“Amyloid‐beta accumulation cycle” as a prevention and/or therapy target for Alzheimer's disease

Abstract The cell cycle and its regulators are validated targets for cancer drugs. Reagents that target cells in a specific cell cycle phase (e.g., antimitotics or DNA synthesis inhibitors/replication stress inducers) have demonstrated success as broad‐spectrum anticancer drugs. Cyclin‐dependent kinases (CDKs) are drivers of cell cycle transitions. A CDK inhibitor, flavopiridol/alvocidib, is an FDA‐approved drug for acute myeloid leukemia. Alzheimer's disease (AD) is another serious issue in contemporary medicine. The cause of AD remains elusive, although a critical role of latent amyloid‐beta accumulation has emerged. Existing AD drug research and development targets include amyloid, amyloid metabolism/catabolism, tau, inflammation, cholesterol, the cholinergic system, and other neurotransmitters. However, none have been validated as therapeutically effective targets. Recent reports from AD‐omics and preclinical animal models provided data supporting the long‐standing notion that cell cycle progression and/or mitosis may be a valid target for AD prevention and/or therapy. This review will summarize the recent developments in AD research: (a) Mitotic re‐entry, leading to the “amyloid‐beta accumulation cycle,” may be a prerequisite for amyloid‐beta accumulation and AD pathology development; (b) AD‐associated pathogens can cause cell cycle errors; (c) thirteen among 37 human AD genetic risk genes may be functionally involved in the cell cycle and/or mitosis; and (d) preclinical AD mouse models treated with CDK inhibitor showed improvements in cognitive/behavioral symptoms. If the “amyloid‐beta accumulation cycle is an AD drug target” concept is proven, repurposing of cancer drugs may emerge as a new, fast‐track approach for AD management in the clinic setting.

However, these ongoing approaches have not shown success in the clinic yet. In translational studies with mouse models, a newer approach targeting amyloid-beta oligomer has shown promise. In 6-to 7-month-old Tg 5xFAD mice, injection of an Aβ oligomer-specific antibody rescued memory performance (Cline et al., 2018).
In late-onset AD, amyloid-beta accumulation and senile plaque development begin in middle age. At the early stage, no immediate visible symptoms appear, and the time window is considered presymptomatic or latent. Ten to fifteen years later, neurofibrillary tangles begin to emerge. However, there is also a 3 to 5-year time window without notable memory or cognitive symptoms (Perrin, Fagan, & Holtzman, 2009). In a later stage of AD, memory/cognitive/behavioral symptoms and amyloid-beta plaques and neurofibrillary tangles are observed. Notably, plaques and tangles in the AD brain appear with differential localizations; plaques are diffusely distributed over the entire cortex, while tau tangles are seen in brain regions related to clinical symptoms and overlap with areas of hypometabolism (Dronse et al., 2017). As AD progresses, both lesions spread throughout the brain, each following a stereotyped pattern. Senile plaques spread from association cortices (Thal phase 1) to allocortical areas, including the hippocampus (Thal phase 2); then to diencephalic nuclei, striatum, and cholinergic nuclei of the basal forebrain areas (Thal phase 3); several brain stem nuclei (Thal phase 4); and cerebellum (Thal phase 5) (Thal, Rüb, Orantes, & Braak, 2002). Neurofibrillary tangles first appear in the entorhinal cortex (Braak stages I/II), then spread toward limbic structures, including the hippocampus (Braak stages III/IV) and association cortices (Braak stages V/VI) (Braak & Braak, 1991). Additionally, amyloid-beta accumulation around blood vessels, also known as cerebral amyloid angiopathy, appears.
Cerebral amyloid angiopathy may play a role in cognitive/ behavioral symptoms via cerebrovascular system dysfunction (Weber, Patel, & Lutsep, 2018). This complex pathology led to a controversy regarding which pathology (amyloid-beta plaques, tau/p-tau tangles, or cerebral amyloid angiopathy) is more critical in AD symptoms and therefore represents a better therapeutic target. As the decades-long development of AD pathology is associated or concurrent with many biological events and general aging, it has been difficult to pinpoint an effective intervention or therapeutic target. As approaches targeting late-stage AD pathologies have not been successful, and as early amyloid-beta accumulation with oligomer formation has begun to be recognized as a critical step with the amyloid-beta oligomer hypothesis, the need to understand mechanisms leading to amyloid-beta accumulation is increasing.

