Can accelerated ageing models inform us on age‐related tauopathies?

Abstract Ageing is the greatest risk factor of late‐onset neurodegenerative diseases. In the realm of sporadic tauopathies, modelling the process of biological ageing in experimental animals forms the foundation of searching for the molecular origin of pathogenic tau and developing potential therapeutic interventions. Although prior research into transgenic tau models offers valuable lessons for studying how tau mutations and overexpression can drive tau pathologies, the underlying mechanisms by which ageing leads to abnormal tau accumulation remains poorly understood. Mutations associated with human progeroid syndromes have been proposed to be able to mimic an aged environment in animal models. Here, we summarise recent attempts in modelling ageing in relation to tauopathies using animal models that carry mutations associated with human progeroid syndromes, or genetic elements unrelated to human progeroid syndromes, or have exceptional natural lifespans, or a remarkable resistance to ageing‐related disorders.

further extending the human healthspan require substantial breakthroughs to understand the biology of ageing at an organismal, organ, cellular and molecular level.
One major drawback in in vivo ageing research is the duration and the cost of the experiments due to the requirement of keeping the animals for most of their lifespan (over 2 years for a typical mouse model) to be able to study the ageing process. Fuelled by the discovery of ageing-related features in a variety of accelerated ageing (progeroid) syndromes, the past two decades has seen both advancement in understanding the underlying molecular mechanism behind these rare diseases and remarkable efforts in utilising progeroid pathological ageing to unveil the mystery of physiological ageing and ageing-related disorders. In the realm of neurodegenerative diseases, some strides in understanding have been helped by the analysis of human patient samples. But advances in this period have also arisen from the emergence of new animal/tissue culture models that age faster while recapitulating key biochemical ageingrelated features such as senescence, nuclear envelope impairment and elevated levels of genomic damage-to name a few. Here, we recap recent attempts in utilising model organisms, which exhibit an accelerated or decelerated rate of ageing, to study neurodegenerative disorders that are predominantly driven by pathological changes of the tau protein (tauopathies). To understand the advantages and disadvantages of using fast/slow ageing models, it is necessary to briefly consider the ageing-dependent nature of tauopathies and widely used animal models in tauopathy research.

| Microtubule-associated protein tau and tauopathies
Tau proteins are a group of low molecular weight microtubuleassociated proteins, of which there are six isoforms produced by alternative splicing of the MAPT gene, located on chromosome 17q21 (Goedert et al., 1988(Goedert et al., , 1989. A highly conserved protein, tau is expressed across many animal species, including human, mouse, rat and bovine species (Adolf, 1989;Takuma et al., 2003). Transcripts of tau are found primarily, though not exclusively, within the central nervous system, particularly in neurons. As would be expected of microtubule-associated proteins (MAPs), the primary physiological function of tau is binding to microtubules, stabilising and mediating their assembly and modulating vesicle/organelle transport along microtubules (Binder et al., 1985;Weingarten et al., 1975).
The term 'tauopathy' refers to a group of neurodegenerative diseases pathologically characterised by the accumulation of insoluble intracellular tau inclusions in the central nervous system, with symptoms of dementia and/or parkinsonism (Kovacs, 2015). Some examples of tauopathy include Alzheimer's disease (AD), argyrophilic grain disease (AGD), globular glial tauopathy, progressive supranuclear palsy (PSP), Pick's disease (PiD) and corticobasal degeneration (CBD; Review by [Götz et al., 2019]). Based upon the primary contributor of pathology, tauopathies can be classified as primary tauopathies, in which tau aggregation plays a prominent role in disease pathogenesis, and secondary tauopathies, where disease progression is mainly propelled by abnormalities of other proteins. Based upon its aetiology, tauopathies can be divided into two subtypes: familial and sporadic. On the other hand, sporadic tauopathies are not hereditary and have delayed onset of symptoms (>65 years in most sporadic AD cases [Rabinovici, 2019]). Not linked to tau genetic mutations or any specific environmental factors, sporadic tauopathies occur in a seemingly unpredictable manner, suggesting a complex and multifactorial nature of its pathogenesis.
