Neuronal Failure in Alzheimer Disease: A View through the Oxidative Stress Looking-Glass

Abstract

There is considerable debate and controversy surrounding the cause(s) of Alzheimer disease (AD). To date, several theories have gained notoriety, however there has yet to be any one that garners universal acceptance. In this review, we provide evidence for the oxidative stress-induced AD cascade that posits aged mitochondria as the critical origin of neurodegeneration in AD.

Introduction

Alzheimer disease (AD) is a neurodegenerative condition responsible for the cognitive deterioration of 26.6 million people in the world and it was projected that the worldwide prevalence of AD will grow four fold to 106.8 million with 1 in 86 people living with AD by the year 2050 where 59% of cases will be in Asia¹. Age is the primary risk factor for AD such that the incidence of the disease is 15% among individuals over the age of 65 and almost 50% among those over 85². While many specific aspects of AD have been documented, there has yet to be any clear understanding of its pathological initiation and progression. To be sure, molecular aberrations in the cell elicit neuronal failure through a variety of well-established mechanisms, including: i) oxidative stress³; ii) abnormal protein folding and aggregation⁴; iii) cell cycle dysregulation⁵; iv) mitochondrial dysfunction⁶; v) synaptic failure^{7, 8}; vi) inflammation⁹; vii) loss of calcium regulation¹⁰; viii) defective cholesterol metabolism^{11, 12}; ix) vascular alterations^{13, 14}; and x) neurotrophin deprivation¹¹. However, the causal relationships governing these factors remain unknown.

AD is described by the successive degeneration of neurons first in the entorhinal cortex of the mediotemporal lobe, followed by those of the CA1 region of the hippocampus, the CA 2/3 regions, the CA4 region, and the neocortex¹⁵. Despite this predictable vulnerability, the molecular mechanisms governing it are unclear.

Amyloid-β (Aβ) and the hyperphosphorylated form of the microtubule-associated protein tau are two hallmark pathologies of AD that are critically involved in neuronal death. While the function of Aβ is unclear, its accumulation into insoluble fibrils and plaques increases over the course of AD progression, causing severe cellular burden. The peptide is formed by the proteolytic cleavage of its protein precursor, amyloid-β protein precursor (AβPP), via sequential enzymatic actions of β-site AβPP cleaving enzyme 1 (BACE-1) and γ-secretase, a protein complex with presenilins 1 and 2 at its catalytic core^{11, 16}. Aβ is associated with multiple cascades that are thought to result in neuronal damage^{17, 18} and as such, Aβ is generally considered the primary mediator of neurodegeneration in AD¹⁹. The Amyloid Cascade Hypothesis maintains this position, although its lack of therapeutic translation has become a major criticism²⁰.

The microtubule associated protein tau is similarly associated with AD progression, and its accumulation in the form of neurofibrillary tangles (NFTs) is a strong correlate of dementia in AD^{11, 21}. NFTs are used as a post mortem diagnostic criterion for AD²². Tau becomes phosphorylated on serine and threonine residues that flank the microtubule binding domain. Upon hyperphosphorylation, tau is unable to bind to and stabilize cytoskeletal microtubules and instead self-aggregates into NFTs in and around the cell²³.

AD is typically a sporadic condition. 90% of all cases are sporadic in nature and are characterized as late-onset AD (LOAD), with an age of onset of around 65 years. Familial AD (FAD), on the other hand, is an early-onset, genetic condition caused by several known autosomal dominant mutations; its age of onset is typically around ages 40–60²⁴. Mutations that yield FAD all involve or result in an alteration in the ratio of Aβ₄₂/Aβ₄₀; this is perhaps the most touted evidence both for and against^{25, 26} the aforementioned Amyloid Cascade Hypothesis²⁷. Dominantly inherited mutations in presenilin 1 (PS1), an essential component in the γ-secretase complex, account for the majority of FAD cases, and more than 170 such mutations have been identified in PS1 on chromosome 14²⁸. In addition to PS1, mutations in presenilin 2 (PS2) (on chromosome 1²⁹) and AβPP (on chromosome 21³⁰) also elicit FAD, although they are less prevalent. These mutations also affect the ratio of Aβ_42/40. Importantly, both LOAD and FAD present identical brain lesions and patterns of neurodegeneration. Notably and somewhat contradictory, LOAD leads to a reduced ratio of Aβ_42/40, whereas FAD leads to an increased ratio³¹.

