Oxidative stress in Alzheimer's disease: Primary villain or physiological by-product?

Abstract

The prevalence of Alzheimer's disease (AD) is increasing rapidly worldwide due to an ageing population and largely ineffective treatments. In AD cognitive decline is due to progressive neuron loss that begins in the medial temporal lobe and spreads through many brain regions. Despite intense research the pathogenesis of the common sporadic form of AD remains largely unknown. The popular amyloid cascade hypothesis suggests that the accumulation of soluble oligomers of beta amyloid peptides (Aβ) initiates a series of events that cause neuronal loss. Among their putative toxic effects, Aβ oligomers are thought to act as pro-oxidants combining with redox-active metals to produce excessive reactive oxygen and nitrogen species. However, to date the experimental therapies that reduce Aβ load in AD have failed to halt cognitive decline. Another hypothesis proposed by the late Mark Smith and colleagues is that oxidative stress, rather than Aβ, precipitates the pathogenesis of AD. That is, Aβ and microtubule-associated protein tau are upregulated to address the redox imbalance in the AD brain. As the disease progresses, excess Aβ and tau oligomerise to further accelerate the disease process. Here, we discuss redox balance in the human brain and how this balance is affected by ageing. We then discuss where oxidative stress is most likely to act in the disease process and the potential for intervention to reduce its effects.

Keywords:

• Alzheimer's disease
• Pathogenesis
• Oxidative stress
• Human post mortem brain tissue

Introduction

Alzheimer's disease (AD) is the most common form of dementia, accounting for 60% of all cases. The prevalence of AD is approximately 1% of the total population in western countries, while the age-prevalence rises rapidly from 5% at 65 years to greater than 25% in individuals over 85 years of age. The worldwide incidence of AD is increasing dramatically with the 36 million current sufferers worldwide predicted to reach 115 million by 2050.¹

The AD brain is characterized by the deposition of insoluble protein as extracellular plaques and intracellular neurofibrillary tangles (NFTs). The former are largely composed of a group of peptides called beta amyloid (Aβ), while the latter are composed of fibrillar and hyperphosphorylated microtubule-associated protein tau (tau). While Aβ and tau deposits are hallmarks of AD and are used for the pathological diagnosis of the disease, dementia results from the regional loss of neurons.

According to the amyloid cascade hypothesis (ACH), oxidative injury is precipitated by the build-up of Aβ oligomers and progresses to tau-mediated neurodegeneration.² It has been proposed that the recent failure of Aβ-modifying treatments in clinical trials reflects that their implementation at the clinical phase of AD is too late to modify disease progression.³ Opponents of the ACH suggest that a factor X precipitates sporadic forms of AD. In particular, the late Mark Smith and his colleagues suggest that Aβ may act initially as an anti-oxidant that is upregulated in response to oxidative stress (OS).⁴ Here we review the literature linking OS and AD and define its likely role in this disease.

Clinical presentation

The clinical diagnosis of AD is based on symptoms and made without the aid of sensitive and specific biomarkers. A diagnosis of ‘probable’ AD is made when there is a worsening memory with deficits in two or more other cognitive domains, such as executive function, attention, language, and visuospatial skills⁵ in the absence of other conditions likely to cause the dementia.

The diagnostic criteria have recently been revised and now incorporate, at least in the research setting, biomarkers of the underlying pathophysiological process.⁶ Clinical biomarkers have been divided into those representing ‘early’ Aβ deposition such as can be demonstrated with the Aβ-binding ligand, ¹¹C-labelled Pittsburgh Compound-B, positivity on positive emission tomography while ‘late’ events include elevated cerebrospinal fluid (CSF) tau protein and regional brain atrophy.

The preclinical period

The average disease duration of AD is 10 years but the prodrome may be up to 20 years.⁷ The latter can be divided into an early ‘Preclinical AD’⁸ and mild cognitive impairment (MCI).⁹ MCI is distinguished from AD in that the individual does not suffer any significant interference in their ability to function at work or in daily activities.⁹ The term ‘preclinical AD’ optimistically infers that ‘early’ biomarkers faithfully reflect the underlying pathophysiological processes.⁸ Indeed, studies suggest that the majority of individuals with MCI have typical AD pathology, but to a lesser degree than those who have dementia.¹⁰