Among all known risk factors for sporadic late-onset tauopathies, chronological age is strongly correlated with the prevalence of the disease. For instance, a recent epidemiological analysis showed a positive correlation between age and prevalence of progressive supranuclear palsy, a primary tauopathy (Viscidi et al., 2021;Figure 1a). The same correlation has also been established in Alzheimer's disease, a secondary tauopathy. An early longitudinal study conducted in the United States demonstrated a strong positive correlation between age and the incidence rate of Alzheimer's disease in both male and female participants (Kawas et al., 2000;Figure 1b). In 2014, a meta-analysis of the prevalence of dementia (62% of which was Alzheimer's disease) in Western Europe illustrated the same correlation (Prince et al., 2014;Figure 1c). Notably, ageing is also an important risk factor in neurodegenerative diseases in which pathogenesis is driven by proteins other than tau.
For example, advanced age also appears to correlate with high incidence of parkinsonism (Figure 1d), a neurological condition primarily caused by defects in α-synuclein (Savica et al., 2013). The sex bias in dementia related to tau or synuclein may be explained by sex hormones (Rajsombath et al., 2019;Sundermann et al., 2020;Yang et al., 2022), or sex-related differences in brain connectome and specific genes/transcription factors (López-Cerdán et al., 2022;Shokouhi et al., 2020).

| Tauopathy rodent models
Model organisms form the foundations of sporadic tauopathy research, as they offer physiologically relevant platforms to study the influences of genetic, nutritional and environmental factors, and observe pathological changes at different time points of disease progression. Importantly, using model organisms with short natural lifespan allows for investigation of late-onset disorders in a timely manner. In the realm of tauopathy research, more than 40 animal models have been developed over the past two decades (ALZFORUM-Animal Models, n.d.). The summary of single knockin transgenic tau rodent models captured in Table 1 highlights two limitations in these models. First, most of these animal models frequently rely on an overexpression of human tau protein (huTau) with disease-related mutations, for example, P301L (Lewis et al., 2000) and R406W (Frost et al., 2016) in FTLD-MAPT (frontotemporal lobar degeneration associated with MAPT mutations). In consequence, they recapitulate processes occurring in genetic tauopathies, but not necessarily the events of sporadic tauopathies. Second, overexpression of mutant tau significantly accelerates pathogenesis and therefore does not necessarily reflect the ageing-dependent nature of idiopathic tauopathies. Indeed, only two out of 23 rodent models in Table 1 have a late-onset cognitive impairment (>12 months in an average lifespan of 2 years for laboratory mice (Dutta & Sengupta, 2016) and 3 years for laboratory rats [Quinn, 2005;Sengupta, 2013]). As mimicking an aged cellular/physiological environment is an important practical and theoretical consideration that facilitates research into the origin of sporadic tauopathies, it is necessary to seek ways to combine the knowledge we inherited from studies using TgTau animal models and the lessons we learned from accelerated/decelerated ageing animal models.

| Accelerated ageing models in tauopathy research
In the field of cellular ageing and neurodegeneration, approaches to study the effects of ageing include the following: first, use of genetically modified animals that express mutated proteins which have been associated with human premature ageing syndromes (also called progeroid syndromes-PS). PS are rare congenital/genetic disorders that recapitulate some pathological features of normal ageing in an accelerated manner and thus provide potential insights into the natural ageing process. Most human PS are caused by either defects in the nuclear lamina, hence, the name 'laminopathies', or deficiencies in the DNA repair machineries. Table 2 offers a non-exclusive summary of human PS. There have been attempts to exploit defects in these two cellular processes to mimic ageing in tauopathy model organisms; their benefits and limitations are summarised in Table 3.
Second, the use of natural genetic variants that occur in animal lines that display phenotypes of premature ageing-these would include animals harbouring the polygenic SAMP8 trait or being homozygous for null alleles of the Klotho gene (Akiguchi et al., 2017; Kuro-o TA B L E 1 Transgenic single knock-in tau rodent models developed in the last two decades (ALZFORUM-Animal Models, n.d.

