Alzheimer's Disease: Redox Dysregulation As a Common Denominator for Diverse Pathogenic Mechanisms

A. Overview of Alzheimer's disease and scope of the review

O riginally described by Alois Alzheimer in 1906, Alzheimer's disease (AD) is the most common cause of dementia, being clinically characterized by cognitive impairment and memory loss. According to the World Health Organization, an estimated 37 million people currently have dementia worldwide, and about 50% of them are being affected by AD (29). AD has a very high personal impact, and a high social cost. In the United States, the total cost of illness, including informal care and caregivers' reimbursement, is estimated at 100 billion U.S. dollars per year (437).

Population demographics are rapidly changing as a result of the increasing life expectancy with a fast increase of aged population, favoring prevalence of dementing disorders. As a group, the main risk factor for AD is age (14, 196, 256, 412, 617). In fact, the incidence of sporadic AD increases exponentially as function of age after the sixth decade of life (256). Epidemiologic data identify two different AD populations depending on age of appearance. The named late onset or sporadic AD is the most prevalent group accounting for over 95% of cases (256). According to the Alzheimer's Association, 11–16 million cases of AD are projected just in the United States by 2050, and 95%–99% of AD cases, besides the rare familial AD, will be observed in the older population, with prevalence above 30% for individuals older than 85 years.

There is near consensus that AD is not a unique nosological entity, as considerable heterogeneity exists in its risk factors, pathogenesis, and neuropathological findings (294). Even more, a mixture of AD and a second pathological identity is present in most dementia patients; being the most frequent, the coexistence of AD and cerebrovascular disease. For that reason, at present days, the recommended preventive measures for AD are similar to those for cerebrovascular diseases (165, 254, 421).

Despite intensive research in neuropsychological evaluation, biomarkers, and imaging techniques, AD still remains as an exclusion diagnosis. Currently, clinical evaluation allows for the diagnosis of probable AD. However, the unequivocal AD diagnosis requires confirmation by anatomopathological analysis, in which the observation of amyloid β (Aβ) plaques and fibrillary tangles in defined regions of the brain, in conjunction with the clinical evolution of the patient, constitute the gold standard (294, 569). AD patients are identified by neuropsychological testing to evaluate neurocognitive performance in various areas and by assessing their cognitive decline and impairment on everyday life activities. During the last decade, an early stage of cognitive impairment without dementia has been defined. On the basis of histopathology, imaging, and cognitive evaluation scores, mild cognitive impairment (MCI), an intermediary stage between a cognitively intact person and AD, is considered an early stage of AD progression (501). MCI is further divided into two broad subtypes: amnesic MCI (affects memory) or nonamnesic MCI (602). The rate of amnesic MCI conversion to AD is roughly 10%–15% per year.

As a neurodegenerative disease, the pathology of AD is characterized by the presence of senile plaques, a heterogeneous set of aggregated Aβ, and by the presence of neurofibrillary tangles (NFTs), constituted by highly phosphorylated tau proteins. In addition, the histopathology involves synapse loss, neuronal loss (hippocampus, entorhinal, and temporoparietal cortex being the most affected brain regions) (656), and glial cell activation (257, 294, 319, 669, 674). Already in his original proposal, Alzheimer stated that these histopathological lesions were a marker of an upstream process instead of the disease cause (163). However, the most popular hypothesis for AD, the amyloid cascade hypothesis, still views Aβ as the cause of the neurodegenerative changes, stating that the first event of the pathogenic cascade of AD depends on Aβ accumulation.

On the 1980s, Aβ and the whole gene of amyloid precursor protein (APP) were characterized on the chromosome 21 (628). Once defined as one of the main components of plaques, Aβ was quickly associated to the early-onset, inherited autosomal dominant dementia (308, 546) familial AD. Later, mutation on two other genes encoding for enzymes processing APP, presenilin 1 (PS1) (441) and presenilin 2 (PS2) (714), were also shown to cause familial AD. PS mutations affect APP processing, resulting in an increasing amount of Aβ available to aggregate (214, 560). The most prominent feature of familial AD is its young age of appearance, as early as the fourth decade of life. Finally, the presence of apolipoprotein E4 (ApoEɛ4) was associated to sporadic AD, which could result in an increase of production or reduction of removal of Aβ (Fig. 1). In addition to these four original genes involved in AD, recent studies show that other genes could be also involved, including ADAM10 (316), ATXN1 (718), CD33 (447), as well as polymorphisms of various receptors, like alpha(2)-macroglobulin (270) and Vitamin D receptors APA1 and TAQ1 (349) and cytokine polymorphisms, such as interleukins (ILs), IL1β, IL1α, IL-6, and tumor necrosis factor α (TNFα) (219, 229, 518, 715).

OPEN IN VIEWER

FIG. 1.

Neurobiological mechanism of Alzheimer's disease (AD). Hypothetical sequence of events in the pathogenic cascade of AD. Aβ aggregates are present in familial and sporadic AD. However, whereas increased production of Aβ is consistently observed in the familial AD, the sporadic form could depend either on increased production or on reduced removal of Aβ—both mechanisms potentially being associated to aging, oxidative stress, and the inflammatory status of the nervous tissue, which can also influence synaptic function and eventually neuronal viability. In addition to the sequential events proposed by the amyloid cascade hypothesis, there is evidence that Aβ and inflammation can directly affect neuronal function, as well as the existence of feedback of neuronal damage and inflammation that favors accumulation of Aβ. AD, Alzheimer's disease; Aβ, amyloid β.

Similarities between familial and sporadic AD prompted the notion that both diseases share a common pathological mechanism dependent on Aβ. However, familial AD comprise around 2% of patients, and appears unreasonable to extrapolate this etiological mechanism to the sporadic form dismissing the possibility that such amyloidosis could be a secondary event to a totally different etiology (617). Indeed, familial AD and sporadic AD could represent the final clinical stages of different diseases (294, 421).

The precise mechanism by which Aβ produces synaptic impairment and neurodegeneration is still under intense discussion. Previous studies led us to propose that glial cell dysfunction is involved in the pathogenics of AD, through the generation of a cytotoxic environment (669) that would favor both Aβ aggregation and neuronal dysfunction and neurodegeneration. Over the years, other hypotheses have been proposed to explain the pathogenesis of AD. In addition to glial cell dysfunction (section I.C), these theories include oxidative stress (section II.A); protein misfolding/aggregation (section II.F), which can result in both Aβ-dependent senile plaques and tau-dependent NFTs (section II.B); mitochondrial dysfunction and impairment of calcium metabolism (section II.C); and neuroinflammation (section II.D), among others. Most of these hypotheses are not mutually exclusive, but could co-participate. Moreover, instead of the highly sequential view regarding the progression of AD traditionally held, an emerging view is that several mechanisms co-exist simultaneously affecting each other at multiple levels, establishing triggering loops that feedback on the cascade of degenerative changes.

On the other hand, despite such mechanistic diversity, common factors can be articulated in a convergent way for generating the disease, one of the axes being glial cell dysfunction (Fig. 2) (669) and another, oxidative stress. In this review, we discuss evidence supporting the role of oxidative stress in the pathogenesis and progression of AD, focusing on elements that allow recognition of oxidative conditions as a common factor for several biological mechanisms. Such recognition is especially relevant for therapeutic strategies capable of targeting effectively on common mechanisms responsible for disease evolution.

OPEN IN VIEWER

FIG. 2.

Disease mechanisms as viewed by the amyloid cascade hypothesis and the glial cell dysregulation hypothesis. Several pathogenic events contribute to neurodegeneration in AD. The amyloid cascade hypothesis states that the disease depends on the accumulation and aggregation of Aβ, and all the other biological mechanisms, including oxidative stress, inflammation, excitotoxicity, and neurodegenerative changes, depend directly or indirectly on Aβ accumulation. In contrast, glial cell dysregulation hypothesis states that the cause of AD depends on changes on microglial cell activation, which become increasingly cytotoxic loosing their defensive functions. Impaired activation results in oxidative stress, persistent and neuronal dysfunction, and in turn will induce production and aggregation of Aβ and additional neuronal dysfunction. Inflammatory activation, either secondary to aging or to various forms of stimuli or injury, can be responsible of glial cell dysregulation. APP, amyloid precursor protein; hyperphos tau, hyperphosphorylated tau.

B. Oxidative stress: General concepts

Oxidants, such as reactive oxygen species (ROS) and reactive nitrogen species (RNS), are normally produced, although at low concentrations, as part of aerobic metabolism. ROS include superoxide radical anion (O₂ ^·−), hydrogen peroxide (H₂O₂), and hydroxyl radical (^·OH), among others. O₂ ^·− can be dismutated to H₂O₂ and oxygen by superoxide dismutase (SOD), an antioxidant enzyme. These reactive species can accomplish functions as intracellular messengers. Nevertheless, reactive species, especially when produced at high concentration, can result in oxidative stress associated with cell damage (Fig. 3). O₂ ^·− can act as an intermediate in the generation of more reactive ROS, like hypochlorous acid and ^·OH; by reaction with nitric oxide (NO^·), it can form the highly oxidizing agent peroxynitrite (ONOO⁻). NO^· is a free radical gas that participates in the physiological regulation of cerebral blood flow and neurotransmission, but also exhibits pro-oxidative, cytotoxic effects by inducing oxidative and nitrosative stress (95, 157). Both NO^· and ONOO⁻ are relatively long-lived diffusible oxidants, thus RNS/ROS-mediated damage may spread over relative long distances. At physiological pH, ONOO⁻ can be protonated to form peroxynitrous acid (ONOOH), which then generates ^·OH and nitrate (NO₂ ^·), leading to lipid peroxidation and nitration of aromatic amino acids (157). In addition, in the presence of SOD, or upon interaction with carbon dioxide (CO₂), ONOO⁻ can also generate nitrosonium cations (NO⁺) that causes the nitration of tyrosine residues in proteins. Finally, since NO^· is capable to release iron from the Fe-binding protein ferritin, it may also facilitate the production of ^·OH via the Fenton reaction (690).

OPEN IN VIEWER

FIG. 3.

Generation and effect of radical species in the cell. Mitochondria are the main producer of ROS in the cell, with NADPH oxidase taking this place under conditions of inflammatory activation. Additional radicals are produced secondary to the activation of AA metabolism and activation on NOS; both pathways are activated by the increase of intracellular calcium. Presence of high amounts of O₂ ^·− and nitrites (NO ·) results in the production of ONOO⁻ that is especially reactive with proteins and lipids. Some transition metals (e.g., Fe²⁺, Cu⁺, and others) as well as Aβ can breakdown H₂O₂ to the reactive hydroxyl radical (Fenton reaction). The hydroxyl radical can subtract an electron from polyunsaturated fatty acid giving rise to a lipid radical cascade resulting in production of lipid peroxyl radicals (LP ·), which is not reduced by antioxidants, and the lipid peroxidation process occurs. Lipid peroxidation products such as HNE can covalently modify proteins and impair their function. LP is reduced within the membrane by the reduced form of vitamin E, resulting in the formation of a lipid hydroperoxide and a radical of vitamin E. Regeneration of vitamin E depends on vitamin C, and intermediate radicals enter the antioxidant GSH cycle. For the lipid peroxidation process, LP reacts with Fe²⁺ to form lipid alkoxyl radicals forming hydroperoxide. Superoxide radical is dismutated by the SOD to H₂O₂. H₂O₂ is most efficiently scavenged by the enzyme GPx, which requires GSH as the electron donor. The GSSG is reduced back to GSH by the enzyme GSHR. Radical species have multiple targets in the cell, including carbohydrates, proteins, lipids, and nucleic acid. Oxidation of these macromolecules is associated to their lost of function. Cells also have several protective mechanisms and pathways, including enzymes like SOD and catalase, GSH, and other antioxidants (vitamin E among others) for the management of oxidative stress. AA, arachidonic acid; COX, cyclooxygenase; GPx, glutathione peroxidase; GSH, glutathione; GSHR, glutathione reductase; GSSG, oxidized glutathione; HNE, 4-hydroxy-2-nonenal; H₂O₂, hydrogen peroxide; LOX, lipoxygenase; NADPH, nicotinamide adenine dinucleotide phosphate; NO ·, nitric oxide; NOS, NO·synthase; O₂ ^.−, superoxide; ONOO⁻, peroxynitrite; ROS, reactive oxygen species; SOD, superoxide dismutase.

Oxidative stress is caused by the imbalance between production of ROS and breakdown of the chemically reactive species by reducing agents and antioxidants enzymes. This imbalance may be due to environmental factors, stressors, or disease. Thus, in practical terms, oxidative stress is determined by excessive exposure to oxidant molecules when there is insufficient availability of antioxidant mechanisms (238).