Neuronal Oxidative Stress: Sources and Vulnerabilities

Aerobic respiration, like any other biochemical or physical process, is not 100% efficient. The mechanism by which mitochondria process carbohydrates and establish a proton gradient for the synthesis of ATP (i.e., the tri-carboxylic acid (TCA) cycle and oxidative phosphorylation) involves the controlled transfer of electrons from strong reducing agents to strong oxidizing agents. Some of the free radicals thus generated escape; rather than appropriately transporting their electron to the next molecule in the cascade, they abandon the inner membrane of the mitochondria and impose alterations on other macromolecules within the cell. These alterations are often detrimental: DNA/RNA oxidation may yield fragmentation and deficiencies in repair machinery^{32, 33}; oxidative modification of enzymes and metabolic signaling proteins may lead to metabolic impairments³⁴; and protein cross-linkages and impaired proteolysis network resulting from oxidative modification may render proteins insoluble and prone to aggregate abnormally^35–37.

The ability of the cell to adapt to oxidative damage is robust, yet finite. Endogenous antioxidants are intended to sequester the formation of free radicals. Superoxide dismutase, for instance, catalyzes the conversion of the potent free radical superoxide to hydrogen peroxide and water. Other important oxidative stress proteins include catalase, glutathione peroxidase, glutathione reductase, and heme oxygenase³⁸. Under normal conditions, the majority of free radicals are sequestered within the mitochondria (i.e., their place of origin). Notably, however, three important factors (age, metabolic demand, and disease) exacerbate the vulnerability of cells to oxidative burden.

Statistically, the longer a cell is respiring, the more reactive oxidative species (ROS) go unsequestered within that cell. It is estimated that, on average, cells utilize 10¹³ molecules of oxygen per day, with 1% of the associated reactions producing unsequestered ROS. Therefore, on average, 10¹¹ molecules of ROS are generated in a given cell every day. Even with sufficiently efficient antioxidant mechanisms, some ROS will unavoidably escape and cause damage to the surroundings which may accumulate with time, especially in long living cells. Age is thus a great risk factor for oxidative stress. Likewise, high metabolic activity, which requires more oxidative phosphorylation per unit time, and thus involves a greater number of toxic species generated per unit time, also renders a cell more vulnerable to oxidative insults. A neuron, which requires energy for axonal transport, vesicular release, ion pump operation, electrochemical gradient maintenance, and the like, requires much more oxygen per unit time than any other cell in the body. Metabolic demand is therefore another risk factor for oxidative stress. Disease conditions that produce mutations in cellular antioxidant defensive machinery provide a third risk factor for oxidative damage.

Altogether, the brain represents a tremendously vulnerable region to oxidative damage. Despite comprising only 2–3% of total body mass, 20% of basal oxygen supply is utilized by the brain. Moreover, neurons are among the longest-lived cells in the body and thus must continue to survive and function for decades with the same metabolic machinery. The presence of transition metals in the brain that catalyze oxidative reactions furthers the appearance of oxidative pathologies. Iron and copper, for instance, increase the likelihood of an oxidation/reduction reaction taking place³⁹, and the regional concentration increases of these metals in the brain compound the risk factors described above. Furthermore, the brain suffers a relative paucity of antioxidant systems as well as an increase in polyunsaturated fatty acids (prime targets for ROS)⁴⁰. Given the aforementioned, an aging brain is tremendously susceptible to oxidative deterioration.

