8.4. ALZHEIMER'S DISEASE: MOLECULAR CONCEPTS AND THERAPEUTIC TARGETS
Alzheimer's disease (AD) is characterized by progressive memory deficits, cognitive impairments and personality changes that are caused by progressive synaptic dysfunction, and much later, death of neurons in the neocortex, limbic system and subcortical regions in the brain. AD is now the most common cause of dementia in the elderly. Worldwide approximately 15 million people are affected. Because of longer lifespans, the number of affected persons has been estimated to triple over the next few decades (115; 116; 117). The expected individual problems and enormous costs for society have increased public awareness of dementia (118). The characteristic lesions of AD, the neuritic plaques, are mainly located in the association cortex, limbic system and basal ganglia. Amyloid peptide (β) is the major constituent of these plaques (approximately 90%) (119).
In addition, AD is pathologically characterized by neurofibrillary tangles (paired helical filaments containing hyperphosphorylated tau) within the cell bodies or dendrites of neurons. Tau is a microtubule-associated protein that binds to tubulin and participates in microtubular assembly and stability of axons. Although most data support a central pathogenetic role of the amyloid rather than the neurofibrillary tangles (e.g., the genetic mutations associated with Aβ ?oversecretion that give rise to AD or the toxicity of amyloid), skeptics argue that amyloid plaques correlate only poorly with the degree of dementia. However, more recent studies have shown that clinical markers for AD correlated well with "soluble forms" of amyloid (120).
8.4.1. EXTRACELLULLAR RELEASE OF Aβ
Aβ is the cleavage product of beta amyloid precursor protein (APP), a type I transmembrane glycoprotein, that contains a large ectopic N-terminal domain, a transmembrane domain, and a small cytoplasmatic C-terminal tail. There are several isoforms differing in the number of amino acids due to alternative splicing of the transcript of the single APP gene on chromosome 21. The pathogenetically relevant Aβ-domain within APP is located partly in the transmembrane domain and partly in the ectodomain. Different APP cleaving proteases, generally termed secretases, determine the relative proportion of amyloidogenic and non-amyloidogenic APP catabolites (121; Fig. 8.3). Thus, proteolytic action of the β-secretase yields the non-amyloidogenic N-terminal domain of APP (soluble APP) and a small 10 kDa C-terminal fragment. Importantly, this β-secretase cut occurs within the Aβ domain, thus precluding the generation of the amyloidogenic Aβ.
In contrast, the pathogenetically relevant Aβ is formed by the cleavage of APP at its NH2-terminus by a γ-secretase, leaving a membrane bound fragment, C99. This fragment can then be further cleaved by a γ-secretase to release the amyloidogenic Aβ in the extracellular space. Vassar et al. (122) recently reported a candidate for the γ-secretase that maps to the long arm of chromosome 11. This "β-site APP cleaving enzyme (BACE)" exhibits all the properties hypothesized for the γ-secretase. Recently, a less potent homologue of this enzyme (BACE-2) has been identified on chromosome 21. In contrast, the γ-secretase still remains to be discovered.
8.4.2. INTRACELLULAR Aβ ACCUMULATION
APP, after reaching the cellular surface, can be reinternalized and transported to the endosomes (123). In these organelles, β-cleavage can occur and, after eventual further cleavage by the γ-secretase at the cell membrane on its way back to the cell surfaces, Aβ can be formed and extracellularly released. However, it has been shown that Aβ also accumulates intracellularly (124, 125). Since presenilin, which is located in the endoplasmatic reticulum, can bind to immature APP (126), it is possible that APP can be intracellularly trapped, resulting in such increased intracellular Aβ ?formation, thereby injuring the neuron. Similar adverse effects may result from the interaction of Aβ ?with an intracellular polypeptide called endoplasmatic-reticulum-associated binding protein (ERAB), which has been shown to be overexpressed in neurons of AD patients (127).
8.4.3. GENES RESPONSIBLE FOR AD
Mutations in several genes cause AD. Mutations in the APP gene, which is localized on chromosome 21, account for 5-20% of the cases with early-onset familial AD (128). Most of these mutations affect the site of action of the different secretases, thereby increasing either the rate of Aβ release or the peptide length that determines the tendency to form fibrils (129). Presenilins (PS) are transmembrane proteins with still uncertain functions. Whereas mutations in the PS-1 gene (130, 131) are the most common cause of early onset familial AD (over 50% of cases), mutations of the PS-2 gene (132) are rather rare (133). Importantly, evidence is growing that the PS-1 and PS-2 variants cause AD via the APP/Aβ pathogenetic pathway, as they also modulate proteolytic activity of the secretases (134, 135).
Consistent with this contention are the increased numbers of amyloid plaques in patients with PS mutations (136) and - on the other hand — their reduced appearance in knock out animals for these genes (137). The evidence that sporadic AD and these three autosomal dominant inherited familial cases lead to the same abnormal APP processing and amyloid pathology (Fig. 8.3) is considered as a major argument for the amyloid theory of AD.