1. Sources of oxidative stress

There are two major sources of oxidative stress, one associated with the chronic formation of ROS derived from the mitochondrial electron transport chain (34), and another related to the acute and high output formation of ROS derived from nicotinamide adenine dinucleotide phosphate (NADPH) oxidase, a mechanism especially relevant during the activation of the innate immune system, including microglial cells in the central nervous system (CNS) (285). NADPH oxidase is a transmembrane protein complex that transports electrons across biological membranes for accomplishing the reduction of oxygen to O₂ ^·−. NOX2, the prototype homolog for the six members NOX family of NADPH oxidases, also known as phagocytic oxidase gp91phox, is activated in macrophagic cells, such as microglia, to catalyze the production of O₂ ^·− during host defense (see section I.C). NOX2 is activated when the cytosolic subunits are phosphorylated and Rac is activated in the cytosol (Fig. 4a), resulting in their translocation to the membrane and formation of the active NADPH oxidase complex with cytochrome b558 (24, 25).

OPEN IN VIEWER

FIG. 4.

Scheme showing the role of NOX2 in ROS production and activation of inflammatory pathways. (A) NOX2 complex: membrane-bound enzyme complex found in the plasma membrane. It is made up of six subunits: a Rho GTPase, in this case Rac1, and five oxidase subunits (NOX2 or gp91-PHOX, p22phox, p40phox, p47phox, and p67phox). (B) NOX2 requires p22phox subunit to be stabilized at the membrane and for recruitment of cytoplasmic subunits for its activation: Phosphorylation of p47phox (organizer subunit) allows its interaction with p22phox to be localized itself at the membrane and recruits the activator subunit p67phox into the complex to make contact with NOX2 and the small subunit p40phox. The achievement of this chain of steps makes possible the interaction of GTPase Rac with NOX2. Once assembled, the complex generates superoxide by transferring an electron from NADPH in the cytosol to oxygen on the luminal or extracellular space. Aβ interacts with NMDA, as well as other putative receptors, resulting in the phosphorylation and activation of the subunits of NOX2 that assemble in the membrane as an active complex, producing ROS. The activation of NOX2 by Aβ appears to be mediated by an increase in intracellular calcium, which directly, or indirectly through ERK1/2 phosphorylation and PKC-dependent cPLA2a phosphorylation, phosphorylates NADPH oxidase subunits. In neurons there is also evidence of AA release in response to stimulation by Aβ and NMDA. ROS produced by NOX2 further activates ERK and PKC. Memantine acts as an antagonist blocking the effect of Aβ on NMDA receptor. Dotted arrows are putative pathways. cPLA2, cytosolic phospholipase A2; ERK, extracellular signal-regulated kinases; GTPase, guanosine triphosphatase; LP, lysophospholipids; NMDA, N-methyl-D-aspartic acid; NOX2, NADPH oxidase; PKC, protein kinase C.

Other enzymatic sources of ROS are xanthine oxidase and uncoupled endothelial NO^· synthase (eNOS). Uncoupling refering to the fact that, in the absence of critical cofactor for NO^· synthesis like tetrahydrobiopterin, NOSs (including eNOS) produce ROS rather than NO^·, with the resulting increased oxidative stress and reduced levels of NO^·. An usual way to measure oxidative stress in biological samples is to determine the level of oxidative stress markers, such as protein carbonyls, 3-nitrotyrosine (3-NT), thiobarbituric acid reactive substance (TBARS), free fatty acid release, iso- and neuroprostane formation, acrolein, 4-hydroxy-2-nonenal (HNE), carbohydrate-mediated advanced glycation end products (AGEs), and 8-OH-2′-deoxyguanosine (8-OHd6) and 8-oxo-7,8-dihydroguanosine (8-OHG) and other oxidized bases, and altered DNA repair mechanisms (79, 405, 529).

4. Redox state and redox buffering

The redox state depends upon the ratios of reduced and oxidized forms of redox couples, which include glutathione/oxidized glutathione (GSH/GSSG), NADH/NAD⁺, NADPH/NADP⁺, cysteine/cystine, thioredoxin_red/thioredoxin_oxid, and glutaredoxin_red/glutaredoxin_oxid. The GSH/GSSG ratio is the primary determinant of the cellular redox state because it is 1000 to 10,000-fold more abundant than other redox couples, and because of its low standard redox potential (Eo=−240 mV). GSH is the most abundant intracellular nonprotein thiol, reaching concentrations of up to 10 mM in mammalian cells, with astrocytes being the major source (266). Furthermore, GSH is also tightly linked to other redox couples, either metabolically or because of its ability to form thiol-GSH mixed disulfides (130, 525). GSH acts as an oxyradical scavenger by scavenging NO^· and other oxidants, thereby protecting cells against oxidative damage by reducing oxidized or nitrosylated protein thiols (241). In addition, GSH is involved in the elimination of toxic oxidation products such as HNE, a product of lipid peroxidation (324, 689), and is able to detoxify various oxidants by directly scavenging free radical or acting as coenzyme in GSH-peroxidase-catalyzed reactions (375). Within the cell, GSH is kept in its thiol-reduced form (>98%) by GSSG reductase, an NADPH-dependent enzyme; additional amounts of GSH are present as GSSG and as GSH conjugates. Maintaining optimal GSH/GSSG ratios in the cell is critical because GSH is a major endogenous antioxidant defense system of the brain, removing hydrogen- and lipid-peroxides (418). Activation of several signaling pathways, including PKB, calcineurin, nuclear factor jB, and MAPK, is associated with changes in redox status of neurons after oxidation of GSH (698).

Peroxiredoxins (Prxs), also called thioredoxin (Trx) peroxidases, is a family of antioxidant enzymes that reduce H₂O₂, ONOO⁻, and a range of organic hydroperoxides using reducing equivalents provided by Trx/thioredoxin reductase (TrxR) system (697), which have been studied as a stress responsive system (21). Several Prx isoforms are induced in the brain in response to insults (330), as well as the presence of aberrant patterns of Prx expression in the CNS of patients affected by neurodegenerative disorders (463), being suggested that they can serve neuroprotective functions.

Trx is an ubiquitous thiol oxido-reductase system that regulates cellular redox balance, undergoing reversible oxidation of the cysteine pair while reducing disulfide bridges of proteins (21), and appears to mediate signal transduction elicited by several growth factors and cytokines (536, 537). Trx plays an essential role in cell function by limiting oxidative stress directly via antioxidant effects and indirectly by protein–protein interactions (158). In fact, cellular redox regulation for many processes is provided in mammalians by interaction between the Trx and GSH systems (147). Together, they form a powerful system controlling redox regulation of gene expression, signal transduction, cell proliferation, protection against oxidative stress, anti-apoptotic functions, growth factor and co-cytokine effects, as well as regulation of the redox state of the extracellular environment (614). Besides the role as a source of reducing equivalents, Trx by itself acts as antioxidant or ROS scavenger (159).

In mitochondria, the main redox buffering systems are GSH, glutaredoxin, and Trx systems. GSH, besides acting as reducing agent and antioxidant, also acts as mediator of several physiologic reactions, including metabolism of xenobiotics, thiol disulfide exchange reactions, and cell signaling (cell-cycle regulation, proliferation, and apoptosis). In the nucleus, GSH maintains critical protein sulfhydryls (PSHs) that are necessary for DNA repair and expression (658). Cerebellar granule neurons in culture exhibit marked functional deterioration and die in response to loss of both mitochondrial and cytoplasmic GSH, but not after cytoplasmic GSH loss alone (585, 700). Additionally, although mtGSH depletion does not affect astrocytes' viability, they become more susceptible to NO^· exposure (444).

The brain is particularly vulnerable to oxidative stress due to (i) its high demand of oxygen, showing the highest oxygen metabolic rate of the economy, consuming around 20% of the total amount of oxygen in the body, (ii) its dependence on oxidative metabolism for obtaining metabolic energy, (iii) its high content of iron, which can catalyze the generation of ROS and RNS, (iv) its content of antioxidant enzymes, which is lower than in other organs (198, 412), and (v) the postmitotic nature of neurons, which make them especially vulnerable by accumulating mutations and impairments that are not removed by cell replacement. In particular, aged, AD or injured brains of any sort, show oxidative modifications in nucleic acids, proteins, lipids, and sugars (Fig. 5); oxidative damage; and changes deriving in loss of function (239, 381, 405). Most of cells have protective mechanisms, responsible for enzymatic breakdown or scavenging of ROS (Fig. 3). For example, H₂O₂ is converted to water and O₂ by CAT or glutathione peroxidases (GPxs). However, in the brain, antioxidant systems are reported to be less functional, which can lead to further increase of ROS and RNS reacting with the various target molecules (239).

OPEN IN VIEWER

FIG. 5.

Modification of proteins, nucleic acids, and lipids by glycation, oxidation, and peroxidation. The various ROS generated during oxidative stress have preferences on the type of macromolecule they oxidize, also generating complex metabolic interactions and newly generated reactive species. Protein modification and protein–protein cross linking by glycation, autoxidation, and the reactive aldehydes generated from lipid peroxidation are synergistic processes. Glycation and lipoperoxidative modification involve adduction of redox-active moieties that potentiate free radical autoxidation. AGEs, advanced glycation end products.

C. Microglia, astrocytes, and oxidative stress

Microglia are the brain-resident macrophages (255, 522, 539) representing its innate immune system, and providing the first line of defense for injury or disease. They colonize the brain at late stages of prenatal life, before the development of the blood–brain barrier (BBB) (522). In the healthy adult brain, microglia display a characteristic ramified morphology, and serve a surveillance role, dynamically monitoring brain microenvironment (161, 458). Microglia can sense a wide range of stimuli, including CNS trauma, ischemia, infection, toxic insult, and autoimmune injury (319, 332, 385, 566, 603, 674), recognizing a wide range of targets, such as peptides, lipoproteins, glycolipids, and nucleotides (450, 496, 661). In addition to abnormally processed, modified, or aggregated proteins (e.g., Aβ), stimuli include inflammatory cytokines, which can trigger microglia activation, damaged neurons, which provide the strongest signal for microglia proliferation, inducing reactive microgliosis, membrane breakdown products, altered molecules (e.g., active forms of matrix metalloproteinase), impaired neurotransmitter function (e.g., elevated glutamate concentration), and cytosolic compounds (240, 385, 450, 522, 563).

When stimulated, microglia become activated presenting structural and morphological changes, enlarging their size (200, 458) as well as modifying their functional properties (374, 385, 671). Microglial cell can undergo at least three activation pathways (and probably several intermediate ones) starting from the quiescent, surveillance state (222, 407, 434): (i) classical activation (M1 type activation), which under certain conditions will tend towards cytotoxicity, (ii) alternatively active/phagocytic/neuroprotective (M2) (222, 407), or (iii) regulatory (434). In fact, activation of microglial cells, far from a single phenotype, represents a continuum change from innate to adaptive activation with the expression of different cytokines and cytokine receptors (645).

Activation of interferon-regulatory factor 5 (IRF5), member of the IRF family, defines commitment to the M1 macrophage lineage (558). IRF5 activates genes encoding for type I interferon, inflammatory cytokines (including TNFs, IL6, IL12, and IL23) and tumor suppressors (331, 473). In mice, IRF4 controls M2 polarization by stimulating the expression of specific M2 macrophage markers (331, 558). For M2 macrophages, activation of nuclear factor-kappa B (NF-κB) subunit p50 has been associated with the inhibition of M1-polarizing genes (500), whereas CREB-mediated induction of transcription factor C/EBPβ has been shown to up-regulate M2-specific genes (548). M2-type induction, through secretion of IL4, IL10, and transforming growth factor β (TGFβ), secrete IL4, IL6, IL10, and IL13, cytokines that promote humoral immune responses and down-regulate M1-mediated responses, inhibiting numerous macrophage inflammatory functions (645). M2 polarization is further subdivided according to the production of IL10 or IL12, depending on specific associations, like wound-healing, tumor-associated, or heterogeneous macrophages (434). A priori, it may appear that M2 activation is the one associated with protective functions. However, there is strong evidence that chronic allergic inflammatory processes like asthma depend on expression of M2 cytokines IL4, IL5, IL9, and IL13 (230, 701). Finally, regulatory macrophages can arise during the later stages of adaptive immune responses, and their primary role appears to limit inflammation (433). There are many different ways to generate regulatory macrophages, although ERK/MAPK has emerged as a potential candidate to be the mediator (384, 434).

Microglia are activated in virtually all CNS diseases (240, 332, 453). Activated microglia produce and secrete a spectrum of inflammatory mediators, such as eicosanoids, cytokines (319, 450, 642), chemokines, ROS, NO^·, small metabolites acting as mediators, proteases like α-antichymotrypsin and α-antitrypsin, and inflammatory markers, such as serum amyloid P and C-reactive protein (45, 361, 385, 450, 453, 603, 642). These inflammatory mediators both regulate innate immune defense and act on neuronal properties, specially modifying synaptic function (10, 11, 173, 570). Furthermore, initially intended to destroy pathogens and injured brain cells as part of their protective response, they are also responsible for bystander damage of neurons at later stages, depending on the environmental context (361, 669). In fact, their cytotoxic activation is often associated to neuronal loss and decline of cognitive function in literature (61, 93, 319). Nevertheless, in addition to deleterious effects, microglia are also capable of secreting trophic factors and modulator cytokines, and in that sense, they have the repertoire to mediate neuroprotection.