| CURRENT AD DRUG RE S E ARCH AND DE VELOPMENT TARG E TS
Mainstream AD drug research and development efforts have been directed toward (i) components directly involved in AD pathology, such as amyloid-beta or p-tau, amyloidogenic proteases, and other amyloid-beta or tau binding proteins, such as receptor for advanced glycation endproducts (RAGE); (ii) components associated with pathology and predicted to be involved in symptoms, such as neuroinflammation, oxidative stress, and mitochondrial dysfunction; and (iii) medicines that can ease cognitive/behavioral symptoms of AD, including neuronal function modulators and neurotransmitters, although they may be palliative and may not address pathological causes nor lead to fundamental cure.
As of 30 October 2019, the Clini caltr ials.gov database lists 2,214 clinical trials involved in AD (https ://clini caltr ials.gov/). The Alzforum website, which maintains a database on potential dementia therapeutics, lists 216 therapeutic reagents for AD and mild cognitive impairment (MCI) (https ://www.alzfo rum.org/thera peutics). However, only five reagents (donepezil, galantamine, memantine, rivastigmine, and tacrine) are approved by the US FDA for use as symptomatic relievers of AD, targeting the cholinergic system and other neurotransmitters. No other agent has been validated as a clinically effective target.

| INTRIN S I C AND E X TRIN S I C FAC TOR S THAT AFFEC T AMYLOID -B E TA AMOUNT
The earliest step of AD pathology development is cerebral accumulation of amyloid-beta, the mechanism of which has been elusive. A majority of sporadic AD patients do not carry mutation in familial AD genes (i.e., APP, PSEN1, or PSEN2) (Lanoiselée et al., 2017), suggesting that other drivers are involved in the development of sporadic late-onset AD cases. Here, we discuss amyloid-beta accumulation as a result of the disturbed balance between Aβ generation and catabolism (i.e., Aβ dyshomeostasis). Consistently, AD patients indicated impaired clearance rates for Aβ42 and Aβ40, compared with controls (Mawuenyega et al., 2010).

Various intrinsic factors can affect Aβ generation and/or catabo-
lism. Amyloid-beta can be generated through intensive neuronal activity (i.e., activity-dependent Aβ generation) (Ovsepian & O'Leary, 2016). Activation of and/or increase in amyloidogenic proteases can play a role in increasing Aβ, as observed in patients with Down syndrome who exhibit early AD-like dementia and express a large amount of BACE1 (Miners, Morris, Love, & Kehoe, 2011). Systemic factors, such as the effect of oral and gut microbes on systemic inflammation, are also gathering interest (Tremlett, Bauer, Appel-Cresswell, Finlay, & Waubant, 2017). Sleep affects accumulation and removal of Aβ (Cordone, Annarumma, Rossini, & Gennaro, 2019). Reports indicate that Aβ catabolism decreases over age. Older mice showed a 40% decline in clearance of injected Aβ, which may be caused by a decline in the efficiency of exchange between the subarachnoid cerebrospinal TA B L E 1 Thirteen among 37 genes on human AD genetic risk loci are functionally involved in the cell cycle and/or mitosis Gene Name Full Name Proposed function (s) and pathway(s) involved* "Cell cycle"/total publications** "Mitosis"/total publications** Reported involvement in cell cycle and/or mitosis (including possible link)

ZCWPW1
Zinc finger CW-type and PWWP domaincontaining 1 Function poorly known, Note: A list of 37 AD genetic risk loci (genes identified as frequently mutated in AD-omics studies) was used. The number of existing publications was examined for each gene, followed by a keyword search with "Gene X and cell cycle" or "Gene X and mitosis." The process provides an estimate for the current total research activity regarding the gene and for the gene's functional involvement in the cell cycle and/or in mitosis. Genes showing a direct or strong functional connection to the cell cycle and/or mitosis are marked in bold. Thirteen among the 37 AD genetic risk loci have indicated functions in the cell cycle and/or mitosis, suggesting the importance of the cell cycle and/or mitosis in AD development. *Based on GeneCards database. **Publication numbers as of 5 November 2019, via PubMed keyword search.