| Transgenic progeroid model type 1: laminopathy models
Laminopathies are a diverse group of diseases caused by genetic malfunctions of proteins associated with the nuclear lamina-a protein meshwork coating the inner surface of the nuclear envelope and forming a part of the nucleoskeleton (Shin & Worman, 2021).
The nuclear lamina is critical for maintaining the structure of the nucleus and the organisation of the chromatin. As the most common form of laminopathy, Hutchinson Gilford progeria syndrome (HGPS) is caused by an autosomal dominant mutation in LMNA, the gene encoding for lamin A and C proteins. The mutation leads to the production of a truncated form of lamin A (termed 'progerin'; Ullrich & Gordon, 2015). Healthy lamin A precursor undergoes a 4-step post-translational process to produce mature lamin A: normal wild-type pre-lamin A undergoes farnesylation at its C-terminal CaaX box by farnesyltransferase (FTase), C-terminal cleavage after the cysteine residue, methylation by protein-S-isoprenylcysteine carboxyl methyltransferase and upstream cleavage by ZMPSTE24 (Davies et al., 2009). In HGPS, due to its permanent farnesylation state, progerin is constitutively retained in the nuclear lamina, acting as a dominant negative and leading to multiple defects relevant to cellular ageing. These include altered mechanical properties of the nuclei and increased nuclear stiffness (Goldman et al., 2004), altered chromatin organisation (Shevelyov & Ulianov, 2019), increased unrepaired DNA double-stranded breaks , disrupted nucleocytoplasmic transport (Kelley et al., 2011;Snow et al., 2013) and increased protein translation rate (Buchwalter & Hetzer, 2017).

Models that recapitulate both tauopathy and laminopathy features
have been built upon mutations in LMNA.
The first attempt of using progerin to accelerate ageing in a mouse model was accomplished in 2012. In knock-in mice that express exclusively lamin A or exclusively progerin, it became apparent that lamin A, prelamin A and progerin are absent in neuronal and glial cells, suggesting that lamin A and its derivatives are not likely implicated in brain ageing (Jung et al., 2012). A neural-specific microRNA, miR9, has been identified to suppress lamin A-and therefore progerin expression in the central nervous system; this finding was confirmed by an independent study performed on induced pluripotent stem cells derived from HGPS patient cells (Nissan et al., 2012). This is in agreement with the absence of Alzheimer's-like pathology in HGPS post-mortem tissues and with the clinical observations that HGPS patients do not develop neurological symptoms such as cognitive deterioration (Ullrich & Gordon, 2015). In Baek et al., 2015, the authors established an inducible HGPS transgenic mouse model that specifically expresses progerin in bone, skin, heart and neurons. This system showed that forced long-term expression progerin in the hippocampal neurons resulted in severe nuclear distortions (e.g., nuclear blebbing and invagination) in cells from all tissues, without any apparent neuropathological defects, such as protein aggregation, adult neurogenesis ability, transcriptomic profile alteration (including lamin B1) and behavioural changes (Baek et al., 2015). It therefore suggests that progerin-induced nuclear dysmorphology in neurons does not necessarily correlate with loss of neuronal functions and dementia-related pathologies. The mechanisms behind this neuronal protection remain unknown. It is worth mentioning that progerin has also been used to assist in modelling pathological phenotypes of late-onset Parkinson's disease in human iPSC-derived neurons (Miller et al., 2013), where expression of progerin in PD iPSC-derived dopamine neurons induced ageing-related markers and disease phenotypes, such as dendrite degeneration, Lewy-body-precursor inclusions and dysregulated protein degradation mechanisms.
Recently, the role of the nuclear lamina in Alzheimer's disease has attracted more attention (Gil et al., 2021). Analyses of AD postmortem brain tissue revealed that lamin A mRNA levels increased at  (Gil et al., 2020). Given these discoveries, the notion that lamin A and its derivatives are absent in neuronal cells and are therefore not involved in the function of the central nervous system deserves renewed scrutiny.
