The role of cell cycle-mediated events in Alzheimer's disease

Abstract

The mechanism(s) underlying selective neuronal death in Alzheimer's disease remain unresolved. However, recently, we and others showed that susceptible hippocampal neurones in Alzheimer's disease express markers common to cells in various phases of the cell cycle. Since neuronal maturation is associated with effective escape from the cell division cycle, emergence out of quiescence may be deleterious. Here, we review a number of current findings indicating that disregulated ectopic re-activation of cell cycle-mediated events, akin to neoplasia, represent an important early pathway associated with neuronal death and, more importantly, one that involves virtually the entire spectrum of the pathological events described in Alzheimer's disease.

Recent observations demonstrate evidence of attempted re-entry into the cell division cycle in neurones in Alzheimer's disease (AD). Progression through the cell cycle is associated with many features associated with AD including the activation of signal transduction pathways, cell phase dependent kinases and transcriptional activation, that leads to cytoskeletal alterations and increases in mitochondrial metabolic activity and DNA replication. Rather than completing the cell cycle, we propose that there is an abortive progression that has been shown to be related to cellular dysfunction, oxidative stress, a possible protective response to apoptosis, end stage phenomena and neuronal death.

Cell cycle abnormalities in Alzheimer's disease

Adult neurones are generally thought as being effectively precluded from division and reside in a state of terminal differentiation. Thus, it is surprising that vulnerable neurones in AD, a neurodegenerative condition characterized by selective neuronal death, have certain phenotypic markers reminiscent of a cycling, rather than a quiescent nondividing, cell (Smith & Lippa 1995; Arendt et al. 1996, 1998; Vincent et al. 1996, 1997; McShea et al. 1997; 1999a, b). Though terminally differentiated neurones may lack the ability to complete the cell cycle, they may be able to proceed variously through to a point prior to the actual event of cellular division. Thus one might expect to observe a characteristic ‘molecular phenotype’ representing a neurone arrested in a transitional phase of the cycling process. Indeed, the presence of growth promoting genes such as CDK4 (Arendt et al. 1996; McShea et al. 1997) and CDK2 (A. K. Raina and M. A. Smith, unpublished observation) imply that vulnerable neurones in AD have emerged from quiescence. Additionally, the presence of p21, highly phosphorylated protein, Ki67, p107, mpm-2 other cell cycle-related proteins (Baumann et al. 1993; Nagy et al. 1997a, b; A. K. Raina and M. A. Smith, unpublished observation) suggest that the vulnerable neurones may be attempting to initiate the early phases of mitosis. However, as yet, there is no evidence suggesting a successful nuclear division nor chromosomal condensation in AD implying that the neurones do not complete mitosis (M-phase) and that the proteins associated with exit from the cell cycle such as p16 are apparently upregulated (McShea et al. 1997).

Cell cycle, τ, phosphorylation and NFT

The increased phosphorylation status of the microtubule-associated protein τ associated with the genesis of neurofibrillary tangles is a poorly understood phenomena (reviewed in Bramblett et al. 1993). In fact, similar, if not identical, forms of phosphorylated τ proteins are generated in mitotic cells (Brion et al. 1985, 1994). This mitotic τ phosphorylation is mediated by the primary regulators of cell-cycle progression, i.e. cyclin dependent kinases (CDKs), and it is of note that several of these CDKs have been localized in vivo in AD and phosphorylate τin vitro similarly to that found in vivo (Freeman et al. 1994; Arendt et al. 1995, 1996; Gartner et al. 1995; Alcorta et al. 1996). Increased phosphorylation during mitosis coincides with microtubule reorganization and it is therefore perhaps not surprising that AD is invariantly associated with microtubular abnormalities. Furthermore, taxanes, i.e. taxol that stabilize polymeric tubulin result in an increase in mitotic specific phosphoepitopes, due to growth arrest in a phase of the cell cycle just prior to mitosis, induction of p21 and the initiation of chromosomal condensation, i.e. features shared with neurones in AD.

Oxidative stress in Alzheimer's disease

There is a significant body of evidence that the degeneration of senescent neurones is associated with oxidative stress (e.g. Smith et al. 1994a, b, 1996, 1997). The brain responds to oxidative assaults in part through the activation of NF-κB (Schreck et al. 1991) and by the induction of specific antioxidant enzyme systems. For example, haem oxygenase-1 (HO-1) is associated with degenerating neurones (Smith et al. 1994b; Castellani et al. 1995, 1996) and specific induction of HO-1 (an inducible isoform), but not HO-2 (a constitutively produced, noninducible isoform), parallels the regional susceptibility seen in neuronal degeneration (Premkumar et al. 1995). Interestingly, and suggestive of a mechanistic connection, markers against cytoskeletal alterations (i.e. abnormal τ), oxidative stress (i.e. HO-1) and cell cycle alterations (e.g. p16) all show complete overlap in degenerating neurones in AD (McShea et al. 1997).