TA B L E 1 (Continued)
fluid and the brain parenchyma (Kress et al., 2014). Levels of neprilysin, an Aβ-degrading protease, decreased with age in both normal and AD patients. This decreasing neprilysin level may act as a trigger for AD (Hellström-Lindahl, Ravid, & Nordberg, 2008). In APP-SL70 mice, the microglial response to increasing amyloid-beta was estimated to be overwhelmed with aging (Blume et al., 2018).
In addition to intrinsic factors, extrinsic or environmental factors can play a role. In a mouse model, recurrent activation of brain herpes simplex virus 1 (HSV1) infection led to amyloid-beta accumulation and other AD pathology (tau phosphorylation, neuroinflammation) (De Chiara et al., 2019), indicating that viral infection can trigger Aβ accumulation and AD pathology. In an AD-omics study, HHV-6A and HHV-7 were identified as prominently associated with human AD across three independent cohorts (Readhead et al., 2018). Reports like these support the theory that pathogens trigger AD (Haas & Lathe, 2018;Itzhaki, 2014;Sochocka, Zwolińska, & Leszek, 2017), as well as the role of amyloid-beta as a protection mechanism against viral infection (Li, Liu, Zheng, & Huang, 2018). Although AD cannot be completely explained by the pathogen theory alone, pathogens may act as a risk factor or have an impact on a segment of patients with AD. A high rate of HSV1 and other infections was observed in AD patients (Sochocka et al., 2017). Since three subtypes of AD were identified based on the spread of neurofibrillary tangles , there may be more than one causal process, leading to distinct pathology of AD subtypes.

| A B RIEF HIS TORY OF E ARLIER S TUD IE S OF ANEUPLOIDY AND AD
Since the 1990s, a long-standing theory has purported that aneuploidy plays a critical role in AD development. In an early thesis noting AD-like dementia in patients with Down syndrome, Potter (1991) hypothesized that (i) aneuploidy (chromosome 21 trisomy) is causal to AD, and that (ii) genes associated with the risk of AD would be involved in the cell cycle, and such genes would lead to the development of aneuploidy when mutated (Potter, 1991).
Following hypothesis (i), the link between aneuploidy and AD was explored using cytogenetics. Earlier studies tested the rate of aneuploidy in peripheral blood lymphocytes and fibroblasts from AD patients. Cells from familial and sporadic AD patients were shown to have more micronuclei than controls. The antifungal drug griseofulvin mitigated the increase in micronuclei in cells from patients with AD, indicating an altered response to genotoxic challenge Trippi et al., 2002). These results may be attributed to increased DNA damage and impaired DNA repair (Coppedè & Migliore, 2009), or perhaps to DNA replication stress (Yurov, Vorsanova, & Iourov, 2011). Newer studies indicated a higher aneuploidy rate in AD-affected neurons (Iourov, Vorsanova, Liehr, & Yurov, 2009). In addition, the presence of hyperploid neurons was noted (Arendt, Brückner, Mosch, & Lösche, 2010) [see Section 6.3 "High degree of aneuploidy in patients with AD and mild cognitive impairment (MCI)" for additional references].
Although a single-cell sequencing report noted conflicting results (van den Bos et al., 2016), and although reported rates of ane-  & Potter, 2013) also cause aneuploidy in neurons. In a later section, "Thirteen among 37 genes on the human AD genetic risk loci are functionally involved in the cell cycle and/or mitosis" (Table 1), we will discuss newer corroborating evidence from contemporary AD-omics.
In Boveri's 19th-century theorem, aneuploidy was predicted to cause cancer (Boveri, 2008). Aneuploidy can be caused by external genotoxic challenges (e.g., radiation, chemicals, and virus) or by an internal defect in molecular mechanisms for genome maintenance that are intimately involved in cell cycle regulations. Mainstream mechanistic studies of the cell cycle emerged rather independently from studies of AD. The conceptual framework that the cell cycle is driven by cyclin-dependent kinases (CDKs) and cyclins emerged by the mid-1980s/early 1990s (Nurse, 2012 Murray & Hunt, 1993).
Cancer, carcinogenesis, and developmental defects associated with aneuploidy have been the primary disease targets of cell cycle studies. These types of investigations were assumed to be useful for finding methods and targets to manipulate cell cycle and cell growth, which could lead researchers to cancer therapeutics. Over the years, this assumption has been shown to be correct. Many validated cancer therapeutics target machineries that are involved in the cell cycle. In the 1990s-2000s, aneuploidy-inducing transgenic genomic instability mouse models, such as chromosome instability (CIN) models and microsatellite instability (MIN) models, were developed. The models were used mainly to assess cancer development (Rao & Yamada, 2013;Rao, Yamada, Yao, & Dai, 2009;Simon, Bakker, & Foijer, 2015) and to examine the relationship between aneuploidy and carcinogenesis (Weaver & Cleveland, 2009;Zasadil et al., 2016). However, most cancer assessment studies do not keep mice until old age. It was only recently that experimental results testing the link between genomic instability and AD in aged mouse models started to be reported .

