Review paper A critical analysis of the ‘amyloid cascade hypothesis’ Richard A. Armstrong Vision Sciences, Aston University, Birmingham, UK Folia Neuropathol 2014; 52 (3): 211-225 DOI: 10.5114/fn.2014.45562 Abstract The ‘amyloid cascade hypothesis’ (ACH) is the most influential model of the pathogenesis of Alzheimer’s disease (AD). The hypothesis proposes that the deposition of β-amyloid (Aβ) is the initial pathological event in AD, leading to the formation of extracellular senile plaques (SP), tau-immunoreactive neurofibrillary tangles (NFT), neuronal loss, and ultimately, clinical dementia. Ever since the formulation of the ACH, however, there have been questions regarding whether it completely describes AD pathogenesis. This review critically examines various aspects of the ACH includ- ing its origin and development, the role of amyloid precursor protein (APP), whether SP and NFT are related to the development of clinical dementia, whether Aβ and tau are ‘reactive’ proteins, and whether there is a pathogenic relationship between SP and NFT. The results of transgenic experiments and treatments for AD designed on the basis of the ACH are also reviewed. It was concluded: (1) Aβ and tau could be the products rather than the cause of neuro- degeneration in AD, (2) it is doubtful whether there is a direct causal link between Aβ and tau, and (3) SP and NFT may not be directly related to the development of dementia, (4) transgenic models involving APP alone do not completely replicate AD pathology, and (5) treatments based on the ACH have been unsuccessful. Hence, a modification of the ACH is proposed which may provide a more complete explanation of the pathogenesis of AD. Key words: Alzheimer’s disease, amyloid cascade hypothesis, β-amyloid, neurofibrillary tangles, disease pathogenesis. Introduction The ‘amyloid cascade hypothesis’ (ACH) [57] is a direct relationship between the development of the most influential model of the pathology of Alzhei- Aβ, in the form of senile plaques (SP), and neuro- mer’s disease (AD) proposed over the last 25 years. fibrillary tangles (NFT) and second, between these As a result, numerous studies of AD pathogenesis lesions and clinical dementia. Hence, deposition of have been carried out [6,8] and potential treatments Aβ is the initial pathological event in AD, leading to proposed and tested based on the ACH [69,113]. the formation of SP, tau-immunoreactive NFT, neuro- It was the discovery of β-amyloid (Aβ), the most nal loss, and ultimately clinical dementia. The ACH important molecular constituent of senile plaques is often regarded as the ‘conclusive model’ of the (SP) [49], and mutations of the amyloid precursor aetiology of early-onset FAD, and sporadic AD (SAD), protein (APP) gene linked to early-onset familial a more complex disorder caused by a variety of fac- AD (FAD) [27,50], which led to the original formula- tors, but resulting from essentially the same patho- tion of the hypothesis [57]. The ACH proposed first, logical cascade [42,127]. Communicating author: Dr Richard A. Armstrong, Vision Sciences, Aston University, Birmingham B4 7ET, UK, phone: 0121-204-4102, fax: 0121-204-4048, e-mail: [email protected] Folia Neuropathologica 2014; 52/3 211 Richard A. Armstrong Since its publication there have been observa- Key observations in the development tions difficult to reconcile with the ACH and con- of the amyloid cascade hypothesis cerns whether it completely describes AD pathogen- The original formulation of the ACH as proposed esis [11,90]. First, transgenic mice, in which genes by Hardy and Higgins [57] is shown in Figure 1. A num- overexpress amyloid precursor protein (APP), do not ber of key observations resulted in the formulation reproduce the predicted cascade [39,90]. Second, SP and subsequent development of the hypothesis: and NFT are separated in brain both spatially [11] 1. Aβ was discovered as the most important mo- and temporally [39,82] questioning the pathogenic lecular constituent of SP [49], emphasising the impor- link between them. Third, Aβ and tau could be the tance of amyloid peptides in AD. reactive products of neurodegeneration rather than 2. Altered proteolytic processing of APP and its their cause, arising as a consequence of oxidative accumulation were shown to be early events in AD stress [126]. Fourth, study of various neuronal injury and were followed by microglial activation, astrocy- biomarkers does not support a central pathological tosis, and dystrophic neurites (DN) [105]. role for Aβ, especially in late-onset SAD [28]. These 3. Mutations of APP gene [27,50] were linked to observations suggest a more complex relationship early-onset FAD, the presence of point mutations in between Aβ, tau, and AD pathogenesis than sug- APP evidence that amyloid deposition was the caus- gested by the ACH. In addition, concentration on the ative factor [56]. Three main groups of APP mutations ACH may have been to the detriment