Oxidative Stress: Prominent and Early Feature of Alzheimer Disease

Free radical-induced damage to macromolecules has been well documented in AD and deposition of redox-active ions, capable of generating the most damaging hydroxyl radicals through Fenton reaction, likely exacerbates both the spectrum of molecules and cellular areas affected. DNA and RNA oxidation, marked by increased levels of 8-hydroxy-2-deoxyguanosine and 8-hydroxyguanosine (8-OHG), suggest a higher frequency of DNA fragmentation and aberration in DNA repair⁴¹. Oxidative modification of metabolic proteins, such as creatine kinase BB, cytochrome c oxidase (COX), and ketoglutarate dehydrogenase complex (KGDH), have been demonstrated by elevated levels of protein carbonyl and nitration of tyrosine residues ⁴² and indicate impaired metabolic activity. Lipid peroxidation, marked by thiobarbituric acid reactive substances, malondialdehydes, 4-hydroxy-2-transnonenal (HNE), and isoprostane, as well as by altered phospholipid composition, suggests altered membrane integrity. Oxidative modification of sugars, marked by increased glycation and glycooxidation, indicates impaired cellular ability to adequately process critical carbohydrates⁴³. Each of these aspects is elevated in AD as compared to control cases⁴².

Moreover, mitochondrial abnormalities, attributed to oxidative stress-related damages, are well-established in AD. As stated above, specific enzymes involved in electron transport and the TCA cycle, such as KGDH and COX, have been documented as modified via ROS in AD^{44 45}. Mitochondrial DNA (mtDNA) modification⁴⁶ and calcium dysregulation^47–49 occur in AD to a great extent, and increased numbers of mitochondria with broken cristae, altered size and shape, and aberrant intracellular localization are elevated in AD neurons^{47, 50, 51}. These latter phenomena are the result of impaired mitochondrial dynamics that can be caused by and also amplifies oxidative stress⁵². Mitochondrial fission and fusion, the ongoing processes that ensure organelle stability within the dynamic cellular environment, rely on critical membrane proteins to successfully occur. Fusion enables the exchange of lipid membrane in inter-mitochondrial components (i.e., mtDNA and fission/fusion proteins) such that the cell is populated by healthy, normal mitochondria⁵³. Fission, on the other hand, coupled with fusion and autophagy, enables the sequestration and elimination of irreversibly damaged mitochondria and mitochondrial content⁵⁴. These dynamics are the necessary means by which the cell dampens its inevitable accumulation of oxidative free radical-induced damage. Over the lifetime of a cell, however, oxidative damages and transcriptional errors (due to alteration of mtDNA) reach a critical threshold. Oxidative stress ultimately propagates throughout the cell as mitochondria become abnormally shaped and localized^{47, 55–58}.

Importantly, increasing evidence demonstrates that the oxidative modifications occurring in vulnerable neurons in AD occur prior to hallmark pathologies of the disease ³⁵. That is, the markers of oxidative damage are found in the cytoplasm of neurons prior to any indication of degeneration in AD brains^{59, 60}. In fact, the hallmark pathologies such as amyloid plaques and neurofibrillary tangles may be an adaptation in response to elevated oxidative stress. More recent studies found that patients who suffered from mild cognitive impairment (MCI), considered as prodromal stage of AD, demonstrated depleted antioxidants along with increased lipid peroxidation in the plasma and lymphocytes⁶¹. Elevated protein/RNA oxidation and increased redox-active iron were also documented in the brain of MCI^{32, 62, 63}. Transgenic mouse models of AD also demonstrate oxidative damage prior to Aβ deposition⁶⁴. The mitochondrial abnormalities described above provide considerable support for this primacy. Specifically, vulnerable neurons in AD and MCI exhibit severe metabolic deficiencies well before any clinical manifestations of disease. Neuroimaging studies and neuropsychological tests have indicated impaired cerebral metabolism prior to any evidence for functional impairment or brain atrophy induced by AD pathologies^{51, 65}. This metabolic impairment is likely the result of the alterations in mitochondrial dynamics and intracellular localization⁶⁶, which, in turn, increases the production of oxidative damage to mtDNA and membrane proteins leading to a vicious cycle.

