Zebrafish is known to have an average lifespan of 42 months (Gerhard et al., 2002; Kishi, 2004), while the lifespan of mice typically ranges from 25 to 26 months (Fahlström et al., 2011) (Table 1). Interestingly, even genetically modified mice designed for longevity live as long as the wild-type zebrafish (Bartke et al., 2001). It is known that there is a positive correlation between body weight and a lifespan in vertebrates (Kuparinen et al., 2023), suggesting that zebrafish, despite their relatively small size, is inherently long-lived species compared to other animals including mice. In addition, humans tend to live longer among mammals of equivalent body weight, while mice demonstrate a relatively shorter lifespan (Roser et al., 2013; Kowalczyk et al., 2020) (Table 1). These observations suggest that zebrafish offers a unique opportunity to replicate aging-related phenotypes that are not observable in shorter-lived species. This aspect of zebrafish biology underscores the potential of using them to explore complex aging processes and their implications for human health, offering insights into longevity and mechanisms underlying age-related diseases.

In the context of age-related neurological diseases, the efficiency of clearance systems in the brain plays a pivotal role in maintaining brain homeostasis. As organisms age, the efficiency of autophagy and the ubiquitin-proteasome system declines, leading to a disruption in proteostasis and subsequent accumulation of abnormal proteins in the brain (Kaushik and Cuervo, 2015). To clear these substances, the brain has several clearance mechanisms including the glymphatic system, which can drain these substances along cerebral small vessels (Nedergaard, 2013). Benveniste et al. (2018) suggest significant differences in the clearance efficiency of the glymphatic system between humans and mice (Table 1). This assertion is supported by the several studies investigating the half-life of amyloid-beta (Aβ), mainly cleared by the glymphatic system (Wang et al., 2023). In humans with Alzheimer’s disease, the half-life of Aβ is around 13 h, whereas in Alzheimer’s disease model mice, even aged individuals show a half-life of merely 2–4 h, indicating that the clearance of Aβ from the brain in mice is at least three times faster than that in humans (Cirrito et al., 2003; Barten et al., 2005; Mawuenyega et al., 2010; Qosa et al., 2014). Research utilizing zebrafish larvae, in which Aβ was injected into the brain, and the amount of Aβ was measured between 5 h-post-injection and 24 h-post-injection, showed only a 40% reduction in Aβ levels, indicating the half-life of Aβ is over 19 h in zebrafish larvae (Jeong et al., 2021). These studies suggest that zebrafish possesses a capability for brain clearance that is similar to humans, while mice exhibit a significantly higher capacity compared to the other two species (Table 1).

One of the possible causes of this difference is variations in heart rate, which directly impact the efficiency of the glymphatic system. It has been demonstrated that the glymphatic system is stimulated upon artificially elevating the heart rate in mice through the administration of dobutamine (Iliff et al., 2013). The resting heart rate of mice is around 500 bpm, while it is around 60–70 bpm in humans (Wang et al., 2017; Benveniste et al., 2018) (Table 1). Therefore, the clearance capacity in the brain in humans might be much weaker than in mice. Notably, the resting heart rate of adult zebrafish is around 120 bpm (Table 1), suggesting that zebrafish may possess brain clearance mechanisms closer to humans than to mice (Mousavi and Patil, 2020). Moreover, it is implied that dysregulation of cerebral blood flow affects glymphatic system (Sepehrinezhad et al., 2023). The cerebral blood flow is decreased with aging in humans and zebrafish, whereas this does not change in mice (Pantano et al., 1984; Wei et al., 2020; Mizoguchi et al., 2023) (Table 1). This suggests that the efficiency of glymphatic system is decreased with aging in humans and zebrafish, but less in mice.

These could imply that neurovascular aging related to neurodegenerative diseases may occur earlier in humans and zebrafish compared to mice. Thus, considering the difference in brain clearance capacity points toward the potential of zebrafish as a more representative model for studying human cerebrovascular aging and clearance mechanisms in the context of aging-related neurodegenerative diseases.

Alzheimer’s disease (AD) is the most common type of dementia. Two main pathological hallmarks of AD are Aβ plaques and neurofibrillary tangles (NFT) (Blennow et al., 2006). The Aβ plaques are formed by deposition of Aβ protein, and the NFT is formed by intracellular tau protein hyperphosphorylation, which is induced by Aβ (Blennow et al., 2006). There are several AD risk genes such as APP, PSEN1, and PSEN2. The single missense mutation of these genes causes the AD pathology in human (Bagyinszky et al., 2016).