Cell cycle and oxidative stress

The suggestion that cell cycle abnormalities are linked to alteration of the redox state (via oxidative or even reductive stress) find support from the following studies. First, free radicals and free-radical generators delay cell replication time (Curcio & Ceriello 1992) whereas, N-acetylcysteine, an antioxidant, suppresses proliferation and DNA synthesis of PC12 cells (Ferrari et al. 1995). Second, accumulation of p53 is synchronous with oxidative stress and causes blockage of the cell cycle at G1 (reviewed in Stewart 1994; Sugano et al. 1995). Third, prior to mitosis, there is both division of organelle nucleoids and organellokinesis such that during late S, G2 and mitotic phases, mitochondrial proliferation is most evident (Barni et al. 1996). While these events are critical for the high energy demands required for actual cell division, in cells where the cell cycle is interrupted, such excess mitochondrial concentrations represent sources of varied homeostatic imbalances, especially those of calcium (Sousa et al. 1997) and the redox state. Notably, there are significance increases in mitochondria in vulnerable neurones in AD (Hirai et al. 1998) and such neuronal populations are the same as reported to have cell cycle abnormalities and consequent oxidative damage in AD. Therefore, it seems likely that arrest of the cell cycle, at a point where mitochondrial mass is highest, poses an elevated and possibly a chronic oxidative threat to the cell, far beyond the blunting capacity of endogenous antioxidants.

APP, presenilins and cell cycle regulation

Genetic-linkage studies have established that mutations in at least three genes, the amyloid precursor protein (APP) located on chromosome 21, and the two homologous genes, presenilin 1 and presenilin 2 (PS1 and PS2) located on chromosome 14 and 1, respectively, are associated with early onset AD (reviewed by Hardy 1997). The exact mechanisms by which mutations in these genes are involved in AD etiopathogenesis has not been resolved. However, one hypothesis that is gaining increasing support is that the mutations in each of the three genes predispose cells to apoptosis. Since progression of the cell cycle and apoptosis are thought to be intimately linked (Evan et al. 1995; Meikrantz & Schlegel 1995; Evan & Littlewood 1998) this could in part account for the ectopic expression of cell cycle regulated proteins as discussed above.

The expression, phosphorylation, and metabolism of APP, have all been shown to be cell-cycle related (Suzuki et al. 1994; Oishi et al. 1997). Therefore, any perturbation of the cell division cycle, as appears occurs in AD, could lead to misregulated expression of APP, including the generation of subsequent proteolytic products. The precise function of APP is not known. APP is a single pass transmembrane protein expressed at the cell surface, with a short C-terminal region located in the cytoplasm (reviewed by Selkoe 1996). APP is proteolytically cleaved by unidentified proteases, resulting in the accumulation and aggregation of proteolytic fragments (of varying length, between 1 and 43 residues) into extracellular senile plaques which are a characteristic lesion of AD-affected brains (reviewed by Hardy 1997). The relationship between presence of these so-called amyloid-βγ (Aβ) plaques and their toxicity in AD is long-standing, but controversial, subject (see Hardy 1997; Neve & Robakis 1998). Recently new evidence obtained by Yamatsuji et al. (1996) has suggested that mutations in APP proteins increase apoptosis of neuronal cells by G-protein signalling.

Human presenilin (PS) proteins 1 and 2 are approximately 67% homologous, and although their function is unknown, recent studies have also implicated the proteins in regulation of apoptosis and in development. Mice disrupted in the PS1 gene die at birth and show defects in the central nervous system and axial skeleton (Shen et al. 1997; Wong et al. 1997). Interestingly, the Sel-12 gene of Caenorhabditis elegans, which is homologous to human presenilins, is involved in Notch-based signalling indicating that presenilins may play a role in cell fate determination (Levitan & Greenwald 1995). The PS proteins are expressed at very low levels which has lead to difficulties in determining their localization in cells. Nonetheless, the majority of reports indicate that presenilins are localized to the endoplasmic reticulum (ER) and Golgi, consistent with secondary structure predictions of a multiple transmembrane protein (see Kovacs et al. 1996; Janicki & Monteiro 1997). However, some reports have indicated that presenilins are localized to the cell surface (Dewji & Singer 1997) and also to the centromere and centrosome of dividing cells (Li et al. 1997). The binding of presenilins to centromeres and centrosomes have led to speculations that presenilins are involved in chromosome segregation and cell division, although direct evidence for this role has not been obtained so far.