The pathogenesis of AD

Aβ accumulation is regarded as the initiating event in AD, but it is the spread of NFT pathology from the medial temporal lobe to the association and finally primary cortices¹¹ that correlates better with the probability and severity of dementia. In comparison, the regional distribution of Aβ plaques varies markedly between patients.¹¹ Similarly, reactive glia, that associate with plaques and NFTs show better correlations with the extent of NFTs than Aβ burden.¹²

The modern research landscape of AD was established when the Aβ peptide(s) was purified from amyloid-containing AD brain tissue.¹³ The parent amyloid precursor protein (APP) was subsequently isolated and sequenced from a human brain cDNA library and comparative studies suggested it was a neuronal surface receptor.¹⁴ The observation that individuals with trisomy 21 (Down syndrome) develop AD pathology around 40–50 years of age¹⁵ suggested that copy number mutations in APP may account for familial forms of AD. Initially APP missense mutations were found in AD families followed by mutations in two other genes that encode presenilin 1 (PS1) and 2 (PS2) but APP multiplications were also eventually found in 2006.¹⁶

APP is degraded via amyloidogenic or non-amyloidogenic pathways. The majority of non-amyloidogenic processing occurs at the plasma membrane¹⁷ while the majority of amyloidogenic APP metabolism occurs within the endosomal–lysosomal system following internalization, a process closely related to synaptic activity.¹⁸ In the amyloidogenic pathway β- and γ-secretases combine to produce Aβ, a collection of peptides up to 42 amino acids in length.¹⁹ The γ-secretase is a multi-unit enzyme that includes either PS1 or PS2.

Although representing fewer than 5% of total AD cases, the discovery that monogenic forms of AD combined with typical AD pathology in Down syndrome led Hardy and Higgins to suggest the ACH for the pathogenesis of sporadic AD.²⁰ The ACH suggests that fibrillar Aβ in plaques disrupts neuronal calcium homeostasis, precipitating abnormal tau hyperphosphorylation, and NFT formation (Fig. 1). The hypothesis was modified² when soluble oligomeric forms of Aβ, that are also referred to as amyloid-beta-derived diffusible ligands (ADDLs), became the more likely pathogenic species.²¹ The modified hypothesis, which also recognizes the involvement of oxidative injury in the cascade, has stood the test of time although recent failures of Aβ-limiting therapeutics in clinical trials have led some to question an early pivotal role for Aβ in AD.

Figure 1. Oxidative stress and Alzheimer's disease: amyloid cascade hypothesis. Accumulation of extracellular soluble beta amyloid (Aβ) oligomers leads to ionic dysregulation, oxidative stress, neurofibrillary tangle formation and finally the death of neurons. Aβ is largely produced in the trans golgi network (TGN) of neurons following internalisation of the amyloid precursor protein (APP) via the endosomal-lysosomal system. Aβ monomers are then transported out of the neuron via the secretory pathway where they may combine with metal ions, such as Cu²⁺, that precipitate their self-aggregation to oligomers and finally plaques. It is Aβ oligomers that are postulated to disrupt NMDA receptor activity causing an excessive Ca²⁺ ion influx into adjacent neurons. High cytoplasmic Ca²⁺ levels disrupt mitochondria leading to decreased efficiency of the electron transport chain and increased superoxide (O₂^−•) production; O₂^−• is also generated by the membrane bound NADPH oxidases. O₂^−• is converted to H₂O₂ by superoxide dismutases (SOD1 and 2). Reactive microglia are another source of H₂O₂ and this two-electron oxidant readily diffuses and can be converted to the potent hydroxyl radical (HO^•). H₂O₂ is problematic for neurons if enzymes such as catalase in peroxisomes or cytoplasmic glutathione peroxidase do not adequately degrade it. O₂^−• can also combine with nitric oxide (NO) to produce peroxynitrite (ONOO⁻). These reactive oxygen and nitrogen species damage DNA, lipid and proteins resulting in products such as 8-dihydro-2′-guanosine (8OHG), 4-hydroxynonenal (HNE) and 3-nitrotyrosine (3-NT) respectively. All three oxidation products are found in both the extracellular plaques and intraneuronal neurofibrillary tangles that characterize the AD brain. Aβ oligomers may also act as pro-oxidants themselves, upregulate the neuronal form of nitric oxide synthase, or bind and reduce redox active metals allowing them to catalyse the conversion H₂O₂ to HO^• via the Fenton reaction. Extracellular Aβ oligomers are generally thought to precipitate the AD cascade but there is increasing evidence that intraneuronal Aβ accumulation is also an early event in AD and it may accumulate in mitochondria causing direct damage and/or catalyse cytoplasmic redox reactions.