Transgenic mouse models have significantly contributed to our understanding of Alzheimer’s disease (AD), with multiple types of transgenic mice of AD overexpressing mutant forms of AD risk genes (Sanchez-Varo et al., 2022; Yokoyama et al., 2022). These models exhibit key features of AD pathology, including Aβ plaques and cognitive decline. However, they have limitations, including artificial temporal or spatial expression patterns in transgenic overexpression systems, leading to complex outcomes that may not accurately represent the human condition. For example, as reported by Jankowsky et al. (2005), analyzing cognitive behavior was difficult due to their severe hyperactivity which is not a human AD symptom, using transgenic mice expressing chimeric mouse/human APP Swedish/Indiana (carrying KM570, 571NL, and V617F mutation). The authors described that hyperactivity, which is not observed in human, might be caused by neuronal alterations due to transgene expression during early postnatal development (Jankowsky et al., 2005). Moreover, most transgenic mice, such as single transgenic mice (carrying mutant APP) or double transgenic mice (carrying both mutant APP and mutant PSEN1), did not show the formation of NFT, despite exhibiting cognitive decline (Metaxas and Kempf, 2016; Drummond and Wisniewski, 2017; Sasaguri et al., 2017). These limitations highlight the need for alternative AD models, which substitute for transgenic models.

To overcome the issues associated with transgenic models, researchers have attempted to create models using knock-in techniques. These techniques aim to mimic wild-type expression levels and patterns more closely. However, the expected Aβ plaque deposition was not detected in APP Swedish (carrying KM670, 671NL mutation) or London (carrying V717I mutation), and memory impairment was not observed in the knock-in model that carries a single APP mutation (carrying V642I) (Kawasumi et al., 2004; Köhler et al., 2005). Subsequent efforts led to the development of APP knock-in mice models incorporating multiple familial Alzheimer’s mutations. These models include the Swedish (NL), Beyreuther/Iberian (F), and Arctic (G) mutations. The APP NL-G-F mice, which harbor all three mutations, began to develop Aβ plaques at two months and showed memory impairment from six months (Saito et al., 2014). However, further research revealed that in some cases, APP NL-G-F mice did not exhibit the expected decline in memory abilities [refer to the discussion in Sakakibara et al. (2019)]. These suggest that introducing mutated forms of APP through knock-in techniques may not fully replicate the symptoms of AD. In terms of PSEN1/2, there are a few studies on familial mutant PSEN1 knock-in mouse models. These mouse models carry a single familial mutation such as L435F, I213T or R278I, and have shown Aβ plaques but exhibit no or mild memory impairment (Lalonde and Strazielle, 2005; Saito et al., 2011; Xia et al., 2015). Therefore, it is difficult to conclude the PSEN1 knock-in AD mouse models fully replicate the AD pathology.

Cerebral Small Vessel Diseases (CSVDs) are the collective term for diseases that affect the cerebral small vessels. Damage to small vessels lead to lesions in subcortical structures like lacunar infarcts, white matter lesions, large hemorrhages, and microbleeds, leading to dementia (Pantoni, 2010). The progression of CSVDs is highly age-associated (Chung et al., 2023). Characteristic pathologies of CSVDs include enlarged perivascular spaces and formation of abluminal protein deposits (Pantoni, 2010). Recent studies have suggested that the glymphatic system plays a pivotal role in the pathophysiology of CSVDs (Mestre et al., 2017; Benveniste and Nedergaard, 2022). The mouse models of monogenic CSVDs such as CADASIL, CARASIL, Fabry disease, and RVCL, are discussed in the following section.

Cerebral autosomal dominant arteriopathy with subcortical infarct and leukoencephalopathy (CADASIL) is a prototypical CSVD caused by mutations in the NOTCH3 gene (Chabriat et al., 2009; André, 2010). CADASIL is characterized by the accumulation of granular osmiophilic material (GOM) and the extracellular domain of NOTCH3 in the vascular walls, leading to the loss of perivascular cell (vascular smooth muscle cell and pericyte), vascular dysfunction, recurrent lacunar infarcts, cognitive impairments, depressive symptoms, and motor deficits (Kalaria et al., 2004; Chabriat et al., 2009; Dziewulska and Lewandowska, 2012). Despite its clinical importance, the precise pathogenesis of CADASIL remains elusive, and there is currently no effective treatment, emphasizing the need for animal models to better understand and address this condition.

Several transgenic mouse models have been developed that can partially mimic CADASIL pathology. Transgenic mice with the rat Notch3 R169C mutation exhibit GOM lesions, Notch3 accumulation, pericyte loss, and memory impairment (Joutel et al., 2010; Ghosh et al., 2015; Ehret et al., 2021). Mice with the human NOTCH3 R90C mutation also showed GOM lesions, vascular dysfunction, and memory deficits (Ruchoux et al., 2003; Lacombe et al., 2005; Liu et al., 2015). However, Ruchoux et al. (2003) showed that these mice did not exhibit significant brain parenchyma damage. These suggest that the observed memory deficits might arise from mechanisms different from those in human CADASIL. Another transgenic model expressing the human NOTCH3 R182C mutation has been established, which develops GOM lesions, but does not show the white matter lesions, changes in cerebral blood flow, or memory impairment seen in human patients (Rutten et al., 2015; Gravesteijn et al., 2020). To more accurately mimic the pathological conditions of human CADASIL, knock-in models are increasingly developed. There are two types of knock-in CADASIL mouse models; one is Notch3 R170C (corresponding to human R169C) knock-in mice, and another is Notch3 R142C (corresponding to human R141C) knock-in mice (Lundkvist et al., 2005; Wallays et al., 2011). However, it is important to note that, to our knowledge, none of these CADASIL knock-in mouse models have yet exhibited memory impairments.