Apart from expressing progerin, ZMPSTE24, as an enzyme essential in lamin A processing, has also been exploited to generate progeroid ageing features in experimental model systems. ZMPSTE24 is an integral membrane zinc metallopeptidase that endoproteolytically cleaves the C-terminal region of carboxymethylated prelamin A and produces mature lamin A (Quigley et al., 2013). Twenty human ZMPSTE24 mutations that reduce the enzyme activity have been identified to associate with three disease categories of increasing severity: mandibuloacral dysplasia type B, severe progeria (atypical 'HGPS') and restrictive dermopathy (Barrowman et al., 2012); the crucial involvement of ZMPSTE24 in laminopathies was further emphasised by a novel mouse model expressing non-cleavable prelamin A (Wang, Shilagardi, et al., 2022). Despite its availability (Bergo et al., 2002;Varela et al., 2005), to the best of our knowledge, ZMPSTE24-deficient mice have not been directly used to study brain ageing in neurodegenerative diseases. Nevertheless, neurological abnormalities, such as microcephalia (de Carlos et al., 2008) and oesophageal achalasia (Yang et al., 2015), have been observed in ZMPSTE24-deficient mice, suggesting potential functions of ZMPSTE24 and lamin A in the nervous tissues.
It is noteworthy that multiple recent evidence has highlighted the involvement of other nuclear lamina proteins in tauopathies. In a TgTau Drosophila model, it has been demonstrated that a pathogenic mutation of tau (R406W) alters the arrangement of B-type lamins, affecting maintenance of genomic architecture, cell cycle regulation and survival of adult neurons (Frost et al., 2016). Analysis of post-mortem human FTLD-MAPT cortex revealed a high incidence of nuclear deformation, indicating that tau mediates nuclear membrane dysfunction (Paonessa et al., 2019). Interestingly, two independent tissue culture-based CRISPRi screening both pointed out potential involvement of BANF1 (Barrier-to-autointegration factor 1) in tau aggregation (Koss et al., 2022;Polanco et al., 2022). As loss of BANF1 has significant implications in nuclear envelope integrity in Subtype-dependent (see Table 3) cell loss and polyploidisation (reviewed by Schumacher et al., 2021).
Unsurprisingly, reduced genomic stability has been a widely reported feature in Alzheimer's disease (reviewed by Lin et al., 2020).
Immunostaining for γH2AX, a phosphorylated form of H2AX that is widely used as a marker of double-stranded DNA breaks (DSBs), showed that DSBs accumulate in neurons and astrocytes in the hippocampus and frontal cortex of AD patients during the progression of neurodegeneration (Shanbhag et al., 2019). Consistent with these observations, in vitro studies with primary mouse cortical neurons revealed that non-phosphorylated tau accumulates perinuclearly upon DSBs formation, followed by accumulation of phosphorylated tau immunoreactive to AT8 antibody (Asada-Utsugi et al., 2022). In the field of pathological ageing, deficiencies in the DNA repair machineries result in many PS such as Werner syndrome (Muftuoglu et al., 2008), Cockayne syndrome (Laugel, 2013) and xeroderma pigmentosum (Lehmann et al., 2011). The wealth of DNA repairdeficient models established for studying these progeroid syndromes offered an opportunity to explore neurodegeneration in a high DNA damage susceptibility system.
Werner syndrome (WS) is a DNA repair-related premature ageing syndrome caused by mutations in a RecQ family DNA helicase, WRN (Muftuoglu et al., 2008). In contrast to progeroid laminopathies, Werner syndrome has a delayed onset-typically recognised by the third or fourth decades of life and is therefore sometimes TA B L E 3 Benefits and limitations of specific progeric models for tauopathy research.