| THE "AMYLOID -B E TA ACCUMUL ATION C YCLE"
From the notion that aneuploidy plays a role in AD development, some hypotheses focusing on the role of mitotic cycle re-entry of neurons evolved. One such hypothesis was the "two-hit hypothesis" that proposes age and mitotic re-entry as two key factors ("hits") for AD development (Webber et al., 2005;Zhu, Lee, Perry, & Smith, 2007;Zhu, Raina, Perry, & Smith, 2004). Based on results from genomic instability mouse models, we proposed a version of the two-hit hypothesis with an emphasis on the role of prolonged mitosis in accumulating amyloid-beta, the "three-hit hypothesis." The "three-hit hypothesis" proposes (I) aging, (II) mitotic re-entry, and (III) prolonged mitosis as three key factors for the development of AD (Rao, Farooqui, Zhang, Asch, & Yamada, 2018). Recent literature led us to an integrative hypothesis, the "amyloid-beta accumulation cycle," incorporating interference in mitosis and the aneuploidogenic role of amyloid-beta ( Figure 2). The rationales for the "amyloid-beta accumulation cycle" are summarized below. inhibitory proteins (e.g., wee1 kinase) in the direction toward mitosis; in degenerating neurons, Cdc25A and Cdc25B show higher activity, while Wee1 shows lower activity (Ding et al., 2000;Tomashevski, Husseman, Jin, Nochlin, & Vincent, 2001;Vincent et al., 2001). In a 3xTg AD mouse model and in human AD patients, hyperphosphorylated retinoblastoma protein, a marker for G1/S transition, co-localized with hyperphosphorylated tau, linking aberrant cell cycle progression with tau pathology (Hradek et al., 2015). In the study, Hradek et al. used 19-month-old animals, thus leaving a question if the pRb accumulation is age-associated.

| Neuronal cell cycle re-entry occurs in AD
Lopes, Blurton-Jones, Yamasaki, Agostinho, and LaFerla (2009)  This model also displayed a significant increase in hyperphosphorylated tau and Abeta, supporting the possibility that cell cycle re-entry may lead to AD-like changes even in animals without a previous alteration of the genes related to Abeta or tau.