Cerebral autosomal recessive arteriopathy with subcortical infarcts and leukoencephalopathy (CARASIL) is a hereditary disease caused by loss-of-function mutations in the Htra1 gene, characterized by baldness, strokes, white matter lesions, and early-onset dementia (Tikka et al., 2014). The abnormal accumulation of extracellular matrix proteins and TGF-β1, which are degraded by Htra1 was observed around small cerebral arteries in CARASIL patients (Hara et al., 2009). While aged Htra1 knockout mice showed the abnormal protein accumulation in cerebral arteries, they have not successfully replicated white matter lesions, strokes, or smooth muscle cell loss seen in the human condition (Beaufort et al., 2014; Kato et al., 2021).

Fabry disease results from mutations in the gene encoding α-galactosidase A (α-GalA), a lysosomal hydrolase enzyme (Germain, 2010). This leads to decreased enzyme activity and the accumulation of its substrate, globotriaosylceramide (GL-3), within the lysosomes of various organs, including blood vessels, kidneys, heart, and dorsal root ganglia (Choi, 2015). The primary symptoms of Fabry disease in humans are burning pain, autonomic dysfunctions, posterior circulation stroke, cognitive impairment, and depression (Bolsover et al., 2014; Choi, 2015). α-GalA knockout mice have been developed to study Fabry disease. While these mice exhibit the accumulation of GL-3 (Ohshima et al., 1997; Bangari et al., 2015), they did not show depressive-like behavior or learning and memory deficits (Hofmann et al., 2017).

Retinal vasculopathy with cerebral leukodystrophy (RVCL) is caused by mutations in a 3′-5′ DNA exonuclease TREX1 (Richards et al., 2007). The primary symptoms in human RVCL patients include activation of immune system, leukoencephalopathy, lacunar infarcts, retinopathy, nephropathy, and migraines (Schuh et al., 2015; Søndergaard et al., 2017). While the pathomechanism remains unknown, vascular basement membranes were found to be thicker and multi-layered (Søndergaard et al., 2017). This suggests that the clearance system in the brain might be impaired in RVCL patients. Frame-shift mutant TREX1 knock-in mice have been developed as RVCL models. Although these mice replicated activation of immune system, they did not exhibit key manifestations such as retinopathy and neurological symptoms (Sakai et al., 2017).

Parkinson’s disease (PD) is a neurodegenerative disorder characterized by the accumulation of α-Synuclein in the neurons of the substantia nigra and striatum, and damage to dopaminergic neurons (Kalia and Lang, 2015; Balestrino and Schapira, 2020). PD primarily manifests as motor dysfunction, and approximately 40% of PD patients suffer from dementia (Cummings, 1988). The exact cause of dopaminergic neuron impairment in PD remains unclear, highlighting the importance of animal model research in elucidating these mechanisms. Genetic mouse models replicating PD often utilize genes considered to be causative, such as LRRK2, PRKN, and PINK1. Transgenic mouse models of these genes consistently exhibit motor dysfunction, and most of these models show age-related cognitive impairments (Magen and Chesselet, 2011; Magen et al., 2012; Pischedda et al., 2021; Dovonou et al., 2023). Several knock-in mouse models harboring mutations of pathogenic LRRK2 variants have also been developed. These models typically replicate the characteristic neuronal damage and motor dysfunctions observed in PD (Chang et al., 2022; Dovonou et al., 2023). It is reported that LRRK2 G2019S knock-in mice successfully mimic the neuronal pathology in striatum and cognitive impairments, further contributing to our understanding of the broader impact of PD on cognitive functions (Hussein et al., 2022).

Huntington’s disease (HD) is a disorder resulting from abnormal amplification of CAG repeats in the Htt gene, leading to the formation of insoluble aggregates and subsequent neuronal loss (Walker, 2007; Ross and Tabrizi, 2011). HD is characterized by involuntary, dance-like movements of the limbs, known as chorea, cognitive impairments, and psychiatric symptoms (Walker, 2007). Studies using transgenic mice that overexpress Htt with amplified CAG repeats have demonstrated the manifestation of motor and cognitive impairments (Lione et al., 1999; Lüesse et al., 2001; Giralt et al., 2011; Kaye et al., 2021). These models have been instrumental in mirroring the symptomatology of HD, providing valuable insights into the disease mechanisms and progression. Similar to transgenic models, knock-in mice carrying HD-like mutations in the Htt gene consistently exhibit stable motor and cognitive impairments (Simmons et al., 2009; Giralt et al., 2012; Menalled et al., 2012; Yhnell et al., 2016). This suggests that, similar to PD, HD is a disorder where phenotypic traits are relatively easier to replicate in mouse models.