Fig. 8.3. Molecular concepts of AD. Genetic and external factors modulate neuronal secretion of soluble Aβ in the extracellular space of the CNS. Different secretases determine the relative proportion of non-amyloidogenic (sAPP) and amyloidogenic (Aβ) APP catabolites. "Plaqueassociated"- proteins in the extracellular matrix promote the conversion of the soluble Aβ to the non-soluble, beta-sheeted fibrils. These fibrils may injure neurons either directly or indirectly via an inflammatory effector phase.
In contrast to the mutations responsible for earlyonset AD, allelic polymorphism of the gene encoding the cholesterol-transport protein, apolipoprotein E (ApoE; 138), is a susceptibility factor for late-onset AD (139). Presence of the ε4 allele of ApoE increases the risk and lowers the age of onset in AD in a dose-related manner (140). Thus, the ε4 allele decreases age of disease onset by 3-6 years while the ε3 or ε2 alleles appear to be neutral or even protective (141). The mechanisms, whereby ApoE polymorphism influences risk for AD are still unclear. It is possible that these ApoE isoforms differentially determine the lipid composition of the cell membrane that is crucial in APP catabolism or that they differ in their ability to promote formation of toxic Aβ-fibrils (Fig. 1). Finally, a further putative mechanism is the differential modulation of the repair of neuronal injury (e.g., neuronal sprouting) (142).
Apart from these genes, the contribution of several further genes to AD is currently extensively studied, e.g., the genes for alpha2 macroglobulin (143) or low-density lipoprotein receptor-related protein (LRP; 144).
8.4.4. FORMATION OF TOXIC Aβ FIBRILS
The classical neuritic plaques consist of fibrillar Aβ aggregates. The aggregation of the physiologically secreted soluble Aβ to large Aβ fibrils is currently considered a crucial event in the pathogenesis of AD (Fig. 8.3). In contrast to the soluble, non-fibrillar Aβ , fibrillar Aβ is associated with neurotoxicity (145; 146), i.e., Aβ oligomers and protofibrils have been implicated in direct toxicity for neuronal cells (147; 148). However, neurotoxicity of Aβ fibrils could also be an indirect effect, since fibrillar but not non-fibrillar Aβ has been shown to trigger glial cells to produce toxic mediators in vitro (149) and in vivo (150). Several glycoproteins such as α1-antichymotrypsin, ApoE, serum amyloid A, transthyretin or complement factors bind to Aβ and can be found in AD plaques.
Interestingly, many of these "plaque-associated" proteins function as "pathological chaperones" promoting the fibrillar conformation of Aβ (Fig. 8.3; 151; 152). Also glycosaminoglycans, which are normal constituents of the extracellular brain matrix, accumulate in diffuse and senile plaques and bind to Aβ(153-156) suggesting a role in plaque formation (157; 158; 159; 160). A recent study showed that injection of fibrillar Aβ is not toxic to the young adult rhesus brain whereas it caused profound neuronal loss, tau phosphorylation and microglial proliferation in the aged rhesus brain (161). This suggests that, complementary to Aβ -secretion, changes of the composition of the local cerebral matrix may contribute to AD.
8.4.5. INFLAMMATORY EFFECTOR PHASE
Further downstream in the cascade of events leading to AD, a chronic non-specific inflammatory effector phase may contribute to neuronal injury. This is supported by the detection of clusters of activated microglia and macrophages at sites of senile plaques (162; 163) and the ability of Aβ to activate such mononuclear phagocytes in vitro (149) as well as in vivo (150). Chronic activation of these inflammatory cells could contribute to neurodegeneration in AD by release of neurotoxic products such as reactive oxygen and nitrogen derivatives, proteolytic enzymes and inflammatory cytokines (149; 164). Indeed, theses factors have been detected in AD lesions (163; 165). Because of its high lipid content, the elevated oxygen consumption, and the relatively poor antioxidative defense mechanisms, the brain is especially vulnerable to oxidative stress.
Apart from their direct toxicity for neurons, products of activated glial or leukocytic cells (i.e., inflammatory cytokines) could also adversely modulate APP metabolism, e.g., resulting in increased Aβ secretion (166). Finally, inflammatory cytokines such as IL-1β or TNF-α have been reported to enhance synthesis of the above described "plaque-associated" proteins by astrocytes, microglia or neurons (167), thereby further enhancing fibril formation.
8.4.6. CLOSING ON AD
As summarized in Fig. 1, AD is the result of a complex cascade of events that, over years and decades, leads to clinical symptoms (168). Therefore, even weakly effective treatments could considerably delay the progression or symptoms of AD when administered in a long-term manner and started in early disease or -ideally - in non-demented high-risk subjects. However, a precondition for identification of such high-risk patients is the availability of surrogate markers for AD; and these are still lacking. However, a relatively uniform CSF pattern of proteins (high tau-, initially increased and subsequently decreased Aβ-amyloid peptide profile in AD) has recently been reported (169). Thus, Aβ is not only the key molecule in the pathogenesis of AD but in the future may be clinically useful as a diagnostic tool.