| Various mitogenic/growth signaling factors are activated in the AD brain
Consistent with Section 6.1, misregulations in various mitogenic/ growth signaling factors, including ERK/MAPK, GSK3, PI3K/AKT, and CDK, are reported in the AD brain (Kirouac, Rajic, Cribbs, F I G U R E 2 (a) The "amyloid-beta accumulation cycle". Normal neurons or glia are challenged by mitotic signaling, which may be associated with age and the microenvironment, such as high reactive oxygen species (ROS), reduced antioxidants, damaged blood-brain barrier, and fatigued stem cells, or other pathogenic conditions, such as diabetic wounds, pathogen infection, or mutated AD risk gene. Mitogenic signaling causes neurons or glial cells to enter the cell cycle and attempt to go through mitosis. In the cycling cells, aneuploidy, an environmental factor, other mutations in an AD risk gene, or already existing extracellular amyloid-beta cause cells to go through prolonged mitosis or a quasi-mitotic state with high mitotic kinase activity, when they accumulate amyloid-beta, BACE, and p-tau. If the state is not resolved, mitotic catastrophe occurs, and accumulated amyloid-beta, BACE, and p-tau are released to the microenvironment. Released amyloid-beta, with its prion-like properties, may function as seeds for subsequent plaque pathology. Extracellular amyloid-beta can provoke inflammation and mitogenic signaling, and can also cause mitotic errors, prolonged mitosis, and aneuploidy. Thus, age-or microenvironmentprovoked mitogenic signaling can trigger a vicious cycle leading to further amyloid-beta accumulation (the "amyloid-beta accumulation cycle") (blue/purple arrows). (b) Cancer drugs that target mitotic re-entry and/or prolonged mitosis may be valid drugs for managing the "amyloid-beta accumulation cycle" and AD. The two-hit hypothesis (Zhu et al., 2007(Zhu et al., , 2004 proposed age and mitotic re-entry as crucial events for development of AD pathology. In light of the apparent importance of prolonged mitosis in this process, we proposed the threehit hypothesis . The "amyloid-beta accumulation cycle" is an integrated hypothesis that emerged from the three-hit hypothesis. The "amyloid-beta accumulation cycle" suggests that a reagent that interferes with amyloid-beta accumulation could be an AD drug. As the cell cycle and mitosis are validated targets for cancer drugs, repurposing of cancer drugs for AD management may emerge as a viable clinical option in the near future. (c) Cerebral amyloid-beta protein can accumulate in mice with an unmodified APP gene under certain conditions. Under normal circumstances, wild-type mice with an unmodified APP gene do not accumulate amyloid-beta in the brain, even in old age (24 months and older). AD modeling in mice has been dependent on introduction of transgenic mutations in genes involved in familial/early-onset AD (e.g., APP, PSEN1, and MAPT), representing early-onset AD models (Jankowsky & Zheng, 2017;Saito & Saido, 2018). A rodent model for sporadic late-onset AD has been an unmet need. Over 96% of all human AD cases are late-onset and sporadic, a majority of which carry no mutation in known early-onset AD genes. Thus, identifying conditions under which amyloid-beta accumulates is valuable to gain mechanistic insights on AD development and to model late-onset AD. A progeria mouse model SAMP8 was reported to accumulate amyloid-beta, yet the causal mutation remains unidentified (Akiguchi et al., 2017). Recent reports began to identify conditions that can cause amyloid-beta accumulation in the mouse brain with unmodified APP or other known early-onset AD gene mutations. Examples of amyloid-beta accumulating conditions include (i) aged Sgo1 −/+ mice, a cohesinopathy-chromosome instability mouse model (Rao, Farooqui, Zhang et al., 2018) Tarozzi, Rimondini, & Hrelia, 2016;Sun, Liu, Nguyen, & Bing, 2003;Swatton et al., 2004;Vincent, Jicha, Rosado, & Dickson, 1997).
Amyloid-beta oligomers are synaptotoxic and can activate the signaling axis that involves the tyrosine kinase ephrin receptor A4 (EphA4) and c-Abl tyrosine kinase. When mutated, c-Abl can act as an oncogene (Vargas, Cerpa, Muñoz, Zanlungo, & Alvarez, 2018). CDK7 can act as a CDK-activating kinase and can activate the major Cdk-cyclin substrates. CDK7 expression is age-dependent and is elevated in hippocampal neurons of AD patients (Zhu et al., 2000). The stress signaling AMPK pathway is also reported to be involved in AD (

| High degree of aneuploidy in patients with AD and mild cognitive impairment (MCI)
Human brains are naturally aneuploidogenic during early development and carry a higher rate of aneuploid cells in adulthood. However, This hippocampal defect may affect neurocognitive performance.

| Mutations associated with familial/early-onset AD can cause mitotic error and aneuploidy, as well as other cell cycle disturbances
Presenilin 1/PSEN1 mutation (e.g., familial AD mutation in presenilin 1 [M146L and M146V]) is linked to familial/early-onset AD.
Overexpression of mutant PSEN1 caused chromosome missegregation and aneuploidy in vivo (mice) and in vitro, with mitotic spindle defects observed (Boeras et al., 2008). PSEN1 P117R mutation is a pathogenic AD mutation that can cause increases in p53 and p21 proteins, G1 phase prolongation, S phase shortening, and decreased apoptosis in human lymphocytes (Bialopiotrowicz et al., 2012).
Lymphocytes are proposed to serve as a surrogate indicator for the development of AD, as these cells are responsive to oxidative stress and other challenges, and are indicative of aneuploidy and cell cycle disturbances that mirror the condition of neurons in patients with AD (Wojsiat, Prandelli, Laskowska-Kaszub, Martín-Requero, & Wojda, 2015). As mentioned in Section 6.5, amyloid-beta can also disrupt the mitotic spindle and inhibit mitotic motors, thus causing mitotic defects and aneuploidy (Borysov et al., 2011). The APOE subtype is associated with AD risk. Knockdown of APOE in APOE-expressing ovarian cancer cells led to G2 cell cycle arrest and apoptosis, suggesting its context-dependent role in cell cycle progression (Chen et al., 2005).