Redox balance and the brain

The brain requires 20% of total blood flow and 25% of the body's glucose utilization. In neurons, energy in the form of adenosine triphosphate (ATP) is produced aerobically through the tricarboxylic acid (TCA) cycle and oxidative phosphorylation (OxPhos) in mitochondria. OxPhos is not totally efficient and up to 2% of electrons are incompletely reduced to yield O₂^−•, instead of H₂O (Fig. 1).

Low-molecular weight anti-oxidants such as glutathione, and metal-containing enzymes such as superoxide dismutase (SOD), catalase (CAT), and glutathione peroxidase (GPX), act to control the redox environment in the brain.²² The first defence against ROS is SOD converting O₂^−• to H₂O₂. Isoforms of SOD include the mitochondrial manganese-SOD (SOD2) and the cytoplasmic copper/zinc SOD (SOD1). CAT and GPX then convert the weak two-electron oxidant H₂O₂ to oxygen and water. GPX in particular deals with H₂O₂ and organic peroxidases by catalyzing the conversion of glutathione (GSH) to its oxidized form, glutathione disulphide (GSSG).

Nitric oxide (NO) is a signalling and neuroprotective molecule in the brain but it can combine with O₂^−• to yield peroxynitrite, a reactive nitrogen species (RNS). Excessive NO competes with the anti-oxidant enzymes for O₂^−• to promote RNS generation.²³

The brain also has relatively high concentrations of redox-active metal ions such as iron (Fe) and copper (Cu) that not only promote ADDL formation but also facilitate the conversion of H₂O₂ to other potent ROS.²⁴ Therefore, despite being well served by overlapping anti-oxidant defence mechanisms, the redox balance of the brain is permanently challenged, a scenario that worsens with age.

The ageing brain

Ageing is the major risk factor for AD but there is also considerable overlap in the pathology and postulated mechanisms for cognitively normal ageing and AD. The pathognomonic entities of AD, namely NFTs and plaques, are both common in the brains of neurologically normal individuals.²⁵ In fact, tau pathology confined to the locus coeruleus and entorhinal cortex is observed in the vast majority of aged brains.^26,27

Importantly, OxPhos not only becomes progressively less efficient with ageing²⁸ but there is a concurrent decrease in anti-oxidant capacity.²⁹ This results in accumulation of ROS that inhibit cellular pathways through damage to lipids, proteins, and DNA.³⁰ Interestingly, treatment of a senescence-accelerated mouse model with the putative anti-oxidant curcumin restored ATP production and mitochondrial membrane potential in brain cells ex vivo³¹. The close link between mitochondrial dysfunction, ageing and AD hints at a possible role for anti-oxidants in modifying AD progression.

Markers of oxidative stress in AD

The determination of OS is achieved by either measuring reactive oxygen and nitrogen species (RONS) directly or by the secondary damage caused to various macromolecules (Fig. 1).

Proteins can also be targeted directly by RONS or non-specifically by damage to individual amino acids. The TCA cycle enzyme, aconitase, is highly sensitive to inactivation by O₂^−• because of its labile iron atom and its proximity to the OxPhos. Therefore, a decrease in brain aconitase activity is a sensitive indicator of excess mitochondrial O₂^−• production. Specifically, O₂^−• inactivates aconitase by displacing Fe²⁺from the Fe-S cluster, which is then available to react with H₂O₂ to generate secondary ROS (e.g. HO^• and ROO^•). Tyrosine is nitrated by peroxynitrite to 3-nitrotyrosine, not only disrupting function but also acting as a marker for excess peroxynitrite.³² Redox sensitive amino acid residues can also be oxidized by ROS to yield carbonyl derivatives and disulphide bonds at cysteine residues that result in structural and functional damage to proteins.³³ Lastly, advanced glycation end products resulting from condensation of reducing sugars with proteins via the Maillard reaction are also often oxidized, further compromising the protein's functional activity.³⁴

Oxidative damage to DNA and RNA produces modified bases such as 8-oxo-2′-deoxyguanosine (8-oxo-dG) and 8-dihydro-2′-guanosine (8OHG), respectively, and both have been reported in cerebral homogenates derived from AD brains.³⁵ Structural alterations in DNA can have functional consequences such as base pair mismatching and changes in DNA repair mechanisms.³⁶ RNA oxidation leads to decreased protein synthesis when damaged transcripts are degraded by RNA surveillance systems or rejected by translation machinery.³⁷

