Advertisement
Apoptotic mechanisms in Alzheimer neurofibrillary degeneration: cause or effect?
Dennis W. Dickson
Departments of Pathology and Neuroscience, Mayo Clinic College of Medicine, Jacksonville, Florida, USA.
Address correspondence to: Dennis W. Dickson, Department of Pathology (Neuropathology), Mayo Clinic, 4500 San Pablo Road, Jacksonville, Florida 32224, USA. Phone: (904) 953-7137; Fax: (904) 953-7117; E-mail: dickson.dennis@mayo.edu.
Find articles by Dickson, D. in: JCI | PubMed | Google Scholar
Published July 1, 2004 - More info
J Clin Invest. 2004;114(1):23–27. https://doi.org/10.1172/JCI22317.
© 2004 The American Society for Clinical Investigation
Alzheimer disease (AD), the most common cause of dementia in the elderly, is associated with senile plaques and neurofibrillary tangles (NFTs), but the relationship between these two neuropathologic lesions has been difficult to discover. Senile plaques are heterogenous lesions composed of extracellular amyloid β protein (Aβ), dystrophic neuronal processes, and reactive glia (1), while NFTs are intracellular lesions composed of filamentous aggregates of the microtubule-associated protein tau (2). Genetic factors have implicated Aβ in the pathogenesis of AD since mutations are found in the Aβ precursor (APP) as well as in enzymes involved in the production of Aβ (reviewed in ref. 3). Moreover, genes implicated in AD by linkage studies encode proteins that degrade Aβ, such as insulin-degrading enzyme. The major genetic risk factor for late-onset AD, apolipoprotein E, promotes Aβ aggregation and colocalizes with Aβ in senile plaques.
The conundrum that has plagued research on AD is that clinicopathologic studies have not shown strong correlations between cognitive impairment and Aβ. In particular, the degree of cognitive impairment in AD is not as closely tied to the amount of amyloid deposited in the brain or the number of senile plaques as it is to the amount of abnormal tau protein in the brain and the density and distribution of NFTs (4, 5). This is not entirely surprising since NFTs are composed of proteins that are part of the neuronal cytoskeleton, which supports vital structural and dynamic neuronal functions. Amyloid, on the other hand, is a relatively innocuous extracellular deposit. It has been suggested that cytoskeletal disruption may act as the proximate cause of progressive synaptic and neuronal loss by interfering with axoplasmic and dendritic transport, which starves the cell of trophic support (6). This dysfunction and eventual death of neurons manifests clinically as cognitive impairment.
A major challenge of the amyloid cascade hypothesis for AD (Figure 1), which posits that amyloid formation leads to neuronal loss and dementia (7), is determining the link between Aβ, the protein most clearly linked to the cause of AD, and tau, the protein that is most clearly associated with clinical manifestations of AD. The studies by Rissman and coworkers in this issue of the JCI (8) suggest that apoptotic mechanisms may be the missing link.
Figure 1Proteolytic processes contribute to the amyloid cascade hypothesis. Aβ is generated from APP by concerted proteolysis by β-secretase, which generates carboxyl-terminal fragments (CTFs) of APP, and then by γ-secretase. The Aβ forms aggregates in the extracellular compartment as senile plaques through a process that depends on proteoglycans and apolipoproteins. The extracellular Aβ oligomers may activate caspases through activation of cell surface death receptors. Alternatively, intracellular Aβ may activate caspases through a process that involves ER stress or mitochondrial stress. One of the consequences of caspase activation is cleavage of tau, which favors conformational changes characteristic of paired helical filaments (PHF-tau). Progressive accumulation of tau leads to cytoskeletal disruption (inset), failure of axoplasmic and dendritic transport, and subsequent loss of trophic support that culminates in neuronal death. The extracellular amyloid deposits in senile plaques also trigger reactive glial changes and neuroinflammation that can also contribute to neuronal loss through production of reactive oxygen species (ROS), NO, and proinflammatory cytokines such as TNF-α and IL-1β. sAPPβ, secreted APPβ.
-
Apoptosis in AD
Apoptosis has been the focus of intense research in the last several decades as a means of controlling cell populations in normal development and inflammation through programmed cell death. Failure to control cell numbers through apoptosis is common in cancer, while excessive apoptosis is viewed to play a role in a number of neurologic disorders in addition to AD, including stroke and Parkinson disease (PD) (9).
Enthusiasm for apoptosis, however, as a mechanism for neuronal death in AD has been tempered in recent years. The initial evidence for apoptosis in AD came from cell culture experiments that were not always physiologically relevant, including exposure of cells to very high concentrations of Aβ or to Aβ peptides that do not exist in nature. Evidence for frank cellular apoptosis in AD is controversial, but there is growing recognition that apoptotic mechanisms may play a role in disease pathogenesis in the absence of overt apoptosis (10). Apoptosis is an attractive mechanism for neuronal death in neurodegenerative diseases for several reasons. Neuronal death in degenerative diseases is selective at the individual cell level and not associated with inflammation. In AD, neuronal loss is prominent in the cerebral cortex and the limbic lobe, while different neuronal populations are vulnerable in other neurodegenerative diseases. For example, in PD, brainstem monoaminergic neurons are affected. Cellular death via apoptosis is not associated with disruption of the cell membrane, and clearance of cellular debris is by facultative tissue phagocytes rather than professional macrophages of the reticuloendothelial system. Apoptosis also has a number of characteristic morphologic hallmarks, such as nuclear condensation and fragmentation, which have been very difficult to identify in AD brains.