As we discuss on section I.D, the inflammation hypothesis stresses the notion that hyperactive microglia are the primary cause of neurotoxicity in AD. In contrast, we propose that neurotoxicity is not a consequence of hyperactive but rather of mis-active, dysfunctional microglia (669). Thus, the innate immune response, initially protective, becomes a chronic mis-activation contributing to brain cytotoxicity (Fig. 6) (455, 559, 669, 703). The distinction is not trivial when interested in the development of therapeutical approaches. Adequately activated microglial cells are needed as scavenger cells in the CNS; however, by not responding to normal regulatory feedback and/or by having an impaired ability to clear Aβ (479, 669), glial cells could develop a predominantly cytotoxic feature. Thus, the aim of therapy should be oriented to potentiate a certain mode of microglial cell function rather than functionally inhibiting the whole microglial cell population, an approach that likely has a major bearing in the limitations of past and current thinking about how one might use immunoinhibitory drugs simply to turn microglia off for therapeutic benefit of neurodegenerative diseases.

OPEN IN VIEWER

FIG. 6.

Microglia drives both neuroprotection and progressive neurotoxicity depending on their activation state. Microglia serves multiple beneficial functions by promoting neurotrophic effects, innate immune responses, and modulation of neuronal function. Nevertheless, microglia can become mis-activated and cause neurotoxicity through mechanisms depending on their activation by inflammatory stimuli, such as oxidative stress, ischemia, or Aβ among others, producing neurotoxic inflammatory factors. In addition, microglia can become overactivated in response to neuronal damage (reactive microgliosis), becoming also toxic to neighboring neurons and perpetuating the cycle of neurotoxicity. Reactive microgliosis could be an underlying mechanism of progressive neuron damage across numerous neurodegenerative diseases, regardless of the instigating stimuli. IL1β, interleukin 1β; PGE₂, prostaglandin E₂; TNFα, tumor necrosis factor α.

Regarding AD, the activation of microglia induced by Aβ (249, 326, 584) results in several cell transformations (282) making glia highly competent for phagocytose Aβ plaques (31, 541, 584). Microglia are intimately associated with Aβ plaques in AD, but not with the diffuse Aβ plaques characteristic of aging (249, 262, 287, 669, 672). The trigger for activation in AD is unclear, but the invasion of plaques by morphologically active microglia is seen in AD transgenic mice models, as well as when Aβ is injected into the brain of otherwise healthy mice or in in vitro experiments (9, 462, 528, 635, 672). Multiple receptors have been reported to mediate this process: for instance, the receptor for advanced glycation end products (RAGE) (706), scavenger receptors (9, 123, 635), the G-protein-coupled chemo-attractant receptor (341), the formylpeptide receptor 2 (262), and toll like receptors (528). Cytotoxic activation of glia can be associated with the activation of reduced NADPH oxidase (665) or the induction of the novo expression of inducible NO^· synthase (iNOS); both critical enzymes responsible for the generation of oxidants. Importantly, Aβ is also clearly indicated as a source of oxidative stress (662), as Aβ activates microglia to produce extracellular O₂ ^·− (30, 510) as will be discussed in section II.B.

Astrocytes and neurons play a critical role in propagating and modulating the nature of neuroinflammatory activation, signaling reciprocally to microglia to either accelerate or slow microglial components of autocrine neuroinflammatory cycling. There is an intense cross-talk between neurons and glial cells manifested not only in the regulation of normal neuronal function but also in disease mechanisms (173, 559, 668). Microglia and astrocytes are strongly activated in AD, producing an array of inflammatory mediators and fulfilling phagocytic functions (9, 45, 453, 541, 669, 672, 674). Such robust activation in AD includes (i) the activation of complement pathway (3, 540), (ii) the up-regulation of cytokines, including IL1, IL6, IL8, TNFα, TGFβ, and chemokines like the macrophage inflammatory protein 1 (MIP-1) (6), and (iii) a strong innate immune cellular system activation (257, 385). Indeed, immunohistochemical studies reveal that activated microglia and astrocytes gather around senile plaques in AD (249, 287, 413) and appear to facilitate amyloid plaque clearance (7, 262).

At the cellular level, it is well established that astrocytes modulate production of cytokines, proteases, phagocytosis, and microglial cytotoxic activation (642, 671). The astrocytic modulation of microglial cell production of RNS and ROS can be lost under certain conditions of pathological astrocytic activation (562). Furthermore, whether microglial cell cytotoxic activation occurs or not will depend also on the interplay of cytokines that are released by microglia and astrocytes (Fig. 7). TNFα acts as a positive feedback signal, helping to potentiate glial cell activation, and as a consequence, enabling them to produce IL1β, and promote the feed-forward activation of astrocytes. This escalating cascade, however, is also under systemic inhibitory control, for example, by IL6, which is known to exert a negative feed-back on TNFα signaling. Inhibition of the stimulatory signals and maintenance or reinforcement of the inhibitory signal loops can be a strategy to prevent escalating microglial cytotoxicity. Another alternative would be the immunomodulation of reactive microglia that brings them back into a more protective activation state.

OPEN IN VIEWER

FIG. 7.

Endogenous control of the cascading activation of microglial cells and astrocytes. Astrocytes—via the cytokine TGFβ—exert a negative feedback regulation of cytotoxic microglia. This is an important endogenous mechanism by which ROS production (especially the highly reactive ONOO⁻) can be controlled and restricted at the microglial cell level. The inhibitory effect of astrocytes on microglial NO^· and ROS release is dependent on the astrocytic differentiation state. Whereas differentiated (regulatory activation) effectively inhibited microglial release of ROS and RNS, highly inflammatory activated astrocytes showed little if any inhibitory effect on the production of reactive species. Cascading glial activation also depends on the level of microglial activation, which is regulated by cytokines released from microglia and astrocytes. TNFα potentiates microglial cell cytotoxic activation. This, in turn, increases the microglial release of IL1β, which promotes the activation of astrocytes. In contrast, the microglial release of cytokine IL6 exerts a negative feedback on TNFα actions. RNS, reactive nitrogen species; TGFβ, transforming growth factor β.

Astrocytes interact with neurons by nourishing, protecting, and modulating growth and function, and providing a robust antioxidative potential (397), favoring survival of neurons in conditions of oxidative stress (187). Regardless the strong antioxidative potential of astrocytes, hippocampal neurons show a more robust effect than astrocytes on the regulation of cytotoxic activation of microglia (259) modulating the production of RNS and ROS (Fig. 8), an effect that was shown to be mediated by TGFβ (259, 655). However, under conditions of strong inflammatory activation, their capacity of regulating Aβ- and APP-induced microglia-mediated cytotoxicity and phagocytic activity of microglia are reduced. These results support the hypothesis that, in neuroinflammation, microglial cell scavenger function is impaired and reactivity is enhanced as an initial step for neurodegeneration (673). Furthermore, under inflammatory conditions, neurotoxicity induced by Aβ or APP is greatly enhanced in the presence of cytotoxic microglia. The observed cytotoxicity could be consequence of a persistent mis-activation, in which elevated concentrations of IL1β and radical species support chronic cytotoxic activation of glial cells and cell damage (519). In agreement with our hypothesis, hippocampal cells exposed to inflammatory stimuli do not retain their ability to modulate production of ROS by microglia (Fig. 8).

OPEN IN VIEWER

FIG. 8.

Inflammatory conditioning of hippocampal cultures abolishes their modulatory effect on the production of ROS by microglia. CM from hippocampal cultures under control condition (not exposed to inflammatory conditions) reduces O₂ ^·− production by microglial cells. Glial cultures were cultured under control conditions (control) or with 10 ng/ml IFNγ, or 1 μg/ml lipopolysaccharide+10 ng/ml IFNγ (LI) in NbB27 or hippocampal-CM obtained under control condition (CM control) or after stimulation with the same inflammatory mediators (CM Inflam) for 24 h. O₂ ^·− production was triggered with phorbol 12-myristate 13-acetate for the last 1.5 h of culture. (A) In mixed glial cultures, microglial cells produced superoxide radicals (dark cells) when exposed to inflammatory molecules. O₂ ^·− production was inhibited when glial cells were cultured with hippocampal CM. However, exposure to inflammatory conditions abolished the modulation of ROS production by the hippocampal cells. (B) Quantification of O₂ ^·− production by microglial cells. Results are expressed as fold-number increase of O₂ ^·− production compared with control cultures. Values correspond to the mean+standard error of the mean of three independent experiments performed in triplicate evaluated by the nitro blue tetrazolium chloride colorimetric assay. **p<0.001 compared with control conditions, ^# p<0.05 and ^## p<0.001 compared with stimulated cultures maintained in fresh Neurobasal/B27 media. CM, conditioned media; IFNγ, interferon γ.

D. Aging and oxidative stress

ROS are an unavoidable by-product of cell metabolism. A small percentage (∼1%–4%) of electron transfer normally goes toward basal production of O₂ ^·− (40). Oxidative stress is increased at basal conditions in aging (61, 350, 571, 581, 631). The increased production of radical species is proposed by the free radical theory of aging. Electron transport chain activity in mitochondria declines with age (470), whereas basal oxidative stress increases with age (350). This means that there is a decrease on electron transfer efficiency, resulting in an increase on the production of ROS, relevant for the accumulation of mitochondrial oxidative damage. In fact, accumulation of ROS and RNS appears to be the most important causal factor for aging (40).

Robust evidence indicates that the magnitude of the imbalance between antioxidant defenses and oxidants, or the level of oxidative stress, widens during the aging process. The age-associated increase in oxidative stress appears to depend on the fact that antioxidant defenses do not keep pace with the age-related increases in ROS production (525). Thus, the balance shifts progressively toward a more pro-oxidant state with aging.

In this context, an age-related loss of GSH can be predicted to have at least two potentially deleterious consequences: (i) increase of basal levels of H₂O₂, leading to a rise in the formation of ^·OH, and in consequence, in greater molecular structural damage. Enhanced lipid peroxidation would result in increased level of malondialdehyde and hydroxynonenals, which can form adducts with DNA or proteins, modifying their structure and function (652). (ii) Protein oxidation can interfere with the catalytic efficiency of enzymes, which could reduce their ability to mount adaptive responses under stress conditions (181). Both types of alterations indeed occur during the aging process (524). Thus, maintenance of an optimal GSH redox state is imperative for limiting the level of macromolecular oxidative damage.

Aging, especially of the nervous system, is also characterized, among other changes, by an inflammatory status that induces functional changes of the immune responses (345, 414, 669, 674, 702) and increased oxidative stress state (525). So, through oxidative stress, increased level of inflammatory cytokines and glial cell activation (674) set the stage for a neuroinflammatory environment that could facilitate progression of neurodegenerative changes. Microglia become activated in a more cytotoxic fashion as function of age. Aged microglia increase their production of inflammatory cytokines (see section II.D) and reduce some of their protective functions, like induced phagocytotic activity—condition that could lead to their failure to efficiently clear Aβ (669). These changes represent a shift in basal cell reactivity and defense activity that can result in a regulatory impairment that can favor the genesis and progression of neurodegenerative processes (563). The role of aged microglia is also relevant to understand therapy failures in AD. It is possible that anti-inflammatory drugs may not work in AD patients because, in the brain, neuroinflammation is mediated by microglia rather than by activation of an eicosanoid pathway. As we mentioned in section I.C, the inflammation hypothesis stresses the notion that hyperactive microglia are the cytoxicity entity, whereas we propose that AD is rather a consequence of dysfunctional microglia (669). Neuroprotective microglial cells are needed in the CNS. However, when glia are dysregulated, cytotoxic activation could predominate.

During aging there are numerous documented changes at cellular, tissular, and organ levels (Fig. 9). There are well-known biomarkers of aging, such as cross-linking of collagen, the accumulation of AGEs, lipofuscin, and accumulation of damage induced by ROS and RNS (40) at the DNA and protein level. Accumulation of AGEs is also observed in neurological disorders such as multiple sclerosis and AD. The receptor for AGEs, RAGE, is increased after oxidative stress, immune and/or inflammatory responses, and upon altered cell functions, suggesting that AGE accumulation is not a process privative of aging (598). Engagement of RAGE induces the release of inflammatory cytokines and free radicals, thus perpetuating a cycle of damage. RAGE is increased in AD, where it is found to be expressed on neurons and astrocytes. In astrocytes, AGE-conjugated proteins form granules, suggesting that astrocytes uptake and degrade glycated proteins (598).

OPEN IN VIEWER

FIG. 9.

Aging-dependent cell changes. Aging induces several functional and structural changes. Several of those changes can be linked to inflammation and oxidative stress. In general, organisms also present adaptive changes to aging for maintaining an adequate function. Genetic factors and environmental factors, depending on diet and other lifestyle factors, have also special relevance for the aging process. If genetic or environmental factors are deleterious, or if adaptive mechanisms fail or are insufficient to compensate aging changes, aging will result in impairment of glial (further promoting activation and cell damage) and neuronal function.