The peroxidation of membranous polyunsaturated fatty acids by ROS yield products such as iso- and neuro-prostanes, acrolein, and the reactive aldehydes, malondialdehyde and 4-hydroxy-2-nonenal (HNE), and these are all found in the AD brain.³⁶ Lipid oxidation products can act directly in the cell or, in the case of HNE, covalently attach to proteins to inhibit their functions. HNE-conjugated proteins, α-enolase and voltage-dependent anion channel 1, are not only seen in AD brain tissue but also Huntington's disease and amyotrophic lateral sclerosis.³⁸ Similarly, increased acrolein and HNE levels have been described in post-mortem brain tissues of patients with MCI³⁹ while CSF isoprostanes have also been described in offspring of female AD patients.⁴⁰ The latter study argues for perturbations in redox balance occurring in individuals with a potential genetic susceptibility to AD.

The mechanism of oxidative stress in AD

In 1994 the Butterfield group demonstrated in vitro that Aβ was a pro-oxidant⁴¹ and that oxidation of the sulphur atom of methionine residue 35 in Aβ_1–42 or shorter fragments was critical to its pro-oxidant and neurotoxic properties.⁴² It was suggested that Cu²⁺converts Aβ to a sulphoperoxy radical.⁴³ As a variation on this theme, Greenough et al.²⁴ consider that it is the binding of the redox-active metals Cu and Fe to Aβ that facilitates their reduction and the consequent pro-oxidant activity of these Aβ-metal ion complexes (Fig. 1).

Soluble Aβ also has been shown to stimulate neuronal nitric oxide synthetase (nNOS) to produce NO, which can then combine with O₂^−• to produce peroxynitrite.⁴⁴ However, nNOS is also activated by NMDA receptor-mediated calcium influx⁴⁵ so NO-mediated oxidative damage is not necessarily due to the direct effects of Aβ.

In addition, Aβ is also thought to damage mitochondria, either indirectly via excessive intraneuronal Ca²⁺ levels or directly by acting as a mitochondrial toxin.⁴² A proteomic study utilizing the triple transgenic (APP/PS1/tau) mouse showed deficits in OxPhos with an Aβ-dependent inhibition of the electron transport chain complex IV but a tau-dependent inhibition of complex I.⁴⁶Aβ is known to accumulate intraneuronally⁴⁷ and it may be taken up by mitochondria, although it seems more likely that putative interactions involve binding to mitochondrial surface proteins such as amyloid-binding alcohol dehydrogenase.⁴⁸

Metals and cerebral oxidative stress

Several transition metals play key roles in brain metabolism, including Fe, Cu, Zinc (Zn), and manganese, but in excess they may be involved in chemical reactions that form ROS including H₂O₂, O₂^−• and HO^•. Studies of post-mortem brain tissue showed that Cu⁴⁹ and Fe⁵⁰ were both elevated in the AD brain. However, more recently this metal hypothesis of AD was questioned when a meta-analysis suggested that there were no global increases in these metals; Zn and Fe concentrations remained unchanged while Cu was actually depleted in the AD brain.⁵¹ Proponents of the metal hypothesis of AD countered by suggesting that there were focal increases in metal content around plaques and NFTs.²⁴

It is thought that haemoxygenase (HO) provides an important source of redox-active Fe²⁺ that must be sequestered in a ‘redox-inactive’ form.⁵² As part of normal homeostasis, constitutively expressed HO-2 regulates pro-oxidant haem by degrading it into free Fe, carbon monoxide and biliverdin/bilirubin. However, when confronted with increased levels of haem, inducible HO-1 is produced and this often occurs in parallel with increases in the intracellular iron-binding protein ferritin, which binds the released Fe.⁵³ Potentially, free haem can cause oxidation of other proteins and lipids.⁵⁴ Similarly, unbound Fe²⁺ can react with H₂O₂ to yield highly reactive OH^• (Fenton's reaction) or interact with O₂^−• and H₂O₂ to form OH^• via the Haber–Weiss reaction (Fig. 1). Interestingly, HO-1 localizes with NFTs in the AD brain.⁵⁵ Accumulating Fe²⁺ and Cu²⁺ ions can bind to histidyl and sulphydryl residues, respectively, within Aβ.²⁴ These Aβ-metal complexes are also able to generate ROS via Haber–Weiss and Fenton reactions.