Progeroid syndromes Models Benefits Limitations
Laminopathy models Hutchinson-Gilford progeria syndrome Progerin knock-in mice (Jung et al., 2012;Osorio et al., 2011) N/A • Progerin absent from neuronal and glial cells • No neurological phenotype • Tau-specific pathology has not been reported Inducible progerin transgenic mice (Baek et al., 2015) • Progerin specifically expressed in neurons • Severe nuclear distortion  Table 4) referred to as 'adult progeria' (Oshima et al., 2017). Thus far, neurodegeneration has not been observed in individuals with WS.
Nonetheless, though rare, non-AD senile dementia, schizophrenia (Goto, 1997), cerebrovascular disease and peripheral neuropathy (Anderson & Haas, 2003)  XPV (or POLH) and ERCC1 (Kraemer et al., 2003). Depending on the disease subtype, XP patients may have different neurological manifestations. Of note, cognitive dysfunction may develop in childhood and advance through later stages in XPA, and peripheral neuropathy has been reported in XPG patients (Anttinen et al., 2008). Table 4 summarises mouse models established for each XP subtype. In contrast to the prominent neurological phenotypes in XPA patients, Xpa −/− mice only displayed mild neurological manifestations, such as delayed neuromotor recovery and increased memory acquisition dysfunction following experimental brain trauma (Tomasevic nerve vacuolisation, brain mass reduction (Dollé et al., 2011) Progressive motor abnormalities and reduced life span: clasping of the hindlimbs, fine tremors and kyphosis, severe locomotor deficits and reduced ability to maintain balance (de Waard et al., 2010) Impaired fear conditioning and impaired water maze performance at 6 months of age (Borgesius et al., 2011) Ercc1 −/− No histopathological neurodegeneration, or of abnormal neuromuscular junctions. Observed uraemic encephalopathy (Lawrence et al., 2008) Poor coordination, ataxia and loss of visual acuity (Lawrence et al., 2008) a Double transgenic models that also include Cockayne syndrome-related mutations. et al., 2012). Interestingly, overt neuropathology and cognitive/behavioural alterations have been observed in transgenic mouse models for XPG and ERCC1 (Barnhoorn et al., 2014;Borgesius et al., 2011;de Waard et al., 2010;Lawrence et al., 2008). Despite that no protein misfolding-related pathology had been reported in XP patients or in these transgenic XP mouse models, the neurodegenerative features observed in XPG and ERCC1 models indicate that defects in nucleotide excision repair mechanism alone may lead to neurodegeneration and conformational assessment of aggregation-prone proteins (e.g., tau, Aβ and α-synuclein) may be a useful future step in XPG-and ERCC1-deficient mouse models.
Cockayne syndrome (CS) is a rare, autosomal recessively inherited genetic disorder characterised by premature ageing-like features, such as severe growth failure and cutaneous photosensitivity. Defects of two genes, CSA (ERCC8) and CSB (ERCC6), are responsible for the disease (Henning et al., 1995;Troelstra et al., 1992). In contrast to other progeroid syndromes associated with DNA repair deficiencies, CS is a degenerative disorder and has more profound manifestations in the central nervous system, including mental retardation, microcephaly, retinal atrophy and progressive neurodegeneration (Spitz et al., 2021).
To study neurological impairments caused by CSA and CSB deficiency, and astrocyte activation in cerebella (Xu et al., 2019). Beyond rodent models, Csb-1 deficiency in C. elegans also associates with locomotion dysfunction and aged-dependent loss of sensitivity to mechanosensory stimuli, possibly through its effects on mitochondrial activity (Lopes et al., 2020). It is noteworthy that protein misfolding and loss of proteostasis have been reported in CS patient-derived fibroblasts (Alupei et al., 2018;Qiang et al., 2021). However, it is currently unclear if these proteostatic defects can be mirrored in CSA or CSB animal models, and whether this may accelerate the progression of proteinopathies in the central nervous system, such as tau aggregation.
Ataxia telangiectasia (A-T) is a neuromotor dysfunction neurodegenerative disorder of childhood caused by the disruption of gene ATM (Ataxia-telangiectasia mutated serine/threonine kinase).