When assessing the effect of aging on CNS function, it is necessary to keep in mind that at every level of complexity, the nervous tissue depends on a finely tuned cross talk of glial cells and neurons, as well as on its capacity to respond to information being delivered from outside of the CNS, via endothelial cells and sensory neurons input. Glia secrete growth factors, such as nerve growth factor (NGF), basic fibroblast growth factor, glioma-derived growth factor, and insulin-like growth factor, as well as cytokines, including ILs, TNFα, and TGFβ. Glial cells and neurons maintain a reciprocal trophic and regulatory interaction, allowing for a regulated activation of the immune response, including limited production of radical species when needed (Fig. 10A). Under this view, cytotoxic activation could depend on the impairment of any of the interacting members of this regulatory cross-talk. In aging, there is a decrease on trophic support, as well as in the modulatory effect exerted by neurons and astrocytes (Fig. 10B). Microglial cell reactivity is greatly increased, which in turn increases the production of inflammatory cytokines, NO^·, and ROS both at basal conditions and in response to any activating stimulus.

OPEN IN VIEWER

FIG. 10.

Modulation of central nervous system inflammatory activation by neuron–glia interaction. The homeostasis of the nervous tissue is maintained by a finely tuned cross talk of glial cells and neurons. (A) Astrocytes (also microglia) secrete growth factors and cytokines, with reciprocal regulation of glial cells and neurons. Inflammatory cytokines are deemed as potentially responsible for cell damage in neurodegenerative diseases, inducing the production of NO · and ROS, which appears to be responsible of cell damage. Factors, such as TGFβ, induced by inflammatory stimulation, modulate the activation of glial cells toward a protective role. (B) In aging, there is reduction on the trophic support and on the modulator activity of neurons and astrocytes. Microglial cell cytotoxic activation is increased, with increased secretion of inflammatory cytokines, NO ·, and ROS. TGFβ level is also elevated, but there are reports indicating that its canonic transduction pathway—Smad 3—is reduced (at least in AD patients). In addition to this increased inflammatory reactivity, in aging there is an increased frequency of noxius stimuli depending on hypoxia, acidosis, and diverse forms of injury. Overall, in aged individuals, neurodegenerative processes could be facilitated by the basal inflammatory activation and for the relative difficulty of microglia to respond to the modulation by neurons and astrocytes, or, as an alternative, by the failure of the modulatory mechanism itself.

Age-dependent mitochondrial oxidative stress involves controversy and there are detractors questioning its significance (see section II.C). In any case, oxidative stress appears to influence life span extension. Life span extension by caloric restriction appears to depend on the reduction of superfluous oxidative metabolism. Furthermore, mitochondrial dysfunction is involved in apoptosis associated to mitochondrial membrane instability occurring when mitochondrial depolarization, oxidative stress, or energetic failure surpass a certain threshold level, with subsequent activation of cell death cascades (264, 371, 445).

A. Oxidative stress and progression of AD

Although the exact mechanism by which Aβ produces synaptic and neuronal loss is highly controversial, there are reports showing that antioxidant compounds block Aβ-induced neurotoxicity in culture and in in vivo and ex vivo AD models, consistent with a role of oxidative stress in AD pathogenesis and disease progression (611, 710).

The important number of proteins oxidatively modified in the MCI brain support the involvement of oxidative stress in dysfunction of glucose utilization and oxidative-dependent impairment leading to energy dysfunction, neuritic length, excitotoxicity, lipid peroxidation, cholinergic dysfunction, pH buffering and CO₂ transport, regulation of the cell cycle, APP processing, tau hyper phosphorylation, antioxidant defense, cell signaling, and protein synthesis (88, 529). The proteomics to identify oxidized brain proteins show that most changes are common for MCI, early and late AD, suggesting that certain key pathways are triggered early and are maintained during the progression of AD (609).

An example of a protein oxidatively modified in both MCI and AD is provided by peptidyl prolys cis/trans isomerase (Pin-1). In addition to its role for the cell cycle, Pin-1 also regulates other biological functions such as protein assembly and folding, intracellular transport, intracellular signaling, transcription, and apoptosis (80). An additional important function of Pin-1 in the specific case of AD is its ability to regulate APP function (484), by regulating the production of Aβ (80, 484). Proteomics have facilitated the identification of of other proteins oxidatively modified that are involved in energy and glucose metabolism. The content of oxidatively modified energy-related proteins in MCI and AD brains correlates well with the decrease of glucose utilization and altered activity of enzymes involved in glucose metabolism (88, 272), including oxidation of proteins involved with glycolysis and tricarboxylic acid (TCA) cycle, and significant decreases in TCA enzyme levels (91, 607). Positron emission tomography (PET) studies showed that glucose uptake is reduced in early AD brains (164, 596). Alpha-enolase, triosephosphate isomerase, and phosphoglycerate mutase 1 are enzymes contributing to the glycolytic pathway, and together with ATP synthase-alpha, are common targets of oxidation in AD brains (109, 489, 608, 610). Alteration of the function of the above-mentioned enzymes may lead to increased production of methyl glyoxal (MG), a small ketoaldehyde compound, derived from triosephosphate, a glycolytic intermediate. MG can react with lysine, arginine, histidine, and cysteine residues and glycate them to form AGEs products and can ultimately lead to altered structure and function of proteins (226, 707). The proteins enolase, neuropolypeptide h3, ATP synthase, and aldolase appear to be oxidized in brain as early as MCI and at every stage of AD (88, 109, 489, 608, 610).

In MCI brain, there are decreased activity and levels of enzymatic and nonenzymatic antioxidants (231), which can lead to increased production of free radicals during the progression from MCI to AD. Coherent with a decreased antioxidant activity, there are elevated protein carbonyls, protein-bound HNE, free HNE, TBARS, and MDA and 3-NT in the brain of MCI subjects compared to age-matched controls (88, 89, 313). MCI subjects also show high levels of isoprostanes in plasma, urine, and CSF compared with those of healthy subjects (406). Further, in MCI, increased oxidative damage in nDNA and mtDNA, with increased levels of 8-OHdG,2,6-diamino-4-hydroxy-5-formamidopyrimidine, 8-OHG, 4,6-diamino-5-formamidopyrimidine, and 5-hydroxycytosine is reported (679). Augmented 8-OHG, reported in cytosol of AD brains, decreases when Aβ and NFT burden increases, suggesting that the oxidative damage to RNA is an early event in the progression of AD (466). Brains from early AD patients also show increased levels of protein nitration, indicative of increased levels of RNS, and elevated protein-bound HNE (530). Interestingly, it has been proposed that Aβ, at low concentrations, has protective effects (112, 342), on the basis that Aβ could be part of the antioxidant defense in conditions of aging and disease (467). This notion is especially relevant considering the therapeutical strategies based on decreasing Aβ levels, which could worsen the disease (492, 590).

Genes whose products participate in reducing oxidative stress, inflammation, and accumulation of toxic metabolites contain the antioxidant response element (ARE). ARE-containing gene promoters include GST, coenzyme Q10 (Q10), NAD(P)H:quinone oxidoreductase, SOD1, TrxR, and HO-1 [also referred to as heat-shock protein (Hsp)32], and GPx (292, 343). The ARE promoter element is bound by several transcription factors, being the nuclear factor E2-related factor 2 (Nrf2) responsible for activating transcription in response to oxidative stress. Nrf2 coordinates the up-regulation of cytoprotective genes via ARE.

Nrf2 is expressed in both the nucleus (predominant expression) and the cytoplasm of neurons in normal hippocampi. In AD hippocampus, Nrf2 levels are reduced (521) and it is not translocated to the nucleus, resulting in a significant reduction in Nrf2 nuclear content. The opposite occurs in PD, where Nrf2 label remains strongly nuclear in nigral neurons (521). These findings suggest that in AD neuronal Nrf2-mediated transcription is not induced despite the presence of oxidative stress.

The ability of Nrf2 to up-regulate the expression of antioxidant genes suggests that increasing Nrf2 activity may provide a useful system to combat oxidative insults. In fact, overexpression of Nrf2 in cell culture protects against oxidative damage elicited by H₂O₂, NO^·, and glutamate (170, 582), whereas neurons from Nrf2 KO mice show an increased vulnerability to neurotoxic insults. Boosting the activity of the Nrf2-ARE pathway by tert-butylhydroquinone treatment or adenoviral Nrf2 gene transfer protects against Aβ toxicity, a neuroprotection associated with increased expression of Nrf2 target genes and reduced phosphorylation of p66Shc, a marker of increased susceptibility for oxidative stress (307, 329). On the other hand, in transgenic AD mice model APP/PS1, the Nrf2-ARE pathway is attenuated at the time of Aβ deposition (307). Increasing levels of Nrf2 in the brain of aged APP/PS1 mice significantly reduce their spatial learning deficits (306).

Nrf2 gene transfer is associated with a robust reduction in astrocytic, but not microglial activation and induction of Nrf2 target gene HO-1. Similarly, whereas Nrf2-ARE regulated gene expression attenuates inflammatory responses, disruption of Nrf2 expression can result in autoimmune disease and enhanced oxidative stress-induced inflammatory response (360). The association between neuropathological abnormalities and changes in expression of ARE-driven genes suggests that Nrf2-ARE signaling pathway could represent an endogenous protective response aimed to limit oxidative neural injury and to restore the redox-balance. Interestingly, recent cDNA profiling studies performed in Nrf2 KO mice, and various in vitro models have led to the discovery of several Nrf2-dependent genes that could be implicated in neuronal protection and survival (327, 344, 358, 359, 582).

There are additional processes that are regulated by cellular stress responses. The brain detects and overcomes oxidative stress by a complex system of longevity assurance processes. Hsps are highly conserved and facilitate correct protein folding. HO-1/Hsp32, an inducible redox-regulated enzyme and its interplay with TrxR, appears to have an important role in cellular antioxidant defense (96). Hsp serve as chaperones that bind to other proteins and regulate their conformation, preventing protein misfolding and oligomerization, protein movement across membranes, availability of receptor, or enzyme activity (497). Hsps are induced under conditions of cellular stress (396), and have been reported in a variety of disorders and injuries, including stroke, epilepsy, cell trauma, aging, and neurodegenerative disease (97, 99). Molecules that activate this defense mechanism are therefore possible candidates for cytoprotection (498). Some of the Hsps include ubiquitin, Hsp10, Hsp27, Hsp32, Hsp47, Hsp60, Hsc70, Hsp70, Hsp90, and Hsp100/105 (100, 172, 306).

Hsp response is also involved in cellular homeostasis during brain development, brain differentiation, cell cycle, apoptosis, growth factors action, mRNA, and protein half-life. Hsp can also be activated by non-noxious stimuli, such as nutritional antioxidants or acetylcarnitine. Acetylcarnitine, through activation of the redox-sensitive transcription factor Nrf2, and its consequent binding to the ARE in the HO gene, up-regulates HO-1 and TrxR, thus counteracting nitrosative stress and NO^·-mediated neurotoxicity (96). Mitochondria are specific targets of the protective effect of Hsp against oxidative injury (394).

The antioxidant protein HO could sense NO^· and thus protect against ROS and RNS insults. This notion is supported by the fact that NO^· and NO^·-related species increase HO activity and induce HO-1 expression, and besides, that cells pretreated with NO^·-releasing molecules acquire increased resistance to H₂O₂-mediate cytotoxicity in association with HO activation (436). In fact, NO^· could act as a signaling molecule that, by triggering expression of cytoprotective genes such as HO-1 and Hsp70, may lead to adaptation and resistance to subsequent more severe nitrosative and oxidative stress (435). Acetylcarnitine treatment of astrocytes exposed to cytokine-induced nitrosative stress restores GSH/GSSG ratio and complex IV inhibition, an effect associated with up-regulation of HO-1 and nuclear translocation of Nrf-2. In in vivo experiments, GSH is a critical factor for induction of cytoprotective Hsps (98).

Hsp60, Hsp70, and Hsp90 alone or in combination protect against intracellular Aβ stress through the maintenance of mitochondrial oxidative phosphorylation and functionality of tricarboxylic acid cycle. Notably, Aβ inhibition of complex IV activity is selectively neutralized by Hsp60. The combined effect of HSPs reduces free radical burden, preserve ATP generation, decrease cytochrome c release, and prevent caspase-9 activation, all important mediators of Aβ-induced neuronal degeneration (663).

Oxidation of specific brain proteins in common for MCI and AD indicates that certain key pathways are affected early and become triggers for additional pathological changes, thus affecting the progression of AD. Similarly, there are also protective changes being induced by oxidative stress. Exploring these deleterious and protective pathways in detail will provide clues for better understanding the pathogenesis and progression of AD and also for the development of effective therapies to treat or delay this disorder.