One limitation in AD research is the lack of understanding of the physiological role of APP. We do know that levels of Zn, Cu, and Fe can all affect APP metabolism.⁵⁶ The APP gene contains a Fe-responsive element in its 5′ untranslated region⁵⁷ and Fe regulatory proteins are able to control its translation. APP also contains a Cu-binding site and can reciprocally modulate the levels of Cu²⁺in the brain.⁵⁸ Both APP and Aβ may be involved in maintaining metal homeostasis in the brain, particularly in the face of increased OS.⁵⁹

Anti-oxidant enzymes

The data from post-mortem AD brains are inconsistent in terms of the levels of anti-oxidant enzymes including SOD1 or total SOD activity, with increases,⁶⁰ decreases,⁶¹ and no differences⁶² found compared to controls. GPX levels have been reported as decreased or showing no difference⁶³ with only one report of increased GPX in the AD brain.⁶⁴ CAT levels have also been reported as either decreased⁶⁵ or not different.⁶⁶

Karelson et al. showed that SOD activity was highly correlated with plaques but not NFTs⁶⁰ but Maeda et al.⁶⁷ suggested that this was mainly SOD2 rather than SOD1. Like APP, SOD1 is located on chromosome 21 and SOD1 is overexpressed in Down syndrome brains.⁶⁸ It seems paradoxical that the anti-oxidant enzyme SOD could be linked with neurodegeneration but the generation of H₂O₂ is potentially toxic for the cell if enzymes such as CAT or GPX are compromised. Consistent with this scenario, Down syndrome neurons can be rescued by CAT expression⁶⁹ while neurotoxic effects of Aβ on hippocampal neurons can be ameliorated by the combination of CAT and SOD expression.⁷⁰ Similarly, it was recently shown that a mitochondrial-targeted CAT/APP double transgenic mouse displayed reduced APP processing and Aβ load with enhanced Aβ-degrading enzyme activity compared to the APP single transgenic.⁷¹

An alternative hypothesis

Results from the CAT/APP mouse model suggest that ROS, and particularly H₂O₂, can induce the upregulation of Aβ rather than the other way around. This view was held by the late Mark Smith and has been summarized in a recent review article.⁴ Briefly, Cuajungco et al.⁷² consider that neurons respond to OS by increasing Aβ production, with Aβ acting as a SOD mimic. Furthermore, Aβ, by binding metal ions, may prevent them from participating in redox cycling reactions.⁷³

The group also maintain that tau is upregulated in response to OS, based firstly on neurons with NFTs in AD brains having less 8-OHG and 3-NT⁷⁴ with similar findings in Down syndrome⁷⁵ and monogenic AD patients.⁷⁶ Second, the phospho-protein tau and other structural proteins act as physiological ‘sponges’ for oxidation products such as toxic aldehydes.⁷⁷ Smith et al.⁷⁸ suggest that plaques and NFTs in elderly controls are likely to reflect increased levels of OS associated with the aged brain and similarly question whether the same metal-laden entities are the source of OS in AD. In contrast, they favour the idea that intraneuronal Aβ is more important and accumulates as a cellular response to OS. Dysfunctional mitochondria are the likely source of ROS because HO^•, which diffuses over nanometre distances only, must be generated in the cytosol in intimate proximity to RNA to yield 8OHG (Fig. 1).⁷⁵ In addition, 8OHG levels correlate with the onset of cognitive impairment in the prodromal stage of AD and are greater than that seen in normal ageing.⁷⁹ As excessive Aβ and tau are neurotoxic, their initial upregulation is likely to provide only transient relief. Such a scenario would still be consistent with the genetic evidence in AD, considering that most monogenic forms of AD manifest by middle age.⁸⁰

Testing the oxidative stress hypothesis

Presently the only therapeutics approved for use in AD are the acetylcholinesterase inhibitors and memantine, a NMDA-receptor antagonist. While offering transient improvements in cognition these therapies do not slow disease progression and disease-modifying treatments are sorely needed. Direct evidence for the OS/free radical hypothesis of AD would include a demonstrable ability of anti-oxidants to modify disease course. In this respect the overall results have been disappointing.⁸¹ Proponents suggest that the variety of trial designs and anti-oxidants used make meta-analyses difficult and that it is still too early to discount a role for anti-oxidants in AD treatment. For example, a trial of vitamin C in an APP double mutant mouse yielded a subtle decrease in Aβ soluble oligomers and protein carbonylation, increased glutathione levels and improved the behavioural phenotype with no amyloid deposition or reactive gliosis.⁸²

