3K; table 1) have been used in the treatment of AD for their ability to inhibit cyclooxygenase 2 (COX2). COX2 converts arachidonic acid to prostaglandin H₂, which in turn is converted to prostaglandins and then to thromboxane A₂. Ibuprofen was also able to disrupt NOX2-mediated signaling, as evidenced by the inhibition of Rac translocation to the membrane [130]. During clinical trials in AD patients, neither ibuprofen [132], tarenflurbil (figure 3L; table 1) [133], naproxen (figure 3M; table 1) nor celecoxib (figure 3N; table 1) [134] slowed disease progression or cognitive decline.

Figure 4. NADPH oxidase and production of ROS.

The assembly of NADPH oxidase subunits (gp91^phox/p22^phox) with cytolosic subunits p47^phox, p40^phox, p67^phox and rac results in the active enzyme responsible for the generation of superoxide.

Several pharmacological NOX inhibitors are currently available [135]; however, none of them have been tested either in vitro, in vivo or in clinical trials in AD research. In a mouse model of amyotrophic lateral sclerosis, apocynin (figure 3O; table 1), a NOX inhibitor [136], markedly increased survival, reduced ROS production and delayed symptoms induced by the superoxide dismutase 1 mutation [137]. Apocynin, or acetovanillone, blocks NOX assembly and therefore inhibits NOX-mediated superoxide formation. These findings suggest that NOX inhibitors may have potent therapeutic effects in neurodegenerative diseases.

Many groups have studied the implication of COX2 [124] and inducible nitric oxide synthase (iNOS), both involved in inflammation, in the treatment of AD. First, COX2 expression is increased in the frontal cortex of AD patients compared to control patients [138]. The same group also reported that overexpression of COX2 enhanced Aβ pathology in transgenic AD mice [139] and Aβ generation in cells, possibly through activation of the gamma secretase activity [140]. Therefore, inhibitors of COX2 have been considered, such as non-steroidal anti-inflammatory drugs (see paragraph above). In vitro, presenilin 2 mutations induced cell death was reduced by COX2 inhibition [141]. iNOS is another target used in the treatment of AD. Its expression was also increased in neuronal and glial cells of human AD brains, especially in the cortex [142, 143] and the hippocampus [144]. In neuronal and glial cells, iNOS produces nitric oxide (NO) by catalysing a five-electron oxidation of the guanidino nitrogen of L-arginine. In turn, nitric oxide can react with superoxide to generate peroxynitrite. In transgenic AD mice, elevation of iNOS and nitric oxide (NO) expression [145] was also associated with an increase of nitrosative stress at the vicinity of amyloid deposits [146, 147]. To prevent nitrosative damage, several groups have been testing the effect of iNOS inhibition in vitro and in vivo [148]. Deficiency of iNOS by genetic deletion reduced mortality and, at late stage, reduced amyloid plaque burden, microgliosis, astrocytosis, nitrotyrosine levels, and peri-plaque tau phosphorylation in APP/PS1 transgenic mice [149]. Other groups however have found that iNOS inhibition may enhance AD pathology [150, 151]. Nevertheless, it is clear that it plays an important role in AD. The pharmacological inhibitor of iNOS, N6-(1-iminoethyl)-L-lysine (L-NIL) (figure 3P; table 1), acts as an analog of L-arginine. It produces a time-dependent inactivation of citrulline- and NO-forming activity of iNOS in the presence of NADPH and oxygen [152, 153]. Moreover, the inactivation of iNOS is irreversible by displacement of the heme prosthetic group [154, 155]. Interestingly, L-NIL improved behavior and decreased cortical amyloid deposition, as well as microglial activation in transgenic AD mice [156].

Another strategy used to prevent ROS-mediated damage is the upregulation of transcriptional factors involved in the antioxidant response. The peroxisome proliferator-activated receptor-γ coactivator 1 alpha (PGC-1α) is an important transcription cofactor involved in energy metabolism [157, 158]. Interestingly, PGC-1α activation is dependent on various insults including the generation of reactive oxygen species. Its expression is reduced in human post-mortem AD brain, and this correlates with the pre-mortem dementia scales (CDR) and numbers of neurofibrillary tangles [159]. The nuclear receptor peroxisome proliferator-activated receptor-γ (PPAR-γ) activates PGC-1α which then interacts with multiple other transcription factors to modulate mitochondrial biogenesis. It has been shown that PPAR-γ may influence gene transcription of BACE1 [160]. Administration of bezafibrate (figure 3Q; table 1), a PPAR pan agonist, reduced behavioral deficit and inflammation in two mouse models of PD, using MPTP and 6-hydroxydopamine [161]. It also prevented the bioenergetic deficit and improved mitochondrial myopathy in mice produced by a deficiency of the nuclear encoded COX10 subunit of cytochrome c oxidase [162]. Pioglitazone (figure 3R; table 1) and rosiglitazone (figure 3S; table 1), two thiazolidinediones (TZDs) that selectively activate PPAR-γ, have been used for the treatment of AD. TZDs modulate the transcription of insulin-sensitive genes involved in glucose and lipid metabolism, especially in muscle, adipose tissue, and liver. These compounds can also bind to the outer mitochondrial membrane protein mitoNEET [163, 164]. In transgenic AD mice, administration of pioglitazone improved cerebrovascular functions and decreased oxidative stress [165], whereas administration of rosiglitazone reduced memory deficit [166]. The latter was also tested in a pilot study on AD patients and its administration had beneficial effects on tests of delayed recall [167]. This finding was not confirmed in a larger trial using 3 different doses of rosiglitazone. It should be noted that apolipoprotein E ε4 noncarrier patients did improve on the highest dose of rosiglitazone [168]. Polyphenols have also been used for their antioxidant properties and their ability to modulate Aβ and tau pathology in transgenic AD mice [169, 170]. Adminitration of polyphenol (grape seed extract) (figure 3T; table 1) also improved cognitive behavior and reduced Aβ oligomerization [171, 172].