B. Aβ, Tau, ApoE4, and oxidative stress

1. Amyloid β

As already mentioned, the amyloid hypothesis (244, 245) proposes that the deleterious effects of Aβ and amyloid plaques are the causative agent of neuronal dysfunction and death in AD (355). The main component of senile plaques are peptides of approximated 4 kDa, being Aβ_1–40 and Aβ_1–42, the most abundant components. In vitro and in vivo studies indicate that Aβ_1–42 is both more amyloidogenic and more toxic than Aβ_1–40 (66, 81, 426). The self-association of Aβ is dynamically structured from monomers passing through dimers and protofibrils to organized oligomeric and multimeric fibrils (167, 169, 262, 570, 574, 677). Almost every Aβ structural entity (169, 236, 244, 351, 574, 671) can lead to the impairment of various neuronal functions ranging from in vitro hippocampal neuron LTP (677, 681), or synaptic pruning (94, 574) to reduced learning ability in vivo (274, 351), and to excitotoxicity and progressive synaptic damage and neuronal death. The progressive accumulation of Aβ adds chronic inflammatory activation to the neuronal dysfunction (especially for some oligomeric forms of Aβ). In sporadic AD, both increased production and reduced removal of Aβ can be linked to aging, oxidative stress, and the inflammatory status of the nervous tissue, which can also influence neuronal function and eventually neuronal viability (206, 669). In fact, signal transduction cascades activated by Aβ can be linked with cell injury and neurodegenerative changes (364). However, aggregated Aβ fails to show this correlation with pathological lesions or clinical compromise (29, 294, 517, 602). In contrast, recent evidence shows that the level of nonfibrillar Aβ appears to be directly correlated with the severity of the pathology (386, 416), suggesting that small Aβ oligomers are the actual toxic species rather than fibrillar Aβ (667).

As many as 30% of cognitively normal individuals in their 70s have Aβ plaques in their brains as revealed by PET scans, without showing signs of cognitive impairment or dementia (193, 431). Similarly, autopsy studies also highlight that Aβ plaques are present in cognitively normal individuals (14, 294). In an autopsy-based cross-sectional study, it was found that Aβ plaques accumulated before the development of cognitive impairment; thereafter, the amount of Aβ in the brain remained nearly constant and was similar across a wide range of patients showing different clinical severity. In contrast, atrophy, neuronal and synaptic loss, and tangles increase with severity of illness (94, 570, 667).

Three broad types of plaques can be recognized. The first type is nonamyloid Aβ-deposits, which can be further distinguished as diffuse and focal. The other two have amyloid structure (i.e., Congo red or thioflavin S positive), a focal type of deposits and the neuritic plaque, which is surrounded by a neuritic corona of neurofibrillary degeneration. In addition, other categories can be recognized depending on structure, morphology, and staining [for a detailed description, see (184, 632)]. Diffuse Aβ-deposits are poorly immunoreactive and ill-limited. There is no association with glial cells and show no inflammatory activation. Thus, diffuse plaques appear to be associated to the aging process and have no correlation with AD. In contrast, in neuritic plaques, the focal deposit constitutes the core, which commonly shows microglia directly associated with it. Indeed, focal deposits with and without amyloid aggregation (234) are associated with morphologicaly activated microglia (19), which is a hallmark of AD pathology (217, 577), although this point is not devoid of controversy (555). The corona surrounding the Aβ deposit contains neuritic and astrocytic components. Neuritic plaques are always associated with congophilic deposits (420). This characteristic could depend either on the fact that microglia themselves are necessary for the transformation of the diffuse deposit into amyloid (691) or because the amyloid conformation is what activates microglia. Furthermore, it is unclear if amyloid plaques originate from diffuse plaques or if they arise by independent pathways (568, 599, 705).

There are also other special plaques, like the cotton wool plaque, formed by N-truncated Aβ (423); the inflammatory plaque, characterized by a strong microglial reaction, surrounding an amyloid core without Aβ reactivity, observed in PS1 mutations (580); and the vascular deposits of Aβ in the vessel walls, of the hereditary cerebral hemorrhages Dutch type, related to a mutation of the APP gene (354). It should be emphasized that plaques are not pathognomonic of AD. They have been described in chronic head trauma (297, 464) and prion disease (478), and they can be also observed in cognitively normal population, and MCI population (23, 68, 505), although the density of neuritic plaques and of NFTs is, on average, higher in early AD cases than in MCI patients (43, 221). The major difference of Aβ deposits found in AD is their distribution (22, 68, 633). In cognitively normal individuals, parenchymal and vascular Aβ deposits are usually restricted to the cerebral cortex, the basal ganglia, the thalamus, and the hypothalamus. In contrast, in AD patients, Aβ deposits are found in the midbrain, brain stem, and cerebellum, in addition to the brain regions potentially affected in cognitively normal individuals (633, 634).

Some propose nowadays that AD can be a two-stage process, in which Aβ initiates a cascade of early damage, which when progresses, becomes independent of Aβ. This independence of Aβ would explain the mild efficacy of therapies targeting Aβ after AD is well established, because in this later stage biological processes like calcium elevation, calcineurin and caspase activation, mitochondrial dysfunction, impairment of axonal transport, neuronal impairments, and inflammation predominate (333). An alternative view held by others is that once Aβ has aggregated in plaques, the damage to the brain is done, and cannot be reverted. Furthermore, plaques represent a source of persistent insult for the brain (either by its mere presence, by its effect on neuroinflammation and oxidative stress, or by being a source of soluble oligomers). Under this view, even if the downstream effects of plaques are targeted therapeutically, if the plaque is not neutralized, the insult persists and the therapy fails to be efficacious (669, 671, 672). In this way, depending upon the hypothesized pathogenic process of AD, the plaque has a different meaning for a therapy to be effective.

Although the mechanism of damage induced by Aβ remains elusive, there is evidence suggesting that oxidative stress is at least partially responsible for Aβ-induced damage (Fig. 12), as it appears to be a mechanism shared with several neurodegenerative diseases (215, 218, 456, 596). In particular, markers of oxidative stress in AD brains co-localize with Aβ plaques (477, 591), situation also observed in animal AD models (591).

OPEN IN VIEWER

FIG. 12.

Induction of oxidative stress by Aβ. Aβ accumulation may be secondary to specific mutations of APP or γ secretase (presenilin-1), one of the enzymes processing APP, in familial AD, or may be due to the presence of apolipoprotein E4 (not indicated in the figure) or another undefined cause in case of sporadic AD, which could result in an increase of production or a reduction of the removal of Aβ. Aβ can promote cellular alterations, including increased ROS production, perturbed energy and calcium homeostasis, intracellular (and extracellular) protein aggregates, and activation of apoptotic cascades. These effects are additives to the age-related increases in oxidative stress, metabolic compromise, DNA instability, and ion homeostasis dysregulation to disrupt neuronal integrity, thereby resulting in synaptic dysfunction and cell death. In addition, although not shown in this figure, changes in glial cell homeostasis occur and contribute to inflammatory processes, oxidative stress, and excitotoxicity. ETC, electronic transfer chain; I (1,4,5)P3, inositol 1,4,5-trisphosphate; JNK, c-Jun N-terminal kinases/stress-activated protein kinase; mtDNA, mitochondrial DNA; P38 MAPK, P38 mitogen-activated protein kinases.

It has been reported that Aβ exerts toxicity that correlates with the generation of cellular H₂O₂ (41), and becomes abolished by activation of SOD (639), O₂ ^·−/H₂O₂ scavengers (73), and vitamin E (247). Conversely, oxidative stress increases Aβ levels (707), implicating the perpetuation of the damage induced by oxidative stress. One current controversy is whether or not Aβ fibrils are required for toxicity. Interestingly, oxidative stress significantly precedes the appearance of Aβ plaques and neurofibrillar tangles in AD brains (427, 429). However, in vitro studies indicate that exposure to aggregate Aβ leads to cytotoxicity due to calcium influx and the induction of oxidative free radical damage (409). Several different hypotheses have been proposed to explain this toxic effect, including the formation of ion channels in cell membranes (516), the spontaneous fragmentation of Aβ to generate peptidyl radicals, and the direct formation of H₂O₂ by Aβ (623), by mechanisms at least partially dependent on a Fenton-type reaction (276).

Respiratory burst of microglia is triggered by several factors, and Aβ also appears to be a potent trigger, indirectly contributing to oxidative stress (119). Aβ is partly targeted to mitochondria and, under pathologic conditions, appears to induce mitochondrial dysfunction and oxidative stress. Oxidative stress, in turn, can induce proteins to adopt insoluble β-pleated sheet conformations. Some authors propose that Aβ is not toxic in absence of redox metal ions (277). Because Cu and Fe are required for electron transport, their chelation could inhibit indirectly oxidative phosphorylation. Accordingly, in vitro studies show that chelation of Cu and Fe inhibited the formation of ROS (77, 263, 589). Thus, the oxidative damage of Aβ would be linked with the presence of redox metals, Cu, Zn, and Fe abundant in mitochondria. Aβ has unusual high affinity for both transition metal ions Cu and Fe and has the capacity to reduce these metals and, subsequently, produce H₂O₂ (134) and OH^· (232). These readicals will contribute to the generation of oxidized amyloid, cross-linked forms of Aβ, which in turn, will promote the oligomerization of monomer Aβ and the formation of insoluble precipitates. Precipitates, in turn, will sequester the ions that enable their aggregation (76, 178, 252, 278). Subsequent extracellular secretion of metal-Aβ complexes would give rise to insoluble Aβ plaques, which presumably would promote microglial cell cytotoxic activation and invasion. In turn, aggregated Aβ triggers ROS production by microglia, thereby potentiating deposition of Aβ. Both Aβ plaques and NFTs are enriched in Zn, Cu, and Fe. Furthermore, Fe, by activating the Fe-response element (IRE) in the 5′ UTR of APP mRNA, or by modulating the activity of α-secretase, can promote the generation of Aβ (543). In concordance with those observations, concentration of metal ions, such as Fe, Zn, and Cu, are markedly elevated in AD patient brains, particularly in those areas most severely affected (76). Hence, Aβ aggregation and oxidative stress appear to be potent reinforcing phenomena (278).

Cu(II) is known to bind to Aβ with a high affinity via histidine and tyrosine residues (281). Cu(II) promotes Aβ-dependent neurotoxicity by promoting the formation of H₂O₂ (134) and influencing its aggregation. Moreover, Cu–Aβ produces·OH (232), at least partly by inducing conformational changes that enable SOD-like activity (111). Furthermore, Cu appears to play a role in generating ROS through its binding to APP, which reduces Cu(II) to Cu(I) through an electron-transfer reaction that could enhance the production of·OH (722).

The role of Zn in the etiology of AD is very intriguing. Several reports indicate that Zn in micromolar concentration inhibits Aβ-induced toxicity (134, 212). The exact mechanism of the protective effect of Zn against Aβ toxicity is unclear; however, one of the hypotheses is that Zn competes with Cu (or Fe) for Aβ binding. Binding of Zn changes Aβ conformation and Cu ions cannot reach its metal-binding sites.

Despite the overwhelming correlations between oxidative stress and Aβ, the fine tune mechanism is still a matter for active research. In addition, Aβ-induced oxidative stress and neurotoxicity can involve methionine-associated formation of ROS (82). Finally, there are also reports that Aβ can increase oxidative stress through its effects on mitochondria, which will be described in the next section.

C. Mitochondrial impairment, mitophagy, and oxidative stress

In neurons, synaptic transmission, cellular remodeling, and cell death are processes tightly associated with mitochondrial function (445, 565). On the other hand, mitochondrial dysfunction, oxidative stress, and cellular death are common denominators in many neurodegenerative diseases (371). The mitochondrial cascade hypothesis proposes that the life span of mitochondria is finite and, when it is overpassed, mitochondrial functions are impaired; this propasal offers interesting insight for understanding the biochemical, histological, and clinical features of AD (619). Evidence supporting this hypothesis is multiple and include altered morphology, depressed metabolic activities, and the release of mitochondrial pro-apoptotic protein in both animal models and cell culture assays of neurodegeneration (Fig. 13) (164, 428). Electron microscopy analysis of mitochondria in AD brains revealed significant morphological alterations (27). Mitochondria are vital organelles and the main energy source for cells; they are also critical regulators of intracellular Ca²⁺ level, and cell death (194, 713). Nevertheless, they are also one of the main intracellular sources of ROS (350). Overexpression of mitochondrial uncoupling proteins would reduce production of ROS and would provide neuroprotection (122, 315, 408, 612). However, a direct link between respiratory uncoupling and these two activities has not been conclusively provided (296).

OPEN IN VIEWER

FIG. 13.