| Forced cell cycle re-entry resulted in amyloidbeta and p-tau accumulation in mouse brains
Transgenic mice in which cell cycle reactivation in neurons is forced by SV40 T antigen via the tet-on/off system show signs of mitotic re-entry (e.g., PCNA, cyclin B1, MPM2) and Aβ deposits and phosphorylated tau in the brain (Park, Hallows, Chakrabarty, Davies, & Vincent, 2007). Expression of SV40 T antigen causes replication stress, mitotic dysfunction, and aneuploidy (Hu, Filippakis, Huang, Yen, & Gjoerup, 2013), suggesting a link among mitotic re-entry, aneuploidy, and AD pathology.

| Amyloid-beta can bind to mitotic motors and microtubules, causing mitotic error and aneuploidy, as well as triggering the stress response
Amyloid-beta can disrupt the mitotic spindle and inhibit mitotic motors (e.g., Eg5, KIF4A, MCAK), causing mitotic defects with prolonged mitosis and aneuploidy (Borysov et al., 2011). The transcriptome of cultured SH-SY5Y cells expressing P301L tau was most affected in the cell cycle and cell proliferation; proteomic analysis on an amyloid-beta (1-42)-injected mouse model revealed that the stress response and metabolism pathways were most affected (Götz et al., 2008). Amyloid-beta injection in a rat model caused pro-apoptotic changes (increased caspase-3, decreased Bcl2/Bax ratio) and activation of stress/mitogenic signaling (increased pERK, pJNK, and NFkB65kd; decreased IkB) (e.g., Dong, Ji, Han, & Han, 2019). Thus, once accumulated, amyloid-beta can be aneuploidogenic by itself and can trigger stress response and cell death.

| Infection with AD-associated pathogens can cause mitotic re-entry, mitotic errors, and/or prolonged mitosis
Various pathogens, including viruses (HHV1-6, HCV), bacteria (Chlamydia pneumoniae, Helicobacter pylori), fungi (Candida albicans), and protozoa (Toxoplasma gondii), have been identified as potential AD risk factors (Sochocka et al., 2017). Herpes simplex virus 1/ HSV1/HHV1 immediate-early protein Vmw110 was shown to inhibit G1/S transition and progression through mitosis (i.e., prolonged mitosis at pseudo-prometaphase), which was in part caused by Vmw110-induced proteasome-dependent degradation of a centromeric protein CENP-C (Everett, Earnshaw, Findlay, & Lomonte, 1999;. Cytomegalovirus CMV/HHV5 infection caused transcriptomic misregulations in cell cycle and mitosis genes, and produced a pseudo-mitosis state in the infected cells (Hertel & Mocarski, 2004). These observations of AD-associated pathogens being able to cause mitotic re-entry, mitotic errors, and/or prolonged mitosis may help to reconcile the aforementioned "AD is caused by pathogen" theory and the "amyloid-beta accumulation cycle."

| WILL ANEUPLOIDY ALONE B E SUFFI CIENT TO C AUS E AMYLOID -B E TA ACCUMUL ATION?
Cohesinopathy-genomic instability model Shugoshin 1 (Sgo1) haploinsufficient mice (Sgo1 −/+ mice) showed spontaneous cerebral amyloid-beta accumulation in old age (Figure 2c; Rao, Farooqui, Zhang, et al., 2018). Normally, amyloidbeta accumulation does not occur in mice. The International Mouse Phenotyping Consortium (IMPC) database reports an abnormal behavior phenotype in Sgo1 tm1a(EUCOMM)Wtsi allele mice, suggesting the likelihood of AD-like cognitive function/behavioral issues with Sgo1 defects (http://www.mouse pheno type.org/data/genes/ MGI:19196 65#secti on-assoc iations). In the Sgo1 −/+ mice, we did not observe a higher amount of APP protein. Thus, accumulation of precursor protein APP was unlikely to be the cause of amyloid-beta accumulation.
However, spindle checkpoint defect-genomic instability model BubR1 −/+ mice did not show cerebral amyloid-beta accumulation (Rao, Farooqui, Zhang et al., 2018), suggesting that aneuploidy alone may not be sufficient to cause amyloid-beta accumulation in a mouse model. Since a major difference in these two chromosome instability-aneuploidogenic models is spindle checkpoint function and prolonged mitosis, prolonged mitosis was proposed to be one of the three critical factors (the "three-hit" hypothesis; Figure 2b) for amyloid-beta accumulation . Thus, types of aneuploidy that are accompanied by prolonged mitosis, such as cohesinopathy and amyloid-beta poisoning, are speculated to further lead to amyloid-beta accumulation.
Whether tetraploidization, another type of aneuploidy, contributes to AD development is a matter of controversy. Tetraploidization was reported to occur in normal and AD brains to a similar degree (Westra, Barral, & Chun, 2009). This finding suggests that the effects of tetraploidization on AD development are limited. A newer paper, however, reported a correlation between neuronal tetraploidization in the cerebral cortex in mice and reduced cognition and AD-associated neuropathology, suggesting a causal role of tetraploidization in the development of AD (López-Sánchez et al., 2017).
For the tetraploidization mechanism, as AD brains abundantly express neurotrophin receptor p75NTR and proNGF (nerve growth factor), their involvement in triggering neuronal tetraploidization, subsequent abortive mitosis, cell death, and hence neurodegeneration was suggested (Frade & López-Sánchez, 2010). Determining the cause-consequence relationship of tetraploidization in AD may not be simple, as they may occur rather simultaneously.