The nuclear factor erythroid-related factor 2/antioxidant response element (Nrf2/ARE) pathway has become another promising target in the field of neurodegenerative diseases, including AD therapeutics [173]. Nrf2 is a transcription factor encoded by the NFE2L2 gene in humans [174] and a regulator of the antioxidant response [175, 176]. Its activity is regulated in part by the actin-associated protein Keap1, which binds to Nrf2 and sequesters it in the cytoplasm. Under oxidative stress conditions, the binding of Nrf2 with Keap1 is disrupted and Nrf2 is released. This release then allows the translocation of Nrf2 from the cytoplasm into the nucleus. While in the nucleus, Nrf2 can bind to promoters with AREs, stimulating the expression of genes that coordinate a cytoprotective response, known as a phase 2 response, such as genes encoding for mitochondrial antioxidant enzymes, and heat shock proteins [177]. It also down-regulates inflammatory genes, such as iNOS and COX2. In transgenic AD mice, decreased expression of Nrf2 and Nrf2/ARE regulated genes correlated with increased amyloid deposition in the brain [178]. In this context, the use of Nrf2/ARE activators may represent a promising avenue in the treatment of AD.

In vitro, in hippocampal cells, activation of the Nrf2/ARE pathway via both tert-butylhydroquinone (tBHQ) and overexpression of Nrf2, through adenovirus-mediated gene delivery, was protective by reducing Aβ₄₂ mediated cell death. These beneficial effects were also associated with increased expression of Nr2/ARE related genes [178]. In vivo, overexpression of Nrf2 through adenovirus-mediated gene delivery (injections performed in the hippocampus) improved memory deficits and decreased soluble Aβ levels as well as astrogliosis. Overexpression of Nrf2 also elevated the expression of Nrf2/ARE genes, such as heme oxygenase 1 [179]. Another class of Nrf2/ARE activators, the synthetic triterpenoid derivatives of 2-cyano-3,12-dioxooleana-1,9-dien-28-oic acid (CDDO) (figure 3U; table 1), has shown benefits in mouse models of neurodegeneration, including AD. These compounds can enhance Nrf2/ARE activation, particularly by disrupting Nrf2-Keap1 binding [180, 181]. Administration of CDDO-methylamide to transgenic AD mice improved memory retention, and reduced protein oxidation, microgliosis, amyloid burden and Aβ₄₂ levels [182]. Similar improvements in behavior and reduced oxidative damage were also found in mouse models of Huntington’s disease [183] and of Parkinson’s disease [184], providing hope for future clinical trials. There is extensive preclinical data showing efficacy of antioxidants in both in vitro and in vivo models of AD. Agents stimulating the Nrf2/ARE pathway in human patients have only recently been tested. However, dimethylfumarate (figure 3V; table 1) has shown efficacy in a phase 2 clinical trial in multiple sclerosis [185].

Again, the use of neuroprotective strategies targeting antioxidant-related pathways has brought positive outcomes in vitro and in vivo. However, their effects in human disease have not been extensively studied. Clinical trials using AD and MCI patients may be of great promise.

Concluding remarks

Oxidative stress plays an important role in AD pathogenesis. Generation and accumulation of ROS within cells are detrimental and can exacerbate the disease progression. Therefore, several strategies have been studied to prevent and/or slow down ROS-mediated damages (figure 5). It should be noted that, independently of the strategy, important factors must be considered in the use of antioxidant drugs, such as their bioavailability (absorption, transport, distribution and retention in the targeted area) and reaction kinetics. They must neutralize free radical faster than the radicals can damage their targets. Timing of the treatment is also extremely critical. For most of the drugs discussed above, beneficial effects have been reported in cell culture and partially in animal models. However, success in human clinical trials is much less frequent. One can argue that treatments are started very early in pathogenesis in animals, whereas in humans pathogenesis may already be well advanced by the time diagnosis is made.

Figure 5. Antioxidant strategies in AD.

In human clinical trials, some studies found slowing of disease progression, whereas others did not find any differences between the same drug and placebo. In fact, understanding and assessing antioxidant capacity in vivo is a challenging task and requires further investigations [186]. The antioxidant system forms a complex network, and treating with only a single or even a few may not be sufficient, or may even imbalance the network in a deleterious way. For this reason, upregulation of a coordinated endogenous network may be more successful. Even though discrepancies exist in data from clinical trials, several ROS-mediated neuroprotective strategies continue to provide hope for neuroprotective treatment of AD. For example, the transcriptional facilitation of the endogenous antioxidant system holds great promise.