Oxidative stress and mitochondrial failure. As part of an Aβ-centered scheme, Aβ either interacts with membrane receptors or crosses the cell membrane and enters the mitochondria, affecting its stability and inducing the production of ROS and RNS. The mitochondrial-binding target for Aβ has been identified to be alcohol dehydrogenase. Their peroxydative attack on cell and organelle membrane lipids yields the mitochondrial toxins HNE and malondialdehyde. Oxidative damage of membrane-bound, ion-specific ATPases and stimulation of calcium (Ca²⁺) entry mechanisms (e.g., NMDA glutamate receptors, MAC of complement, and ion-selective amyloid pore formation) cause cytosolic and mitochondrial Ca²⁺ overload. Aβ directly attacks electron transport complex IV (Cyt-c) and α-ketoglutarate and pyruvate dehydrogenase, which are key enzymes of the Krebs-cycle, and damages mtDNA. Lipid peroxidation products also promote tau phosphorylation and aggregation, which in turn inhibit complex I. Exaggerated amounts of ROS and RNS are generated at complexes I and III. As the mitochondrial membrane potential collapses and permeability-transition pores (ψm) open, caspases are activated. Aβ also induces the stress-activated protein kinases p38 and JNK, as well as p53, which are further linked with inflammatory activation and apoptosis. MAC, membrane-attack complex; NMDAr, glutamatergic N-methyl D-aspartate receptor.

For neurons to generate the large amounts of energy they require, mitochondria undergo fission and fusion (referred as mitochondrial dynamics) to generate smaller or elongated organelles (684). Normal mitochondrial fission and fusion facilitate biogenesis of mitochondria, repair of defective mtDNA, and redistribution of mitochondria to sites requiring high-energy production (141, 203, 325). The complex morphology of neurons adds additional complexity to mitochondria dynamics, which should be moved into different neuronal regions for satisfying local demands of energy and calcium buffering. Indeed, mitochondria are highly mobile in neurons and become abundant in neuronal segments with high metabolic demands such as active growth cones and synapses (305), where mitochondrial biogenesis is particularly required for synaptic transmission and maintenance of synaptic structure (141, 357, 365, 388). The dynamic balance of mitochondrial fission and fusion not only controls mitochondrial morphology, length, size, and number but also regulates mitochondrial function and distribution, having profound impact on neuronal function (139, 684). In AD, mitochondria accumulate in the somata and are reduced in neuronal processes of some neuron populations (684) likely contributing to the observed synapse dysfunctions. Impaired mitochondrial dynamics is manifested in altered mitochondrial morphology and reduced cell bioenergetics and thus contributes to neuronal injury in many neurodegenerative disorders, including AD, PD, and Huntington's disease (HD) (35, 65, 325). Mitochondrial abnormalities correlate with dystrophic neurites, the loss of dendritic branches, and the pathological alteration of the dendritic spines in brains of AD patients (28, 619, 620).

The mitochondrial cascade hypothesis establishes that in sporadic AD, Aβ deposition, NFT formation, and neurodegeneration may be consequences of mitochondria malfunctioning. Mitochondrial fragmentation is an early event during apoptosis that precedes cytochrome c release and caspase activation (201). Excessive mitochondrial fission is correlated with increased ROS production (684, 685, 716), but there is no total agreement on what causes mitochondrial malfunction in sporadic AD. It is possible that accumulation of oxidative stress-induced mtDNA damage during aging and subsequent mitochondrial dysfunction may serve as a trigger. There is increased oxidation in both mtDNA and nDNA in frontal, parietal, and, especially, in temporal lobes of AD cases compared with age-matched controls, with a 10-fold predominance of mtDNA oxidation over nDNA oxidation (680). There are also mtDNA mutations associated with increased incidence of AD (128). It has been found that AD brains show increased unique mtDNA mutations compared with control cases. These mtDNA mutations are enhanced in an age-dependent fashion and preferentially occur in mtDNA regulatory elements, such as the control region (128).

Therefore, in neurodegeneration, abnormal mitochondrial morphology is frequently observed (27, 265, 684), and patients in early stage of AD exhibit declining mitochondrial energy metabolism and ATP production (366, 526, 686). Posttranslational modification of mitochondrial fission and fusion proteins contributes to altered mitochondrial dynamics. Phosphorylation, ubiquitination, sumoylation, and proteolytic cleavage of the mitochondrial fission protein Drp1 (also referred as DLP1), which is found in AD brains (148, 684), regulate mitochondrial fission by affecting its activity in culture (140, 448, 688, 711). Excessive accumulation of nitrosative stress triggers abnormal mitochondrial morphology in brains of neurodegenerative patients via S-nitrosylation of Drp1 (148). Exposure of neurons to aggregated Aβ_25–35 induces an acute impairment in mitochondrial axonal transport (550) and striking mitochondrial fragmentation before neuronal demise. In fact Aβ is associated to S-nitrosylation and hyperactivation of Drp1 (148), which leads to mitochondrial fragmentation (684, 685), bioenergetics impairment, and synaptic damage in cell models of AD (449). Besides, Aβ treatment is associated to abnormally distributed mitochondria (35, 685), suggesting that Aβ may trigger excessive mitochondrial fission in AD patients, which can lead to an increase in NO^·. Aβ-induced axonal and dendritic swelling and Aβ-decreased axonal transport in hippocampal neurons (572) can contribute to altered mitochondrial distribution. Other studies also suggest that overexpression of APP and treatment with amyloid-beta-derived diffusible ligands induce mitochondrial fragmentation and abnormal distribution without cell death (684, 685), suggesting that Aβ-induced abnormal mitochondrial dynamics may play a role at early of AD pathogenesis. In fact, synaptic dysfunction is one of the early and most robust correlates of AD-associated cognitive deficits (684).

Aβ inhibits enzyme complex IV, a major source of ROS generation, even in in vitro preparations lacking biological membranes (105). In vivo, Aβ interacts with several proteins belonging to the outer mitochondrial membrane and affects the transport of nuclear-encoded mitochondrial proteins, such as subunits of the electron transport chain complex IV, as well as inner mitochondrial membrane and mitochondrial matrix proteins. Reduction of the complex IV reduces the amount of hydrogen that is translocated, thus lowering mitochondrial membrane potential. These events cause abnormal mitochondrial electron activities, resulting in the reduction of complex V activity, the decrease in ATP levels, and the increase in ROS generation [reviewed in (186)]. The 50% inhibition of complex IV by Aβ is, likely, the cause for a similar reduction in ATP generation. However, inhibition of the respiratory chain and depletion of ATP could result in oxidative stress through the production of ROS from electron leak at complexes III and I or through the previous formation of ROS damage complex IV. Moreover, ROS induce peroxidation of mtDNA and mitochondrial lipids. The complex of Aβ bound to alcohol dehydrogenase changes mitochondrial membrane permeability and reduces activities of respiratory enzymes, exacerbating ROS production and leading to mitochondrial failure. As already discussed, Aβ binding also activates Fis1 (fission protein) and promotes increased mitochondrial fragmentation, and favors the opening of the mitochondrial permeability transition pore (PTP) (186).

As we age, we accumulate wear-and-tear from oxidative mitochondrial damage, especially accumulation of toxins that reduce cell metabolic activity, favoring neurotoxicity and ROS-induced oxidative stress. This triggers the 3-R response (Fig. 14) characterized by Reset of neurons trying to repair themselves by manufacturing Aβ, which behaves as a signaling molecule that reduces energy production; removal of neurons by undergoing programmed cell death when faced with oxidative stress; and replacement of cells that faced with metabolic stress undergo cell division (619). Neurons, however, are terminally post mitotic and die if they try to divide. In fact, one of the proposed mechanisms for the generations of NFTs (others are discussed in the previous section) is this attempted remodeling of the cytoskeleton, furthering neuronal dysfunction (179).

OPEN IN VIEWER

FIG. 14.

Histopathological lesions of AD according to the view of the mitochondrial cascade hypothesis. Mitochondria serve different function as the individual ages, regulating cell proliferation, differentiation, and senescence. In agreement with this specific functions, production of ROS, as well as mutations in the mtDNA and nDNA, will increase as the individual ages. Mitochondrial dysfunction initiates compensatory events that result in the neurodegenerative sequel of AD. These compensatory responses will result in the formation of Aβ plaques and neurofibrillary tangles, and eventually will lead the cell into neurodegeneration. G2-M arrest, arrest of the cell cycle during the gap between DNA synthesis and mitosis; nDNA, nuclear DNA. Bold arrows indicate the aging process.

Mitochondria are both source and target of ROS, because an important part of oxidative damage occurs near the site of their production (658). Exposure to high levels of oxidants, especially in presence of calcium, can also induce the opening of the mitochondrial PTPs, a voltage-sensitive, nonselective ion channel that opens between the mitochondrial matrix and cytoplasm (126, 719). The opening of the mitochondrial PTPs appears to be a mediator of cell death due to an inhibition of the electron transport chain. This in turn results in oxidative stress and increased leakage of Ca²⁺ currents that induce apoptosis, by the release of cytochrome c (264, 300). In situ hybridization to mtDNA, immunocytochemistry of cytochrome c oxidase (Cyt-c), and electron micrographs of brain biopsies confirm that mitochondrial abnormalities are present in AD brains (110, 265). These studies also show that in AD patients, neurons with increased oxidative damage had increased levels of mtDNA and Cyt-c in the cytosol, which correlated with the significantly reduced number of mitochondria in these tissues.

In some AD cases, decreased Cyt-c activity was accompanied by increased free-radical generation and reduced energy metabolism (27, 265, 370). Analysis of mitochondria, isolated from platelets of AD patients, showed a decrease in Cyt-c activity and in ATP levels, associated with an increase in ROS (101). In transgenic AD mice with mutations of the mitochondria-specific antioxidant enzyme MnSOD, the partial enzymatic deficiency of mitochondrial MnSOD exacerbated AD phenotype, increasing amyloid deposition (356), tau phosphorylation (419), and accelerating behavioral impairment (192). Conversely, MnSOD overexpression attenuated the AD phenotype (183).

Inhibition of the electron transport chain (208) or enhancement of beta secretase (BACE) activity induced by oxidative stress (626) increases APP processing toward production of Aβ, strengthening the notion that not only Aβ lays at the origin of AD. Distinct mitochondrial enzyme deficiencies have been reported in AD brains, including decreased activity of pyruvate dehydrogenase complex and the enzymatic step linking the glycolysis metabolic pathway to the citric acid cycle (490, 709). As mentioned in section II.A, glucose metabolism impairment is part of the constellation of mechanisms involved in the pathogenesis of AD.

Interestingly, decreasing tau levels prevented deleterious Aβ-induced effects in hippocampal neurons (675). Notably, Aβ injections amplify tau pathology (64, 224), whereas absence of tau abolishes Aβ toxicity (288, 289), indicating that mitochondria show enhanced vulnerability toward Aβ insult in tau transgenic mice (162, 185), and suggesting the existence of a synergistic action of tau and Aβ pathologies. When expression of truncated tau was combined with Aβ at sublethal concentrations, a decrease in moving mitochondrial population (401, 637), a decrease in oxidative stress levels, and an enhancement of Aβ-induced mitochondrial potential loss were observed in neurons (514).

It is also proposed that reduced cell energy promotes tau phosphorylation in both sporadic and autosomal dominant AD (618). Accordingly, induction of tau phosphorylation is produced by prolonged fasting, complex I inhibition, carbon monoxide (CO) inhibition, mitochondrial uncoupling, and hibernation (state associated with mitochondrial uncoupling) (189, 621, 708). Tau may also contribute to mitochondrial fragmentation in AD brains. Tau is an in vitro substrate for the mitochondria caspase-3 (152, 209). Caspase-cleaved tau and tau truncated at Asp-421 to mimic caspase cleavage (T4C3) induced mitochondrial fragmentation and mitochondrial fission in a calcineurin-dependent manner (515). The expression of caspase-cleaved tau fragments in neurons can impair mitochondrial function, leading to changes in mitochondrial membrane potential, dysregulation of mitochondrial calcium level, mitochondrial integrity, and increase in mitochondrial O₂ ^·− (515). Conversely, it has been shown that antioxidant treatments inhibit glycogen synthase kinase 3β, the major kinase for tau phosphorylation (624), and that MnSOD overexpression can reduce tau phosphorylation back to normal, indicating the role of mitochondrial O₂ ^·− in tau dysfunction and, by extension, in AD.

