Lysosomal Dysregulation in the Murine App^{NL-G-F/NL-G-F} Model of Alzheimer’s Disease

Highlights

• LAMP1 accumulates at amyloid beta plaques in App^{NL-G-F/NL-G-F} (knock-in) mouse brain.

• Lysosomal hydrolases are enriched at amyloid beta plaques in App^{NL-G-F/NL-G-F} mice.

• Some lysosomal network proteins are elevated in App^{NL-G-F/NL-G-F} cortex.

• Lysosomal network dysfunction in App^{NL-G-F/NL-G-F} mice resembles human AD.

Abstract

Lysosomal network dysfunction is a prominent feature of Alzheimer’s disease (AD). Although transgenic mouse models of AD are known to model some aspects of lysosomal network dysfunction, the lysosomal network has not yet been examined in the knock-in App^{NL-G-F/NL-G-F} mouse. We aimed to determine whether App^{NL-G-F/NL-G-F} mice exhibit disruptions to the lysosomal network in the brain. Lysosome-associated membrane protein 1 (LAMP1) and cathepsins B, L and D accumulated at amyloid beta plaques in the App^{NL-G-F/NL-G-F} mice, as occurs in human Alzheimer’s patients. The accumulation of these lysosomal proteins occurred early in the development of neuropathology, presenting at the earliest and smallest amyloid beta plaques observed. App^{NL-G-F/NL-G-F} mice also exhibited elevated activity of β-hexosaminidase and cathepsins D/E and elevated levels of selected lysosomal network proteins, namely LAMP1, cathepsin D and microtubule-associated protein light chain 3 (LC3-II) in the cerebral cortex, as determined by western blot. Elevation of cathepsin D did not change the extent of co-localisation between cathepsin D and LAMP1 in the App^{NL-G-F/NL-G-F} mice. These findings demonstrate that perturbations of the lysosomal network occur in the App^{NL-G-F/NL-G-F} mouse model, further validating its use an animal model of pre-symptomatic AD.

Introduction

It is estimated that there are 50 million people worldwide living with dementia, which is associated with a cost of US$1 trillion (Patterson, 2018). Alzheimer’s disease (AD), the most common cause of dementia, has no effective disease-altering treatments. Despite hundreds of clinical trials to evaluate treatments in humans, only four drugs, which do not alter the course of the disease, have been approved (Cummings et al., 2014, Patterson, 2018). Pathogenic mechanisms, as well as most hereditary contributions to the sporadic form of AD remain to be elucidated. It is therefore crucial that authentic animal models are generated and characterised for biomedical research, to determine the causative disease pathways.

Transgenic mice that overexpress amyloid precursor protein (APP) and/or presenilin-1 have been successful in modelling amyloid plaque deposition and accompanying cognitive deficits. These models, such as 5XFAD (Oakley et al., 2006) and TgCRND8 (Chishti et al., 2001) have been instrumental to AD research. More recently, mouse models of AD have been generated through the knock-in of humanised amyloid beta and familial AD mutations, rather than transgenic overexpression of APP and/or presenilin-1. One such model is the App^{NL-G-F/NL-G-F} mouse, which has humanised amyloid beta sequence with three additional familial mutations: Swedish (NL), Arctic (G) and Iberian (F) incorporated into the App locus. This mouse has severe amyloidosis, with onset of plaque deposition in the cerebral cortex commencing by two-months of age and spreading to subcortical regions by four-months (Saito et al., 2014), preceding behavioural changes (Masuda et al., 2016, Sakakibara et al., 2018, Whyte et al., 2018, Latif-Hernandez et al., 2019, Mehla et al., 2019). App^{NL-G-F/NL-G-F} mice also develop microgliosis and astrogliosis, (Saito et al., 2014, Mehla et al., 2019). However, a number of findings from transgenic, APP-overexpressing mouse models of AD are not replicated in knock-in models, leading Sasaguri et al. (2017) to suggest that re-examining findings obtained in transgenic mouse models in knock-in models is good practice. The App^{NL-G-F/NL-G-F} model is still being characterised, and it is important to establish which features of human AD are recapitulated in this mouse.

Progressive perturbation of the lysosomal network (comprising endo-lysosomal and autophagic pathways) is a prominent neuropathological feature of AD (Cataldo et al., 2000, Nixon and Cataldo, 2006, Peric and Annaert, 2015, Nixon, 2017, Whyte et al., 2017), while genetic variation in lysosomal network genes is associated with AD (Lambert et al., 2013, Gao et al., 2018). Examples of the manifestation of lysosomal network dysfunction in AD include enlargement of RAB5-positive endosomes, (Cataldo et al., 2000, Nixon, 2017) and altered expression and activity of lysosomal proteins (reviewed in Whyte, et al., 2017). For example, in human AD patients, cathepsin D and β-hexosaminidase A protein levels are elevated in a substantial portion of neurons from vulnerable brain regions such as pyramidal neurons in layers III and V in the neocortex and the CA2 and CA3 fields of the hippocampus, as well as a smaller portion of neurons from less severely affected regions such as Purkinje cells, striatal neurons, and neurons from the thalamus and medulla (Cataldo et al., 1994, Cataldo et al., 1995, Cataldo et al., 1996). Furthermore, extracellular amyloid plaques are associated with dystrophic neurites that are filled with lysosomal network cargo vesicles (Terry et al., 1964, Nixon et al., 2005). In humans, lysosome-associated membrane protein 1 (LAMP1) can be observed histologically at most amyloid plaques (Barrachina et al., 2006, Hassiotis et al., 2018). Lysosomal hydrolases, including cathepsins, are also observed extracellularly, in association with amyloid plaques (Bernstein et al., 1989, Cataldo and Nixon, 1990, Cataldo et al., 1991, Cataldo et al., 1996). Abundant LAMP1 staining around amyloid plaques is modelled well in transgenic mouse models of AD (Condello et al., 2011, Gowrishankar et al., 2015). However, there are some inconsistencies in the existing literature in relation to the presence of lysosomal hydrolases at amyloid plaques in AD mouse models: cathepsin D has been observed around amyloid plaques in APP_SWE/PS1-dE9 mice (Ta et al., 2013); however, Gowrishankar and colleagues (2015) reported that cathepsins B, D and L were only very weakly detected around plaques in 5xFAD mice compared with prominent staining in surrounding neurons.

The lysosomal network has not been examined in an App knock-in line of mice, such as the App^{NL-G-F/NL-G-F} model. It is important to determine whether this mouse model exhibits disruptions to the lysosomal network similar to those observed in human AD and transgenic mouse models to determine the scope of relevance for this knock-in model. Lysosomal abnormalities have been proposed as a useful criterion for evaluating how well animal models of AD recapitulate the human condition (Cataldo et al., 1994). Here, we describe evidence of lysosomal network dysfunction in App^{NL-G-F/NL-G-F} mice. The App^{NL-G-F/NL-G-F} mouse model therefore provides an additional tool to probe the lysosomal network in AD and will permit spatio-temporal investigations of lysosomal network pathology at various disease stages.