Clinically, A-T patients represent a wide variety of symptoms including progressive cerebellar ataxia, oculocutaneous telangiectasia, variable immunodeficiency, radiosensitivity, susceptibility to malignancies and increased metabolic diseases (Amirifar et al., 2019). In an ATM knockout Drosophila model, it has been demonstrated that reducing the function of ATM significantly enhances tau-induced neuronal apoptosis and exacerbates tau neurotoxicity (Khurana et al., 2012). To further increase the genotoxic stress, a novel double mutant mouse model, Atm R35X/R35X /Aptx −/− , has been generated; it not only develops a progressively severe ataxic phenotype but also exhibits significantly perturbed cerebellar Purkinje neurons (Perez et al., 2021). Of note, examination of Alzheimer's disease post-mortem brain samples revealed an elevated level of ATM compared to the age-matched control (Katsel et al., 2013), suggesting that activation of the DNA damage checkpoint is a shared feature between A-T and AD.

| Non-progeroid models of accelerated ageing
Apart from the aforementioned transgenic animal models where the progeric alleles are artificially introduced to the genome to mimic ageing, there are other models to study accelerated ageing of the brain. Here, we briefly review Klotho-deficient mice, SAMP8 mice and African Turquoise Killifish. growth factor receptor binding activity [Gaudet et al., 2011]) and has a wide variety of cellular implications (e.g., mitochondrial dysfunction [Sahu et al., 2018]). In the context of neurodegenerative disorders, the level of Klotho in the central nervous system is generally negatively correlated with the severity of proteinopathies. For instance, a longitudinal study in 2020 suggested that increased serum level of Klotho (in Klotho-VS heterozygous individuals) is associated with reduced AD risk in APOE4 carriers from 60 to 80 years of age (Belloy et al., 2020). It was later reported that Klotho-VS heterozygosity was associated with a lower cross-sectional and longitudinal increase in amyloid-related tau pathology and tau-related memory defects (Neitzel et al., 2021). Echoed with these findings were two recent clinic observations suggesting that higher levels of CSF Klotho were associated with lower CSF Aβ42 and tau burden (total tau and phosphorylated tau; Driscoll et al., 2021;Grøntvedt et al., 2022). In the experimental realm, Klotho-deficient mice display overt phenotypes compared to their wild-type littermates, such as hypogonadism, premature thymic involution, ectopic calcification, impaired bone mineralisation, skin atrophy, hearing loss and neurodegeneration Klotho transgenic mice, demonstrated that elevated level of Klotho protects hAPP mice against premature mortality and cognitive impairments (Dubal et al., 2015). These studies altogether suggest a neural-protective role of Klotho and this mechanism deserves further investigation in simplified tau aggregation reporter platforms, such as in the HEK293 tauRD-YFP cell line (Sanders et al., 2014), and animal models carrying pathogenic tau mutations.

Klotho-deficient (KL
In pursuit of polygenic traits, senescence-accelerated mouse prone (SAMP) substrains were established by Kyoto University through phenotypic selection from AKR/J breeding colonies.
SAMP8 is one of the nine major SAMP substrains and shows features of rapid ageing. Functionally, SAMP8 mice show impairment of memory, deteriorations in learning ability (Miyamoto et al., 1986), emotional disorders (Miyamoto et al., 1992), altered circadian cycle and water consumption (Miyamoto, 1997). Pathologically, brain stem spongy degeneration, blood-brain barrier dysfunction and loss of cholinergic neurons are major anatomical changes in the central nervous system. Despite the absence of amyloid plaques and neurofibrillary tangles, SAMP8 mice show many cellular and molecular characteristics that mimic physiological brain ageing, such as increase in phosphorylated tau and increase in oxidative stress, severe cellular senescence, downregulation of glucose metabolism (Akiguchi et al., 2017) and age-dependent neuroinflammations (Fernández et al., 2021). Moreover, amyloidβ granules are present in the stratum radiatum of the CA1 region of hippocampus in SAMP8 mice, which contain Aβ42, Aβ40, tau, MAP2 and α-synuclein peptides (Manich et al., 2011). An attempt to combine the characteristics of the APP/PS1 transgenic mouse model with a senescence-accelerated background of SAMP8 mice revealed cognitive abnormalities, amyloid plaque formation and other AD markers (e.g., neuroinflammation, hyperphosphorylation of tau), but not neurofibrillary tangles (Porquet et al., 2015). To the best of our knowledge, a SAMP8-based mouse model that expresses human tau has not been established but it may be a useful model.