Alzheimer-Specific Factors and Oxidative Stress

Besides the fact that oxidative stress characterizes aging, the greatest risk factor for AD, genetic and pathological factors associated with AD are also implicated in the regulation or being regulated by oxidative stress. For example, presenilins are needed for the import of around 50% of cellular copper and zinc⁶⁷ and increased presenilin 2 expression increases DNA fragmentation and produces apoptotic changes⁶⁸, which are both important consequences of oxidative damage. The translation of amyloid precursor protein (APP) is regulated by iron influx via an iron-responsive element (IRE) RNA stem loop in its 5'-untranslated region⁶⁹. APP is the dedicated ferroxidase in neurons responsible for the export of irons and the elevation of redox-active iron in neurons in AD is likely due to reduced ferroxidase activity of APP inhibited by zinc⁷⁰. ApoE is a strong chelator of copper and iron, both of which are important redox-active transition metals⁶⁸. Apo E has a beneficial effect against free radicals since it reduced neuronal death caused by hydrogen peroxide and β-amyloid through antioxidant activity in an isoform-dependent manner with E4 being the least effective⁷¹. Interestingly, apo E is sensitive to attacks by free radicals, with the responses being isoform dependent such that E4 is most vulnerable⁷². Indeed, cerebral cortex of autopsied brain samples from patients with APP or presenilin gene mutations or apoE4 allele demonstrated higher oxidative stress^{73, 74}.

Aβ may itself elicit oxidative damage to surrounding cells through, for example, activation of microglia and astrocytes that activate multiple ROS pathways⁷⁵, including an upregulation of L-tryptophan metabolism through the kynurenine pathway⁷⁶. Several intermediates in this pathway, namely 3-hydroxykynurenine and quinolinic acid, are neurotoxic, and the Aβ-induced increase in their production subjects surrounding neurons to oxidative damage^{77, 78}. Aβ-induced lipid peroxidation yields several reactive aldehydes (including HNE) that are capable of harmfully modifying membrane proteins. HNE production specifically disrupts iron homeostasis by impairing glucose transporters^79–82. Iron and copper accumulation within neuritic plaques (chiefly composed of Aβ) further facilitate free-radical generation⁸³. When Aβ binds catalytic amounts of copper, hydrogen peroxide is created, which contributes to oxidative stress as it produces hydroxyl radical via the Fenton reaction.

Aβ has been demonstrated to have antioxidant capacities intracellularly and may actually be secreted as a compensatory measure against oxidative stress^{35, 84}. That is, in vivo and in vitro studies of neuronal responses to oxidative stress indicate an increase in Aβ production⁸⁵, which is followed by a corresponding reduction in oxidative stress^{86, 87}. Moreover, markers of oxidative stress, such as 8-OHG, are known to be reduced in neuronal populations characterized by Aβ deposition, and elevated in vulnerable regions lacking Aβ ^{86, 87}. The data clearly suggests that the neurons, in the former scenario, have been salvaged from oxidative stress by Aβ.

Tau hyperphosphorylation has been demonstrated as a pathological result of oxidative stress^{59, 84, 87–89}. It is intriguing to note that most neuronal loss during the course of neurodegeneration occurs where the levels of oxidative stress are the highest and subsequent deposition of NFTs decreases these levels⁵⁹. Of note, and supporting this, neurons that accumulate NFTs are able to survive for decades and be functionally integrated in cortical circuits^{90, 91}. Mechanistically, phosphorylation of tau antagonizes apoptosis by stabilizing beta-catenin⁹². Regardless, the abnormal accumulation of hyperphosphorylated tau in the form of NFTs occurs subsequently to oxidative stress-induced damages.