Aβ and NFTs induce deficits in the function of the respiratory chain: acting independently as well as synergistically on mitochondrial dysfunction—dysfunction that is intrinsically tied to oxidative stress. Both events occur early in aging and in the pathogenesis of aging-related neurodegenerative diseases. In triple transgenic mice (triple AD) that combine P301L tau transgenic pR5 mice and APPswPS2N141I double-transgenic APP152, it was possible to determine functional consequences of the combined Aβ and tau pathologies. Proteomic and functional analysis showed deregulation of 24 proteins of which one-third were mitochondrial proteins mainly related to complexes I and IV. Deregulation of complex I was tau dependent, whereas deregulation of complex IV was Aβ dependent. Synergistic effects of Aβ became more robust as animals aged (538). A similar synergistic effect of aging and Aβ pathology was also observed in double Swedish and London mutant APP transgenic mice. Pronounced mitochondrial dysfunction in adult Thy-1 APP mice, with reduction of mitochondrial membrane potential and reduced levels of ATP, was already observed at 3 months of age, when elevated intracellular, not extracellular, Aβ deposits are present. Mitochondrial dysfunction was associated with higher levels of ROS, an altered Bcl-xL/Bax ratio, and reduction of complex IV activity (251). Aberrant expression of proapoptotic molecules probably contribute to neuronal loss in the late stages of AD pathology. Up-regulation of mitochondrial apoptotic molecules, including Bax and other pro-apoptotic members of the Bcl-2 familiy, is observed in AD (323, 391, 446). It has been proposed that mitochondrial protection and reduction of oxidative stress are important components of a neuroprotective treatment (352). Addition of Vits C and E effectively scavenged intracellular Aβ-mediated ROS generation, but was less effective in protecting the activity of oxidative phosphorylation enzymes and ATP generation (663). These data argue that Aβ directly interferes with oxidative phosphorylation resulting in oxidative stress.

In addition, Aβ appears to inhibit directly or indirectly key mitochondrial enzymes in the brain and isolated mitochondria, such as Cyt-c (107, 250, 527). It also reduces the activity of the respiratory chain (102, 488), pyruvate dehydrogenase, α-ketoglutarate dehydrogenase, and decreasing NADH, and, when entering the mitochondria, compromises mitochondrial function by inactivation of MnSOD, with subsequent increase of reactive species (13, 335). Other proposed target for Aβ is the PTP (105, 106). Additionally, a mitochondrial dehydrogenase/reductase, the amyloid-β peptide alcohol dehydrogenase (ABAD), that interacts with Aβ has been identified (400). Interactions between mitochondrial functions and Aβ were also validated in AD patient brains and in a transgenic mouse model overexpressing APP/ABAD, which also has elevated levels of oxidative stress and impairment of memory (387). In vitro, it has been shown that Aβ impairs mitochondrial movement inside dendritic spines (549) in cultured hippocampal neurons (550).

D. Neuroinflammation and oxidative stress

Neuroinflammation is a choreographed process involving numerous cells, including neurons, astrocytes, macrophages (mainly microglia), and lymphocytes. In contrast to the classical inflammation involving the peripheral immune system (hot, painful, swollen, and red being its key features, reflecting intense vascular reactivity), neuroinflammation is defined mainly by the activation of microglia and astrocytes, as well as by the increased levels of many molecular mediators, including cytokines like IL1β, TNFα, and TGFβ, all of which increase with aging (248, 414, 669, 674). Microglial cell cytotoxic activation and elevated levels of inflammatory cytokines further induce the secretion of more cytokines and ROS (674).

Herrup proposes that age-dependent changes gradually take their toll on the brain (260, 674), changing, among several things, glial cell reactivity (669). Under those conditions, stimulatory event or injury will be capable of initiating the disease process. The injury triggers a protective response, but the age-related impairment of normal homeostatic mechanisms determines persistance of responses, associated, for example, to a robust induction of oxidative stress that determines progression of degenerative changes (260, 669).

Neuroinflammation emphasizes crosstalk between neurons and glia and establishes a complex interaction with oxidizing agents through redox sensors localized in enzymes, receptors, and transcription factors. The redox status modulates the participation of cytokines in signaling processes, which are critical mediators of oxidative stress, neuroinflammation, and neurodegeneration in the CNS (314, 439). Oxidative stress, in turn, results in increased inflammatory cytokine production. Additionally, neuroinflammatory cytotoxic activation induces oxidative changes, creating a vicious cycle between oxidative stress and neuroinflammation [comprehensive review in (544)]. This close relationship between neuroinflammation and altered redox state has physiological implications in the maintenance of the cellular homeostasis and survival (557).

Neuroinflammation, whereas cause or consequence of neurodegeneration, can be visualized as a consequence of oxidative stress, depending on its various origins (534). Neuroinflammation promotes increased formation and aggregation of Aβ and NFT. Increased Aβ plaque and NFT then exacerbate inflammatory stress, establishing a dysregulated activation cascade with a positive feedback loop that rapidly and irreversibly progresses toward cytotoxicity (649). Besides, microglial activation combined with impaired ability to cope with increased oxidative stress in the aging brain becomes deleterious (121), producing a free-radical-mediated oxidative stress that results in neuronal injuries (61).

In the 1990s, findings of activated complement factors, cytokines, and a wide range of related receptors in the brain of AD patients led to the concept of neuroinflammation and the inflammation hypothesis for AD, which states that the innate immune response of the brain is likely to be involved in the pathology of degenerative diseases of the CNS (57, 455, 674). This new pathogenic mechanism determined the re-consideration of the role of Aβ aggregation in the pathology of AD at least for some groups, which considered glial cells to be a leading factor for this disease (669). However, for most scientists studying AD who adhere to the amyloid hypothesis, neuroinflammation is still considered as a consequence of the activation of microglia by Aβ, which initiates an inflammatory cascade, leading to degenerative changes of neurons (6) in AD (257), PD (267), and HD (57). However, the exact role of inflammation in the pathology of AD and its mechanisms in terms of the cell type that are involved—microglia, astrocytes, or other cells—are still debated.

E. Neurotoxicity, excitotoxicity, and oxidative stress

Atrophy, neuronal loss, synaptic loss, and morphological alterations by NFT, in addition to functional impairment, increase with severity of AD. Many of the pathophysiological mechanisms discussed here are involved directly or indirectly with neurodegenerative changes of the disease. Aβ at its various molecular forms, inflammatory mediators, and radical species induce functional changes and damage, either locally, at process terminals, or along the neuron as a whole, leading to loss of dendritic spines, remodeling of neurites, and inflammatory changes. Cytoskeleton changes also have a key role in neuron energetic and in the conformation and alteration of synapses and dendritic spines. Mitochondrial dysfunction, energetic impairment, calcium mishandling, and accumulation of aggregated proteins impose additional burdens, further promoting neuronal loss, with generalized neurodegeneration, ultimately leading to brain atrophy.

2. Excitotoxicity

Ionotropic glutamate receptors, especially the NMDA subtype, have been implicated in AD (570). Glutamate, the major excitatory neurotransmitter of the mammalian CNS playing important roles in development, when massively released, induces neuronal death. Necrotic death of neurons caused by glutamate excitotoxicity in AD and PD depends on increased Ca²⁺ influx, activation of Ca²⁺-dependent enzymes, induction of ROS and NO^· (Fig. 15), and oxidative stress. There is evidence that Aβ can induce glutamate release from primary cortical neurons and, in turn, alter signaling cascades that are required for long-term potentiation (646). As an additional mechanism, there is also evidence that activation of NMDA receptor is linked to ROS production from NADPH oxidase (Fig. 4b), which in turn leads to altered synaptic plasticity and memory (321).

OPEN IN VIEWER

FIG. 15.

Aging and neuron vulnerability in neurodegenerative diseases. Once triggered, the death of neurons appears to be unavoidable. Excitotoxicity and other death signals activate intracellular cascades involving increased levels of ROS and Ca²⁺, production of PAR4 and p53, and translocation of pro-apoptotic BCL2 family members (BAX and BAD) to the mitochondrial membrane. These events are followed by increased mitochondrial dysregulation and release of cytochrome c into the cytosol. Cytochrome c forms a complex with APAF1 and caspase 9. Activated caspase 9 activates caspase 3, which, in turn, cleaves proteins that result in changes in the plasma membrane, cytoskeleton, and nucleus. Caspases (e.g., caspase 8) can also be directly activated through death ligands, and can act in the absence of mitochondrial changes. Apoptosis can be inhibited at different stages through anti-apoptotic mechanisms, such as IAPs or BCL2 and BCL-XL. In general, cell fate is decided by a balance between survival factors and potentially harmful factors. AMPAR, glutamatergic α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid receptor; APAF1, apoptotic protease-activating factor 1; atPKC, atypical protein kinase C; BAD, BCL-associated death promoter; BAX, BCL2-associated protein X; BCL2, B-cell leukemia/lymphoma 2; FADD, Fas-associated death domain protein; IAP, inhibitor of apoptosis protein; PAR4, prostate apoptosis response 4 factor.

Neuronal excitation secondary to the stimulation of NMDA receptors is linked to the activation of the Ca²⁺-dependent cytosolic phospholipase A2α (cPLA2α) and release of arachidonic acid (AA), which besides being a precursor for prostanoid synthesis, is also a retrograde messenger implicated in memory formation. Activation of cPLA2α depends on phosphorylation by MAPK and PKC. Overactivation of cPLA2α is involved in different forms of neurodegenerative diseases, including AD (613).

Increasing evidence supports a link between Aβ and the NMDA receptors (592). This link involves a novel oxidative mechanism for Aβ-mediated neurotoxicity through crosstalk with the NMDA receptor contributing to the oxidative stress hypothesis and synaptic failure that underlies the pathogenesis of AD. Similar to NMDA, oligomeric Aβ can induce ROS production from cortical neurons through activation of NADPH oxidase (575). ROS derived from NADPH oxidase triggers the activation of ERK1,2, phosphorylation of cPLA2α, and AA release (Fig. 4b). In addition, Aβ-induced AA release can be inhibited by memantine, an NMDA receptor antagonist, suggesting that Aβ triggers these effects through the NMDA receptor (575).

An alternative mechanism suggested to contribute to the loss of neuronal cells by excitotoxicity induced by oxidative stress is based on the observation that in vitro and in vivo exposure to Aβ increases the content in protein carbonyls, HNE, and 3-NT in synaptosomes and neuronal cells (338). In both MCI and AD brain, the glutamate-related enzyme, glutamine synthetase (GS), is found to be oxidatively modified (88, 108), which reduces its enzymatic activity. GS is important in the conversion of extracellular glutamate into glutamine. Hence, oxidative modification and altered function of GS may acutely exacerbates the excitation of postsynaptic neurons, Ca⁺² accumulation, and free radical formation, all contributing to synaptic dysfunction and excitotoxic neuronal cell death (85).

F. Ubiquitin proteasome system, unfolded protein response, and autophagy in oxidative stress

A. Nonsteroidal anti-inflammatory drugs

Therapeutic approaches with NSAIDs, regardless the fact that neuroinflammation is involved in the pathogenic mechanisms of AD, has proven to have some effect in prevention but is ineffective for patients who already have AD. However, therapies with COX-2 inhibitors target a single pro-inflammatory factor (PGE₂) among a large number of inflammatory factors released by activated microglia, which could explain why an anti-inflammatory monotherapy could be inefficacious.

B. Modulation of microglial cell activation

Existing evidence indicates that microglia is important for maintenance, repair, and defense, and even that microglial overactivation is deleterious. Therefore, an effective therapy should target early attenuation of the microglial response to levels that are beneficial, rather than the elimination of the microglial response. Because microglial function, as well as their potential deleterious effects are regulated by NADPH oxidase and the production of intracellular and extracellular ROS (60, 61), this enzyme complex appears as an ideal therapeutic target. Recently, several peptides, antibiotics, and small molecules have been identified as inhibitors of NADPH oxidase and, hence, potential neuroprotective agents (150, 509). Inhibition of NADPH oxidase activation might be a more efficacious approach because it targets the predominant enzyme complex by which microglia produce high amounts of ROS. Its inhibition would down-regulate multiple inflammatory factors, including PGE₂ (683). In addition, although they are not discussed here, because they are outside the scope of this review, some cytokines released by microglia have been considered as possible targets of drug therapy for AD patients (226, 257, 348).

C. Activation of antioxidant pathways

An alternative approach for decreasing the production of ROS would be the activation of antioxidant pathways, an approach that could be particularly important for the brain, an organ that has relatively weak antioxidant defenses. Moreover, aging leads to some loss of the free radical scavenging ability depending on endogenous mechanisms. Even more, reduced cellular expression and activity of antioxidant proteins are fundamental triggers for AD. Among cellular antioxidant defenses, Hsps have been regarded as cytoprotectors from oxidative damage in neurodegenerative diseases. HO-1 polymorphisms have been considered associated with increased AD susceptibility. Deregulation of the HO system has been associated with the pathogenesis of both AD, and brain aging (405, 477). Among the stress proteins, HO-1 is encoded by the HO-1 gene that is regulated by redox, and its activation represents a protective system potentially active against brain oxidative injury. Its expression in AD patient brains is significantly increased in neurons and astrocytes of the hippocampus and cerebral cortex. In individuals with AD, HO-1 co-localizes in senile plaques and NFTs being its spatial distribution, similar to that found in the pathological expression of tau. There are reports that strongly suggest that activation of HO-1 is strongly protective against oxidative damage and cell death in neurons. Thus, modulation of HO-1 should represent a potential pharmaceutical strategy for the treatment of neurodegenerative disorders (517).