to the study have been identified, i.e., those which act at: (a) the of other possible aetiologies including: (1) pertur- beta APP cleaving enzyme 1 (βACE1) site, (b) the gam- bation of vesicular trafficking, (2) disruption of the ma APP cleaving enzyme site, and (c) the mid-domain cytoskeletal network, and (3) disturbance of mem- of the Aβ region. brane cholesterol [37]. These views are supported by 4. Formation of Aβ peptides was directly linked a critique of the ACH by Teplow [133] who argues to APP processing with two events being necessary that ‘insufficient rigour’ has been applied to studies to generate Aβ [116]: (a) cleavage by βACE1 which of Aβ in AD based on the ACH and ‘confusion not cleaves APP on the amino side of Aβ resulting in clarity’ has been the result. a large secreted derivative and an Aβ membrane- This review critically examines various aspects associated C-terminal derivative (CTFβ) and (b) cleav- of the ACH including its origin and development, age by g-secretase which cleaves CTFβ to release the role of amyloid precursor protein (APP), wheth- the Aβ peptide [63]. A variety of Aβ peptides can er SP and NFT are related to the development of be formed as a result of cleavage of APP [54], the clinical dementia, whether Aβ and tau are ‘reactive’ most common being Aβ42 [131], found largely in dis- peptides, and whether there is a pathogenic rela- crete Aβ deposits [139], and the more soluble Aβ40, tionship between SP and NFT. The results of trans- also found in association with blood vessels [86], genic experiments and treatments for AD designed and which may develop later in the disease [35]. In on the basis of the ACH are also reviewed. Finally, addition, Aβ43 may be formed and has a particular- a modified scheme based on the ACH is proposed ly potent amyloidogenicity [115]. Mutations of APP which may provide a more complete description of within the Aβ coding region can also result in the the pathogenesis of all forms of AD. deposition of Aβ38 in vessel walls, especially in those Gene mutation (APP, PSEN1/2), Genetic risk factors Toxic Aβ Neuronal Clinical (Apo E) NFT peptides loss dementia Environmental factors (head injury, diet) Aβ – β-amyloid, Apo E – apolipoprotein E, APP – amyloid precursor protein, PSEN1/2 – presenilin genes 1 and 2, NFT – neurofibrillary tangles Fig. 1. The amyloid cascade hypothesis (ACH) in its original form. 212 Folia Neuropathologica 2014; 52/3 Aging and dementia cases with extensive cerebral amyloid angiopathy 8. In Down’s syndrome (DS), which replicates many (CAA) [88]. Early toxic soluble oligomers could also of the features of AD pathology [5,9], water soluble be involved, which vary with a type of mutant, thus peptides ending in residue 42 precede the formation providing a genetic basis for variations in pathogen- of SP and also increase with age [114]. esis observed among FAD cases [46,48]. 9. Aβ peptides are toxic and can induce cell death 5. Subsequently, the most common form of ear- depending on cell type [55], Aβ toxicity being medi- ly-onset FAD was linked to mutations of presenilin ated by the necrotic rather than the apoptotic path- (PSEN) genes PSEN1 [123] and PSEN2 [78]. Full length way [20,76]. PSEN protein is composed of nine trans-membrane 10. Aβ may induce the phosphorylation of tau by: domains located on the endoplasmic reticulum mem- (a) directly interacting with a domain of APP (714- brane. Endoproteolytic cleavage of PSEN and assem- 723) [47,125], (b) by inducing tau protein kinase I bly into g-secretase complex is followed by transport and subsequently, tau proteins recognised by the to the cell surface, thus potentially influencing APP antibody Alz-50 [132], (c) as a result of synergisms processing [61]. Hence, mutant PSEN1 could enhance between Aβ and tau [100,101], or (d) by directly alter- 42-specific-g-secretase cleavage of normal APP re- ing the phosphorylated state of tau [23]. sulting in increased deposition of amyloid-forming 11. Senile plaques and NFT acquire several ‘sec- species [129]. Alternatively, deficiency of PSEN1 may ondary’ constituents (Table I) which may be involved inhibit the normal cleavage of APP [34,141]. PSEN1 in the ma turation of Aβ deposits into SP [7] includ- also appears to alter the ratio of Aβ species, the ratio ing silicon and aluminium [82], acute-phase proteins of Aβ40/42 being lower and the ratio of Aβ42/40 higher such as a-antichymotrypsin and a2-macroglobulin in PSEN1 cases compared with SAD [60]. [40,81,136] and their mediator interleukin-6 (IL-6) 6. Studies of FAD cases caused by APP717 (va line– [122], intercellular adhesion molecules such as cell isoleucine) mutation showed them to have signifi- adhesion molecule 1 (CAM1) [40], apolipoprotein E cant numbers of tau-immunoreactive NFT support- (Apo E) which is present in the earliest stages of ing a link between APP and degeneration of the SP formation [145], apolipoprotein D (Apo D) [36], cytoskeleton [75].