African Turquoise Killifish (Nothobranchius furzeri) is a species of small freshwater fish native to southeast Africa, primarily found in Zimbabwe and Mozambique. Due to its small size and rapid life cycle, this fish has recently gained a resurgence of interest as an experimental model system in ageing science (Smith et al., 2017).
Compared to other species, this species of fish reaches sexual maturity in 3-4 weeks (Terzibasi et al., 2008) and has a very short lifespan (9-16 weeks for the GRZ strain, and 23-28 weeks for the MZM-0403 strain [Terzibasi et al., 2008]), 6-10 times shorter than that of zebrafish (Kishi et al., 2009). Their small size, fast reproductive cycle and easy-to-care-for nature also make them an ideal species for laboratory experiments. Furthermore, the African Turquoise Killifish genome contains 96.8% of core eukaryotic genes and has been well annotated (Valenzano et al., 2015), aggregates in the brains of aged killifish (7 months) and such aggregates can be propagated in DDX5 aggregation reporter yeast (Harel et al., 2022), suggesting that transmissible prion-like protein aggregates accumulate in the brain during vertebrate ageing.
Despite that misfolded tau was not observed in protein aggregates that accumulate in old killifish brains, this pioneering observation encouraged more thorough examination of aged African Turquoise Killifish for tau-specific pathologies. Indeed, (a) DDX5 is directly involved in exon 10 splicing of the MAPT gene (Kar et al., 2011); (b) similar to zebrafish (MacRae & Peterson, 2015), rapid reproduction and small size of African Turquoise Killifish allows it to be used in phenotype-based screening in drug discovery; (c) existing protocols for genome engineering (Harel et al., 2016;Hartmann & Englert, 2012) allows for establishment of transgenic killifish expressing human MAPT gene.

| Naked mole rat
The naked mole-rat (NMR; Heterocephalus glaber) has attracted considerable biogerontological interest for multiple reasons. First, they have extraordinary longevity. As the longest-lived rodent, the NMR has a maximal lifespan of more than 30 years in a laboratory environment (Ruby et al., 2018), 10-fold greater than the allometrically predicted figure for a mouse-sized rodent (Edrey et al., 2011).
Third, the NMR possesses unusual resistance against diseases. For instance, the NMR has a low susceptibility to spontaneous cancer and experimentally induced tumour growth (Liang et al., 2010), yet the mechanism behind this observation is currently controversial (Hadi et al., 2020;Tian et al., 2013). In the context of neurodegenerative diseases, low susceptibility of protein aggregation for tau and amyloidβ has been observed in the central nervous system of naked mole rat (Edrey et al., 2013;Orr et al., 2015). NMRs offer a unique slow-ageing system to study modulators of tau aggregation due to high sequence similarity between human tau and NMR tauprotein alignment comparison illustrates 95% overall identify and 100% identity at the functional microtubule-binding domain (Orr et al., 2015). Unlike what is observed in human, NMR tau undergoes a progressive increase in molecular weight during development, a transition from a 62/72 kDa doublet to a 72 kDa singlet at 3 weeks of age, and to an 88 kDa band later (Orr et al., 2015). Interestingly, despite that NMRs have a considerable level of total tau and phosphotau, NMR tau remains localised in the axon and does not cause pathologies (Orr et al., 2015).  (Bocklandt et al., 2011), this approach has been well validated in predicting biological age in many other species including mice (Stubbs et al., 2017), rats (Levine et al., 2020), great apes  and ruminants (Caulton et al., 2021) to name a few.