D. Mitochondrial antioxidants

Considering the important component of cell death observed in AD and the fact that mitochondria also have a key role in production of ROS and in apoptosis signaling, therapeutic strategies aimed to reduce mitochondria damage appears as an interesting approach for the treatment of neurodegenerative diseases. Several compounds targeting mitochondria are currently undergoing clinical trials. Mitochondrial antioxidants appear to be especially efficient at preclinical level. Coenzyme Q10 (CoQ10, an electron carrier embedded in the inner mitochondrial membrane, part of the electron transport chain of oxidative phosphorylation), has been shown to have neuroprotective effects in several neurodegenerative diseases by attenuating mitochondrial dysfunction. However, the poor capability of these oral antioxidants to cross the BBB has slowed down their therapeutical use, focusing the investigation on more soluble, shorter chain antioxidants of the family of CoQ10 derivatives, such as idebenone [6-(10-hydroxydecyl-2,3-dimethoxy-5-methyl-1,4-benzoquinone], decylubiquinone, and the ubiquinone derivative MitoQ10, which appears to be the most successful. MitoQ10 has the advantage of being accumulated within mitochondria, where it is reduced into its active ubiquinol form, which can prevent mitochondrial oxidative damage (383).

Szeto-Schiller (SS) peptides represent another class of mitochondrial antioxidants that offer the advantage of localizing within mitochondria at all mitochondrial membrane potential; a characteristic that may improve their therapeutic use. In vivo experiments have revealed that SS peptides are protective in a mouse model of ALS, increasing survival and motor performance, and decreasing cell loss (430). SS peptides also protect dopaminergic neurons against 1-methyl-4- phenyl-1,2,3,6-tetrahydropyridine neurotoxicity, both in vitro and in a PD animal model (430).

A drug rising high expectations over the last few years has been latrepirdine (dimebon), known antihistaminic drug, with a novel mitochondrial mechanism of action, associated to the improvement of mitochondrial stability by preventing mitochondrial permeability pore transition, in addition to effects as an NMDA receptor inhibitor. However, a phase III AD trial of Dimebon in 598 patients failed to show any significant improvement in primary or secondary outcomes (50).

E. Free radical scavengers

Another area of therapeutic development has been the use of free radical scavengers, in order to slow neuronal damage associated with Aβ-induced oxidative stress, acting as neuroprotective agents. Various herbs and fruits contain active phenolic substances endowed with potent antioxidant and chemopreventive properties, which have been studied for potential neuroprotective effects. A promising antioxidant already being tested in clinical trials is curcumin, a polyphenolic substance derived from the root Curcumin longa. Epidemiological studies have suggested that curcumin, one of the most prevalent nutritional and medicinal compounds used in India, appears to be responsible for the significantly reduced (4.4-fold) prevalence of AD in India compared with the United States. Consistent with these data, convincing evidence has been provided that chronic (6 months) dietary curcumin given to an AD transgenic mouse model resulted in a suppression of indices of inflammation and oxidative damage in the brain of these mice and reversed Aβ-induced cognitive deficits (220, 723). Curcumin attenuates oxidative damage in cortical neurons by reducing the intracellular production of ROS and protecting mitochondria (723). Curcumin inhibits lipid peroxidation and effectively intercepts and neutralizes ROS (superoxide, peroxyl, and hydroxyl radicals) and NO·-based free radicals (NO · and ONOO⁻). In addition, it has the potential to reduce cytochrome c release, block caspase-3 activation, and alter the expression of Bcl-2 family proteins (412, 517, 723). Of particular interest is the capacity of curcumin to inhibit COXs and to reduce the activation of nuclear transcription factor NF-kB, key in inflammatory signaling pathway. As for its ability of increasing antioxidant capacity, low concentrations of curcumin show a robust induction of the expression and activity of HO-1 in vascular endothelial cells, in rat astrocytes, and in cultured hippocampal neurons (412, 517, 723). Such ability of curcumin to induce HO-1 has been associated to its strong antioxidant and anti-inflammatory properties (220). Preincubation of neurons with a low concentration of curcumin in culture enhanced cellular resistance to oxidative damage. Also, depending apparently on a different mechanism, curcumin is effective protecting cortical neurons against apoptotic death induced by Aβ.

Curcumin is highly lipophilic and might cross the BBB, and, regardless its low bioavailability, because it is rapidly metabolized, curcumin may reach high brain concentrations to decrease Aβ aggregation at least in animal models. Nowadays, there is a randomized, controlled trial study of curcumin (ClinicalTrials.gov Identifier: NCT00528151), but it is oriented to a different degenerative disease and not to AD patients.

Curcumin is safe, is well tolerated, and appears to lack side effects at doses up to 12 g/day (220, 268). However, there are reports showing some effects that potentially could be deleterious. Inhibition of NF-κB by curcumin (17) could affect survival mechanisms associated to inflammatory activation. Similarly, curcumin is a specific inhibitor of iNOS (369), effect that also can depend on its inhibition of NF-κB, responsible for initiating the expression of iNOS. As inhibition by curcumin attenuates induction of NO production, this effect leads to an overall increase in ROS levels due to a failure in the scavenging of superoxide by NO.

Moreover, in PD animal models, exposure of rat mesencephalic cells to curcumin induces the expression of LRRK2 at the mRNA and protein level; LRRK2 overexpression is strongly associated with pathological cellular inclusions found in several neurodegenerative disorders (472). Curcumin has also been identified as an activator of apoptosis. Subtoxic concentrations of curcumin in a PD model sensitize mesencephalic cells to PQ-mediated apoptosis (471). Apoptosis appears to be enhanced via a mechanism that is dependent on the generation of rROS (52, 160, 303, 605, 636, 696), although other reports show that curcumin inhibits ROS, and consequently apoptosis, through its well-known antioxidant properties (142, 144).

Curcumin, although a potent antioxidant with ROS scavenging properties, is able to increase ROS production in different models (293, 451). Interesting enough, although cell death after the PQ+curcumin exposure was an ROS-dependent event, antioxidants were able to significantly protect against cell death generated by the exposure of cells to PQ alone, but were unable to prevent cell death in PQ+curcumin co-treatment, suggesting that the co-incubation sets cells into a no-way-back cascade leading to cell death that could not be retrieved by antioxidants (471).

Other plant-derived phenolic agents, with analogous chemical structures to curcumin, produce also a strong activation of HO-1 expression and enhancement of cellular defense against oxidative stress. On the same line of antioxidants, in principle, iNOS inhibitors could be used for AD treatment. However, there are no specific drugs being tested (476). There is evidence of beneficial effects of resveratrol on Aβ cytotoxicity in culture. However, protection by resveratrol was not associated exclusively to inhibition of iNOS expression, but it also attenuated PGE₂ release, suppression of COX-2 expression, and suppression of nuclear translocation of NF-κB. The multiplicity of anti-inflammatory effects precludes the assignment of resveratrol protective effect to iNOS inhibition (320). There are also reports on the in vivo effect of three NSAIDs, lysine clonixinate, indomethacine, and meloxicam on NO production and iNOS expression in rat cerebellar slices (174). However, as resveratrol, reduction of iNOS and nNOS levels is only one of several anti-inflammatory effects of these NSAIDs (174). Nevertheless, the decrease in NOS expression after in vivo treatment with NSAIDs could represents a novel mechanism of therapeutic action of these NSAIDs, and may also be of importance in the prevention of NO^·-mediated neuronal injury and as new strategies to reduce the risk of AD. Currently, a phase-III, randomized, double-blind clinical trial oriented to AD patients (ClinicalTrials. gov identifier: NCT00678431) to evaluate primary efficacy of resveratrol for disease progression is being done. Results in terms of the disease outcome are not available yet.

Antioxidant treatment with vitamin E (α-tocopherol) has been pursued with phase II and III clinical trials. Despite initial promise, recent randomized, controlled trials in MCI fail to report any benefit (286, 493). A problem of the antioxidant studies is the dissimilar nutritional status of patients, as well as the potential need of using co-administration of several antioxidants. Furthermore, α-tocopherol may also have different effects within any given population of AD patients (390), recognizing groups of responders and nonresponders α-tocopherol. In a recent meta-analysis of 19 clinical trials, only small trials show either an increase or a decrease in all-cause mortality and that the overall effect is near zero (422). In an effort to understand the lack of effect, AD patients were stratified into α-tocopherol respondents and nonrespondents (389), based on measures of plasma GSSG, the oxidized form of the common antioxidant GSH. Even at high doses of α-tocopherol, half of the patients failed to lower the plasmatic levels of plasma-GSSG, and showed a continuous deterioration in their cognitive performance (389). The other half of the patients, those showing decrease of plasmatic levels of plasma-oxidized GSH with α-tocopherol treatment, had no significant difference in the cognitive performance compared with the first group.

Considering the increase of oxidized macromolecules with aging and AD, one wonders why the antioxidant α-tocopherol is ineffective. The answer may be that a lipid-soluble antioxidant like α-tocopherol, with a high membrane/water partition, could be ineffective protecting against oxidation of aqueous phase nucleic acids and proteins. In addition, if the oxidized α-tocopherol is not removed from the membrane, it will either accumulate or pass the electron to another lipid, inducing further damage to the membrane. Because oxidized α-tocopherol can be recycled by passing the extra electron to a water-soluble electron acceptor such as vitamin C or other compounds, combining antioxidants could be a good strategy for improving their antioxidant effects. There are some reports showing better results with the combined use of vitamins E and C for enhancing the reduction in lipid peroxidation seen in AD patients after treatment with α-tocopherol (513). Effects may depend on the intrinsic antioxidant potential of vitamin C or on its interaction with α-tocopherol, as it is known to recycle α-tocopherol radical to maintain its antioxidant properties. However, several trials have failed in showing improvement of cognitive performance. Failure could depend on the fact that even more complex mixtures of antioxidants, including flavonoids and polyphenols—which have shown efficacy in animal models of aging—may be needed, or on the fact that aging persons have an increasingly oxidized plasma redox potential, which can interfere with removal of radicals (302).

A recent study showed that serum α- and γ-tocopherol levels were not associated with the risk of developing AD (616). However, the use of various tocopherol forms, other than just α-tocopherol, may be key conditions for AD (432). A dementia-free cohort was followed for 6 years to detect incident AD, focusing on α-, β-, γ-, δ-tocopherols and α-, β-, γ-, δ-tocotrienols. The authors found that the risk of developing AD was reduced only for high plasma levels of β-tocopherol, while the other subunits had only a small effect (402).

Another possibility for the failure of simple antioxidants to reverse damage could be that we are aiming at the wrong target: ROS signaling is required for processes as diverse as transcription factor activation, insulin signaling, endothelial function, or LTP model of memory. Also, ROS damage could depend on a variety of co-morbidities. We may need a dose–response targeted reductive shift in the metabolically oxidized redox potential seen in many aging individuals. Thus, vitamin E therapy trials are needed, with a special focus on individual monitoring of redox potential to avoid toxicity and assess biomarkers of efficacy (389).

F. Cyto- and neuroprotective therapies

As previously outlined, Nrf2-induced ARE-Nrf2 signaling leads to a broad cytoprotective response associated with enhanced antioxidant capacity and attenuation of neuroinflammation (660). In view of the multiple mechanisms of its protective effects, Nrf2-ARE pathway could represent a potential therapeutic target for AD treatment (306). In vitro and in vivo studies show that monofunctional inducers exhibit profound neuroprotective effects in an Nrf2-ARE-dependent manner. In contrast, in Nrf2 KO mice and after Nrf2 disruption, the chemically induced ARE-mediated transcriptional response, as well as the chemopreventive efficacy of monofunctional inducers, is completely abolished (344, 417). In addition, it has been shown that curcumin and pyrrolidine dithiocarbamate, capable of inducing Nrf2, alleviate cognitive defects in AD transgenic models (202). Because clinically safe compounds are widely available in the form of chemopreventive agents, the Nrf2-ARE signaling pathway can become a key drug target for neurodegenerative disorders.

A. Future directions of research in this field

The role of microglia in AD has received increasing attention as possible avenues for therapies that modify disease outcome in AD. For example, minocycline studies point out the complexity of the effects of microglial manipulation in AD models. Various effects of minocycline on microglial activation and cognitive function predominate during different phases of the life spans of experimental animals, suggesting that microglial inhibition may be beneficial at one phase of disease progression but detrimental at others. A critical area of research involves the dissection of these mechanisms that may lead to the development of drugs with more clearly defined profiles of efficacy and fewer side effects. It is clear that causes and consequences of microglial cell activation are complex, and pharmacologic manipulation should be directed to selected properties of microglial cell function oriented to the promotion of protective functions while minimizing oxidative and neuroinflammatory damage.

An interesting combination for immunomodulatory therapy would be the use of antioxidants. Once again, antioxidant therapy should take into account that low concentrations of radical species serve physiological functions, including important roles in cell signaling. Thus, any treatment should be oriented to modulate ROS and RNS production rather to abolish them. Furthermore, if it probes possible to target microglia and oxidative stress modulations as an effective therapy for AD, other neurodegenerative diseases may prove similarly susceptible to their benefit.