Oxidative Stress and Alzheimer disease: Ubiquity vs. Specificity

Disease is defined by the deficits in specific functional output underlied by anatomical and biochemical/structural changes elicited by insults and adaptations and/or failure of adaptations. Interestingly, oxidative stress is a prominent biochemical change not only found in AD, but also found in almost all major neurodegenerative diseases including Parkinson disease, amyotrophic lateral sclerosis and Huntington disease⁹³. Since oxidative stress is considered a major contributor to aging, the greatest risk factor for all the age-related neurodegenerative diseases, it is perhaps not surprising that oxidative stress is believed to be a causative factor or at least an ancillary factor in the pathogenesis of these diseases. However, given the different neuronal populations involved and distinct pathology that developsd in these various neurodegenerative diseases, such a ubiquity of the presence of oxidative stress raises the obvious question of specificity of its role in the pathogenesis of each of these diseases. For example, how oxidative stress specifically leads to neuronal death in hippocampus and cortex in AD while in substantia nigra in PD? One possibility is that the selective neuronal death may be caused by increased oxidative stress in selective brain regions in these diseases. For example, in AD, large body of evidence support the elevated oxidative stress in disease-affected area such as hippocampus and cortex but not in disease-spared area such as cerebellum⁴². Similarly, heightened oxidative stress is consistently demonstrated in substantia nigra but not in other unaffected brain region in PD⁹⁴. If this is the case, the question becomes how selective oxidative stress occurs in these diseases, especially in the familial cases where causative mutations in specific genes affect all cells? It is likely the specific characteristics of various neuronal populations involved in different diseases contribute to such selectivity. Neurons with long axons and multiple synapses have higher metabolic demands that may render them more prone to oxidative attacks at a steady state level and thus become more vulnerable to additional disease-related changes. For example, CA1 neurons of hippocampus demonstrate higher levels of superoxide anion than parietal cortical or CA3 neurons which is at least one of the reasons that they are more vulnerable to global ischemia-induced dell death⁹⁵. Dopaminergic neurons are exposed higher steady state levels of oxidative stress produced by metabolism of dopamine which make them more vulnerable to insults that affect the dopamine metabolism⁹⁶. It is also possible the selectivity is due to the specificity of initiating factor(s). Such initiating factor(s) may not be a single event, but more likely a complex interaction between the genetic/epigenetic background of each individual and the environment which may determine the specific neuronal population affected. Overall, different diseases may each display a distinct oxidative stress pattern that likely will lead to a different homeostatic balance that is finely tuned to minimize cell death and more studies are clearly necessary to understand the role of oxidative stress in each disease.

The Age-Related Mitochondrial Cascade of Alzheimer Disease

Combining the evidence listed above reveals the following: i) AD is primarily an age-related disorder; ii) the brain is the most metabolically demanding region of the body; iii) free radical release invariably results from the tremendous amount of oxidation/reduction reactions that take place within neurons; iv) resident antioxidant systems, though robust, are ultimately finite in capacity; v) mitochondria are the metabolic centers of the cell and experience the earliest detriments in AD; vi) Aβ peptides and NFTs accumulate after oxidative stress has already incurred severe damage, and seem to exert a compensatory response to cerebral oxidative stress. Fitting the pieces together, we assert that over the course of a lifetime, statistically guaranteed free radical damage accumulates within mitochondria, affecting mtDNA, membrane proteins, and calcium homeostasis. The antioxidant response of neurons is able to stagnate the effects of this stress for years; metabolic deficits resulting from mitochondrial dysfunctions and perturbed localizations do not appear until well into adult life, and are the very first aspects of dementia. However, the increased presence of transition metals and polyunsaturated fatty acids in the brain render neurons particularly susceptible to damage, and once mitochondrial mutations become widespread within the cell, compensatory responses, such as Aβ secretion, begin to appear to fight further damage. Eventually, secreted Aβ becomes subject to oxidative stress (i.e., dityrosine cross-linkages inhibit its solubility, promoting its aggregation^{62, 97}). The cascade worsens as Aβ then induces neuronal detriments, producing further oxidative stress, inflammation, and synaptic dysfunction. The well-established pathophysiological inclusions of AD, including cell cycle aberration, inflammation, synaptic dysfunction, and loss of calcium regulation, become increasingly relevant, and affected neurons ultimately die. Thus, after decades of stress and increasing malfunction, cognitive decline spreads and yields clinical AD (Figure 1).