There is substantial evidence that presenilins are involved in the regulation of apoptosis (Deng et al. 1996; Guo et al. 1996, 1997; Vito et al. 1996a, b; Wolozin et al. 1996, 1998; Janicki & Monteiro 1997; Zhang et al. 1998). Overexpression of presenilins in terminally differentiated neuronal cells results in perturbation of calcium homeostasis, oxidative stress, and increased vulnerability to apoptosis induced by a variety of agents (reviewed by Mattson et al. 1998). More interestingly, a common aspect of the transgenic expression of different FAD presenilin mutations is that they all increase sensitivity of cells to apoptosis compared to the expression of wild type presenilin genes. A similar increase in apoptosis is found following overexpression of the FAD PS2(N141I) mutant compared to wild type presenilin 2 in dividing HeLa cells (Janicki & Monteiro 1997), suggesting that the mechanisms of presenilin induction of apoptosis in dividing cells and postmitotic neurones may be similar. More recently, by BrdU labelling studies of transfected HeLa cells, we have shown that the overexpression of both presenilin 1 and 2 proteins results in G1 phase arrest of the cell cycle (S. Janicki and M. J. Monteiro, unpublished observation), consistent with the well established notion that apoptosis is preceded by cell cycle arrest (reviewed by Elledge 1996). Cell cycle arrest induced by PS overexpression is potentiated upon expression of the FAD PS2(N141I) mutation compared to wild type PS2 (S. Janicki and M. J. Monteiro unpublished observation). The potentiation of cell cycle arrest by the FAD PS2 mutant is revealing since it might have been predicted that such a mutation would decrease cell cycle arrest to account for the increased reexpression of cell cycle regulated proteins in AD. Although this possibility is still not excluded by the transfection results in HeLa cells, we suggest alternative explanations.

One possibility we propose is that in AD, neurones are attempting to re-enter the cell cycle but are blocked from entry at the G1/S phase of the cell cycle by presenilin (and possibly APP) mutations. This blockage at G1 would result in accumulation of cell cycle proteins as is seen in AD. Another possibility is that presenilin and APP mutations induce proapoptotic cascades which result in the upregulation of cyclin-dependent kinases, since the latter appear to play dual roles in controlling cell proliferation as well as in death signalling (see Park et al. 1998). The induction of cyclin-dependent kinase inhibitors, such as p16, p19, p21 and p27, in AD may reflect a protective response elicited by the cells to these death stimuli. There is good evidence for upregulation of cyclin dependent kinases in cultured neurones upon insult by a number of apoptotic inducing agents (Park et al. 1998). Interestingly, p16, p21, and p27, which are inhibitors of CDK4 and CDK6 kinases protect neurones from these apoptotic insults (Park et al. 1998), and many of these same proteins are reexpressed in AD neurones (Ledesma et al. 1992; Smith & Lippa 1995; Ardent et al. 1995, 1996; Vincent et al. 1996, 1997; Nagy et al. 1997a, b; McShea et al. 1997, 1999a,b).

Cell cycle, growth factors, and oncogenic stimulation

Given the apparent importance of mitotic mechanisms in AD, one is left with a key remaining question, namely to determine what factor(s) stimulate the neurones to re-enter the cell division cycle. A likely explanation is the inappropriate expression, or release, of a growth or cellular differentiation factor(s), and, in this regard, it is noteworthy that elevated levels of nerve growth factor (NGF), transforming growth factor-1 (TGF) and basic fibroblast growth factor (bFGF) are found in AD (Gomez-Pinilla et al. 1990; Crutcher et al. 1993). Indeed, in AD, there are significant alterations in signal transducing pathways that respond to growth factor stimulation such as the ras and mitogen-activated protein kinase pathways (McShea et al. 1999a). More importantly, signalling leading to emergence from quiescence, increased phosphorylation of τ, re-expression variously of a developmental phenotype, and dysregulation of cell-matrix homeostasis (M. A. Smith, unpublished observation), is in effect homologically similar to proximal events associated with neoplastic transformation. Indeed, ectopic mitogenic signalling is frequently associated with early events in neoplasia lending the possibility that a neoplastic-type transformation occurs in neurones in AD.

Conclusions

In summary, mitotic neurones and degenerating neurones in AD, effectively share a cell-cycle related phenotype. Whether re-entry into the cell cycle reflects a primary pathogenic pathway or, is instead, a protective response to cell death stimuli, remains to be determined.