Another feature of apoptosis is intranucleosomal cleavage that produces DNA nicks that can be labeled by specific techniques, such as TUNEL. Initial reports of extensive neuronal TUNEL labeling in AD have been difficult to confirm, and much of the labeling appears to be related to damage to DNA that occurs as a postmortem artifact (11). Another approach to studying apoptosis is to examine enzymes that are involved in mediating programmed cell death. About a dozen such proteases have been identified; they share the property of being cysteine aspartyl proteases and are referred to as caspases. Caspases participate in apoptosis through initiation of intracellular cascades and in executing the final outcome, including proteolytic cleavage of cytoskeletal proteins and proteins of the nuclear scaffold. Both membrane (e.g., spectrin) and cytosolic (e.g., intermediate filament) cytoskeletal proteins are targets of caspase cleavage. Apoptosis is initiated through intracellular mechanisms that often involve alterations in mitochondria or endoplasmic reticulum and by signaling through cell membrane death receptors — the so-called intrinsic and extrinsic apoptotic pathways (12) (Figure 2).
Figure 2The interface of the two major molecules implicated in AD pathogenesis with molecular mechanisms of apoptosis. The two major pathways to cellular apoptosis are intrinsic and extrinsic. The extrinsic pathway involves signaling through cell surface death receptors, such as the TNF receptor, which are regulated by decoy receptors and Fas-associated death domain–like interleukin-1β–converting enzyme inhibitory proteins (FLIPs). Direct binding of Aβ or Aβ oligomers to death receptors remains to be shown, but the pattern of activation of downstream caspases (e.g., caspases 2 and 8) supports involvement of the extrinsic pathway in Aβ-mediated apoptotic processes. Alternatively, intracellular Aβ produced in the ER may lead to ER stress, or binding of Aβ to a mitochondrial alcohol dehydrogenase may lead to mitochondrial stress. Both entries into the intrinsic pathway may activate downstream apoptotic mechanisms. While details of the upstream mechanisms and mediators remain to be defined, activated executioner caspases 3 and 7 are capable of cleaving tau protein, which may favor formation of NFTs. Asterisk indicates possible sites of action of Aβ. Caspases 1 and 5 are involved in cytokine activation.
Activation of caspases has been reported in AD using antibodies that are specific to the activated forms of the enzymes, which themselves are activated by proteolysis (13). Increasingly, apoptotic mechanisms are viewed to play a role in neurodegeneration in the absence of overt apoptosis. The process may even act in discrete subcellular domains such as synaptic termini (14). If caspase activation plays a role in AD, it is important to learn how activation occurs, where in the neuron this process takes place, and whether or not it leads to neuronal death.
-
Proteolysis in AD and other neurodegenerative disorders
The first successful attempts to produce fibrils from tau used constructs with little more than the microtubule-binding domain, but more recent efforts show that full-length tau also forms fibrils in vitro. The microtubule-binding domain appears to be important for tau-tau interaction, and structural studies of the tau filaments isolated from Alzheimer brains reveal that the filaments have a fuzzy coat that can be cleared by protease treatment, leaving intact tau filaments. Analysis of the filaments with the fuzzy coat shows full-length tau, while the core contains the microtubule-binding domain of tau and a small number of flanking sequences (24).
If proteolytic events are crucial to the conformational changes that characterize early tau pathology, it is important to determine the proteases that catalyze this reaction. Tau undergoes proteolysis by at least two types of proteases — caspases and calpains (25, 26). Calpains are calcium-dependent proteases, and their activation by calcium fits well with pathogenetic scenarios that invoke increases in intracellular calcium as critical events in neurodegeneration. Caspases, on the other hand, are activated by conditions that favor apoptosis as discussed above. Interestingly, there is experimental evidence to suggest that both types of proteases can be activated during apoptosis (25). Moreover, proteolytic products generated by caspases and calpains can be difficult to differentiate from each other (27). While calpains are activated in nonapoptotic cell death, they may also play a role in the signal transduction that leads to apoptosis, particularly in the setting of endoplasmic reticulum stress.
Understanding the relationship of proteolysis of tau protein to subsequent conformational changes and fibrillar aggregation is likely to shed light not only on AD, but also on other common neurodegenerative disorders, including PD. In AD and PD there is compelling evidence that proteolytic processing events contribute to aggregation of abnormal proteins derived from normal cellular proteins (Table 1). Both APP and tau are subject to proteolysis in AD, which produces proteins with abnormal conformations that aggregate into fibrillar intracellular and extracellular lesions. Similarly, the protein that accumulates in Lewy bodies within degenerating monoaminergic neurons in PD, α-synuclein, is also subject to proteolysis, since amino- and carboxyl-terminal truncated forms of α-synuclein are enriched in diseased brains (28). As with AD, it remains to be determined with certainty if the proteolytic events are necessary and early events that lead to neurodegeneration. Alternatively, they may be secondary processes in degenerating neurons subjected to activation of apoptotic mechanisms or flooding of intracellular compartments by calcium due to other underdetermined pathogenic processes.
Copyright © 2024 American Society for Clinical Investigation
ISSN: 0021-9738 (print), 1558-8238 (online)