| THIRTEEN AMONG 37 G ENE S ON THE H UMAN AD G ENE TI C RIS K LO CI ARE FUN C TI ONALLY INVOLVED IN THE CELL C YCLE AND/OR MITOS IS
Analyzing AD brains in a comprehensive and hypothesis-free manner with a combination of various -omics, imaging, and other biomarker analysis techniques has been proposed by the "Alzheimer Precision Medicine Initiative (APMI)" to advance understanding of AD, to identify dysfunctional systems and predictive markers, and to develop remedies against neurodegenerative disorders (Hampel, Toschi, et al., 2018;Hampel, Vergallo, et al., 2018). Genome sequencing projects of human AD patients and meta-analysis of the reports have revealed genes/loci that are frequently mutated in AD patients, that is, AD genetic risk loci (Beecham et al., 2014;Carrasquillo et al., 2015;Chouraki & Seshadri, 2014;Jansen et al., 2019;Kim, 2018;Kunkle et al., 2019;Lambert et al., 2013;Van Cauwenberghe, Broeckhoven, & Sleegers, 2016;Zhang, Gaiteri, et al., 2013), in addition to known familial AD mutations, such as PSEN1/2, APP, and APOE variants. The ALPK2, and BZRAP-AS1 were identified with international transethnic cohorts   (Table 1).
Using genome-wide association studies (GWASs), Han, Huang, Gao, and Huang (2017) identified functions of the genes and categorized these functions as "regulation of beta-amyloid formation," "regulation of neurofibrillary tangle assembly," "leukocyte-mediated immunity," and "protein-lipid complex assembly" signaling pathways. With the protein-protein interaction network and functional module analyses, they also identified "hub" genes and "bottleneck"  Table 1.
From supporting evidence of the involvement of the cell cycle and mitotic re-entry in AD development, we hypothesized that some of the genes identified as AD genetic risk loci are functionally involved in the cell cycle and/or mitosis. We performed a series of literature searches using "cell cycle" or "mitosis" as keywords for each of the genes. The search revealed possible functional involvement in the cell cycle or mitosis regulation for 13 among 37 genes (Table 1).
This result provides additional support to the long-standing hypothesis that human AD development is associated with, influenced by, or caused by misregulations in the cell cycle or mitosis via gene mutations, at least in part. The hypothesis warrants further investigation.

| EFFEC TS OF CELL C YCLE INTERFERING DRUG S ON AD MODEL S
The aforementioned reports suggest that two major AD pathological features, plaques and tangles, are caused by or associated with cell cycle misregulation toward mitosis. Existing pharmacological reagents can interfere with the cell cycle and its machineries, leading to the question of whether these drugs affect neuronal health and AD symptoms and pathology.

| Antimitotic drugs can be neuroprotective against tauopathy
With exceptions of mitotic kinase or mitotic motor inhibitors, most antimitotic drugs target microtubule dynamics and mitotic spindles. Taxanes, including paclitaxel/Taxol, are microtubule stabilizers, while vinblastine and vincristine are microtubule destabilizers.
Both classes of antimitotic drugs have a demonstrated history of use in cancer chemotherapy (Florian & Mitchison, 2016 (Shemesh & Spira, 2011). In testing drugs in animal models, bloodbrain barrier penetration and cerebral drug availability must also be considered (Brunden et al., 2011). In the PS19 tau transgenic mouse model of tauopathy, the brain-penetrant antimitotic drug epothilone D reduced the burden of tau pathology . Another brain-penetrant microtubule stabilizer, dictyostatin, also produced improvement in CNS/brain measures (Brunden, Lee, Smith, Trojanowski, & Ballatore, 2017;Makani et al., 2016). Note that most antimitotics target microtubule dynamics. The possibility remains that their neuroprotective effects occur through microtubule and microtubule-binding protein-mediated signaling and/or axonal transport, rather than cell cycle effects (Brunden et al., 2017;Trojanowski, Smith, Huryn, & Lee, 2005