A recent trial looked at the anti-oxidant combination of vitamins E, C and α-lipoic acid or coenzyme Q in AD patients.⁸³ This trial was novel in that it assayed biomarkers of OS in the CSF. Although there was no effect of treatment on CSF biomarkers related to amyloid or tau pathology, F₂-isoprostane levels decreased in subjects receiving vitamins E, C, and α-lipoic acid, although not coenzyme Q. Unfortunately, F₂-isoprostane concentration also correlated with decreasing cognitive performance on a Mini-Mental State Examination (MMSE). Moreover, a Cochrane Database Systematic Review for vitamin E⁸⁴ found no convincing evidence of benefit in treating AD or MCI. Possibly, vitamin E may only inhibit lipid peroxidation with no effects on the oxidative damage to proteins and nucleic acids.³⁴ A review of vitamin C supplementation in AD suggests that avoiding vitamin C deficiency via a balanced diet is more important than supplementation in AD prevention.⁸⁵

Given the lack of clinical success of anti-oxidant supplements, research has refocussed on limiting the production of ROS such as modulating NAPDH oxidase activity.⁸⁶ Recent studies have also suggested that selenium, the metal co-factor for GPX, might be deficient in AD.⁸⁷ One report in an AD mouse model suggested that selenium supplementation can attenuate tau pathology, but it appeared to stabilize a complex formed between tau and its major protein phosphatase, PP2A,⁸⁸ rather than augmenting GPX activity.

One area where some clinical success has been reported is the trials with metal chelators. As noted Cu, Fe, and Zn can bind to Aβ and precipitate aggregation while Cu or Fe can form complexes with Aβ that enhance H₂O₂ production. The first metal chelator trialled, clioquinol, had initial encouraging results in phase II studies;⁸⁹ however, no improvement in cognitive performance could be established and the phase II trial was abandoned.⁹⁰ A clioquinol-like drug, PBT2 was subsequently developed and it performed well in a phase IIa trial with improvements in cognition on some neuropsychological tests.⁹¹ However, a Cochrane Review in 2012 noted that no improvements in cognition were seen with either the MMSE or ADAS-Cog scale.⁹² The current PBT2 phase II trial is due for completion in December 2013.

Conclusion

The working model for the pathogenesis of sporadic AD is the ACH, which states that the accumulation of soluble Aβ oligomers has various deleterious effects on neurons, including oxidative injury, that eventually result in tau-mediated death. An alternative (free radical) hypothesis suggests that OS precedes Aβ oligomerization and that Aβ and tau are upregulated because they can act as anti-oxidants. The latter hypothesis is certainly not as popular but is supported by reasonable evidence. However, it is also possible over the long preclinical and clinical phases of AD that both hypotheses could be operative with Aβ and OS combining to cause cellular damage or that Aβ and OS alternate in their pathogenic roles as the brain attempts to compensate for the disease process.

Understanding the different stages of AD is relevant to treatment regimens, likely preclinical markers and outcome. At present diagnoses are made on clinical signs that result from neuronal loss. The mechanisms responsible for cell loss at the clinical stage of AD might be quite different to those operating in early degenerating neurons. The continuing failure of the very promising Aβ-modifying therapies may be related to these therapies being administered too late, and/or other factors being the precipitating driver of AD pathogenesis. If that driver is OS, then there is probably less chance that anti-oxidants will be therapeutically beneficial if only used in clinical cases as the available data suggest.

In reality the predictable spread of AD pathology across the brain means that neurons of different regions will be at different stages of degeneration. Therefore, therapy is likely to require a suite of therapeutics that target ‘preclinical’ and ‘clinical’ pathogenic mechanisms. One caveat here is that a protective host response in one region could have reached deleterious proportions elsewhere. It is also possible that we are yet to sufficiently understand the physiological roles of APP, Aβ, and ROS to predict the chain of events in AD. Clinical trials based on our most likely therapeutic targets are an essential avenue to prevent an AD epidemic, but similar enthusiasm is required to tease apart the pathogenesis of this most complex of diseases.

Acknowledgements

A Judith Jane Mason and Harold Stannett Williams Memorial Foundation Medical and Scientific Research grant supported this work. The authors thank Mr Stephen Kum Jew and Mr Claude Dennis for their assistance with immunohistochemistry and Professor Jillian Kril for her useful comments and suggestions.