Figure 1.

Oxidative stress-induced cascade of Alzheimer disease. As stated in the text, the inevitable generation of free radicals within neurons eventually compromises the integrity of mitochondria, leading to a cascade of destructive events that elicits the hallmark features of Alzheimer disease. After years of accumulated burden, neuronal death and cognitive decline result.

Though its logic is quite appealing, this description of AD, like all previous hypotheses posited as explaining the disease, suffers some significant shortcomings. First, it does not adequately explain the predictable and consistent spread of neurodegeneration from the entorhinal cortex of the mediotemporal lobe, though the hippocampus, and into the neocortex. Perhaps this orderly passage reflects the metabolic demands of distinct regions within the brain. That is, the entorhinal cortex may require the highest metabolic activity due to its significant involvement in memory formation and retrieval (an ongoing process that occurs continuously throughout life). It would thus be the first victim of the oxidative stress accumulation that leads to degeneration by a matter of simple probability. The remaining neurons in the connective pathway would suffer by connective association, and the disease would spread. Notably, no evidence for this postulation exists, although its potential validity merits further investigation.

The mitochondrial cascade also seems to fail to explain the critical role of AD mutations in FAD and Down’s syndrome. A more careful look, however, ameliorates this misperception. In particular, the timeline of LOAD is such that Aβ accumulation and aggregation occurs late in adult life, incurring deficits on cognition typically after age 65. The mitochondrial cascade attributes this latency to the prolonged period of oxidative stress accumulation that takes years to elicit any real metabolic deficits and subsequent Aβ secretion. Once Aβ is secreted, however, and becomes oxidatively processed into insoluble fibrils, it exerts damage to surrounding cells via a variety of mechanisms. FAD represents a shortcut in this otherwise lengthy process. That is, mutations that increase the cleavage and processing of AβPP subject neurons to an early secretion and aggregation of Aβ. Affected patients thus skip the necessary accumulations of mitochondrial damage that typically take decades to elicit AD pathologies. Rather, Aβ is aberrantly cleaved regardless of the endogenous oxidative environment, and FAD patients experience clinical symptoms much earlier than do LOAD patients. Again, little experimental evidence has supported this postulation; however, it seems quite likely given the mountains of data.

Therapeutic Implications and Conclusions

We are currently faced with a strong incentive to produce a therapeutic agent that prevents AD onset. Based on the evidence here presented, the most likely method of adequate prevention should come from antioxidant administration. Preventing the accumulation of damaging species within vulnerable neurons would ideally protect them from the deleterious cascades that ROS accumulation yields. Unfortunately, there has been little success with antioxidant therapies thus far⁹⁸. It must be emphasized that the failure of antioxidant clinical trials does not necessarily nullify the hypothesis that oxidative stress being a critical contributor to the disease because of the many limitations such as that in patient selection (it would be preferable to select patients with low antioxidants) and in monitor/analysis of the respondents (it would be preferable to monitor oxidative stress surrogate markers before and during the treatment to make certain the therapy compliance) involving such clinical studies. Better antioxidants, especially the ones that target the major ROS production sites, are needed. Several agents under investigation, such as the MitoQ, acetyl-L-carnitine, and α-lipoic acid, do show potential; however, an effective method of treatment is far from complete^{47, 99–110}. Based on the many prospective studies and the complexity of the redox system in vivo, a balanced combination of several antioxidants may also be needed to exert a significant effect on the prevention of AD but it appears not much has been done on this aspect. Although agents that ameliorate secondary pathologies to AD are under increasing scrutiny²², these treatments would only stagnate the course of disease progression and could not achieve its full eradication. To do so, the community must acknowledge all aspects of disease pathogenesis and establish an unbiased understanding of its progression. Only then can research be adequately focused such that appropriate therapeutic intervention is possible. We here pose the oxidative stress looking glass as our best hope.