| CDK inhibitors ameliorated AD symptoms in animal AD models
CDK5 is a unique member of the CDK family. Unlike canonical cell cycle driving CDKs, such as CDK1 and CDK2, Cdk5 is inactive in the cell cycle, but is specifically expressed and predominantly active in postmitotic neurons. Its role in AD has long been postulated (Dhavan & Thai, 2001). In physiological conditions, CDK5 binds with its activator, p35, and plays roles in the development of CNS and movements of neurons. However, once neurons experience pathogenic challenges, Cdk5 associates with p25, which is generated from p35 by calpain-dependent cleavage, and becomes hyperactivated. CDK5/p25 causes aberrant hyperphosphorylation of various substrates that include APP, tau, and neurofilaments.
Flavopiridol/alvocidib is a potent and specific inhibitor of CDKs 1, 2, 4, and 7 in vitro, showing a clear blockade of cell cycle progression at the G1/S and G2/M boundaries (Senderowicz, 1999). In the hCOX-2 transgenic mice, overexpression of human COX-2 in murine primary hippocampal neurons accelerated beta-amyloid-mediated apoptosis.

| Targeting mitogenic signaling
Mitogenic signaling is another potential target for AD drugs. tideglusib, an inhibitor of GSK3 (Lovestone et al., 2015); and lithium, a GSK3 inhibitor (Forlenza et al., 2011). Although this "targeting mitogenic kinase/signaling" approach has not become mainstream in AD research and drug development, the approach has been established in cancer research and drug development, with success against oncogene-addicted cancers (e.g., imatinib/Gleevec targeting Bcr-Abl tyrosine kinase, trastuzumab/Herceptin targeting Neu receptor). Due to organ-specific issues (e.g., blood-brain barrier [BBB]-mediated drug delivery), cancer drug repurposing for AD therapy may not be straightforward. Still, given the abundance of accumulated resources for targeting mitogenic signaling, the approach may hold great future potential. regimens (Matsos & Johnston, 2019). However, there is currently little evidence that the treatments caused amyloid-beta accumulation or triggered AD-like dementia. Instead, the cognitive decline is attributed to damage to the BBB, oxidative stress, and cytokine dysregulation; thus, this chemobrain phenomenon is believed to be mechanistically closer to vascular dementia (Ren et al., 2019;Ren, St Clair, & Butterfield, 2017). Consistently, severe chemobrain is associated with cytotoxic DNA-damaging drugs, rather than CDK inhibitors. An inverse relationship between cancer and AD is known; that is, cancer survivors have less likelihood of developing AD . The inverse relationship is proposed to be related to a balance in cellular tendencies toward cell death or growth (Shafi, 2016). Mechanisms in cell survival/death regulation, that is, p53, Pin1, and the Wnt signaling pathway, were discussed as potential therapeutic manipulation targets (Behrens, Lendon, & Roe, 2009). In addition to innate cellular tendencies, we speculate that the inverse relationship may be in part due to a therapeutic or intervention effect of cancer chemotherapy drug on preclinical AD. When cell cycle-managing chemotherapy drugs are to be repurposed for AD prevention and/ or therapy, we suggest designing clinical trials with careful and conservative dosing, with a less likelihood of BBB damage and/or chemobrain induction.

| SUMMARY
Increasing evidence, including recent -omics data from human patients with AD, points to a critical role of the cell cycle and mitosis, leading to the "amyloid-beta accumulation cycle," in the development of AD pathology. CDK inhibitors tested on animal models of AD showed symptomatic relief, corroborating the notion that the cell cycle and mitosis are targets of AD drug research and development. With further support, existing cell cycle and mitosis-targeting drugs, many of which are clinically used as cancer drugs, may be successfully repurposed as AD drugs in the near future.

CO N FLI C T S O F I NTE R E S T
No conflicts of interest declared.