Importantly, epigenetic clocks have also been developed for specific tissues, including brain. Biological age has been shown to correlate with tau load in the normal human brain (Maroni et al., 2020), and with risk of dementia in humans (Levine et al., 2015;Wu et al., 2021), although some studies have cast doubt on this association . Given the inherent discrepancy between chronological and biological age in accelerated ageing models, being able to determine their biological age would be useful for a proper interpretation of the results. In mouse brain tissue, for which a specific epigenetic ageing clock has been developed (Coninx et al., 2020), the triple transgenic Alzheimer's Disease (3xTg-AD) mouse model showed increased biological age relative to chronological age-matched brain controls. Thus, these methylation-based 'epigenetic clocks' might be a useful tool to measuring the biological age of tissues including brain in accelerated ageing animal models of dementia.
2.5.3 | Outlook: Searching for the molecular origin of tau misfolding using animal models In recent years, animal models with experimentally induced ageing features have emerged as a promising avenue for exploring sporadic tauopathies and other age-dependent neurodegenerative diseases.
This expanding interest has been supported by the rapidly improving gene editing technologies, extensive experience in husbandry and high-quality genomic and transcriptomic sequencing data. Yet, despite increasing research on the determinants of tau aggregate formation, it is still not clear which molecular mechanisms drive the production of abnormal tau in idiopathic tauopathies. Thus far, most efforts have been put into in vitro experiments where tau aggregation is primarily driven by the presence of polyanionic inducers (Ingham et al., 2022) or specific combinations of salt ingredients (Lövestam et al., 2022)-the physiological relevance of these findings will need to be validated in human patient samples and in animal models. To the best of our knowledge, despite that many ageing model organisms have neurological manifestations, such as microgliosis in ERCC1-deficient mice (de Waard et al., 2010) and amyloidβ granules in SAMP8 mice (Manich et al., 2011), none of these canonical models displays spontaneous accumulation of tau in the central nervous system. It follows that acceleration of ageing driven by progeric mutations or polygenic traits have not yet recapitulated an environment that accurately mimics an aged brain. This obstacle thus begs the question 'is it possible to establish alternative models to study delayed and accelerated brain ageing?' Indeed, non-canonical ageing model organisms have been elegantly summarised by Holtze et al., 2021. The wide range of lifespan of these models (ranging from <1 month for C. elegans to potential immortality for S. mediterranea) may help gain a more holistic understanding of the ageing process in the nervous tissues. It should be noted that tau misfolding may give rise to structurally distinct pathogenic species in different tauopathies (reviewed by [Han et al., 2022]). A future model should thus be expected to exhibit heterogeneity of tau conformers at an advanced biological age, similar to a low-tau expression transgenic mouse model previously reported (Daude et al., 2020;Eskandari-Sedighi et al., 2017). Apart from these animal models, certain experimental manipulations have also been employed to investigate the effects of ageing on neurodegeneration, such as parabiosis (a specialised type of blood transfusion procedure that surgically combines young and aged animals to study their physiological interactions [Hu et al., 2022;Kim et al., 2020;Wang et al., 2018;Zhao et al., 2020]) and caloric restriction (Brownlow et al., 2014;Cox et al., 2019;Müller et al., 2021). While current model organisms with ageing features have not yet fully recapitulated the physiological features associated with sporadic tauopathies, the early molecular events at the headwater of idiopathic proteinopathies may become better elucidated by more thorough characterisation of the cellular environment in naturally aged brain and combined use of transgenic animal models with age-related experimental manipulations.

ACK N O WLE D G E M ENTS
The authors are grateful to Dr. Anne Janssen and Denny Yang for helpful discussions and proofreading of the article.

CO N FLI C T O F I NTE R E S T S TATE M E NT
D. L. is a co-founder of Adrestia Therapeutics and a scientific advisor for Shift Bioscience and Adrestia Therapeutics. The other authors declare that they have no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Not applicable.