Long-term effects of amyloid-beta deposits in human iPSC-derived astrocytes

.


Introduction
Alzheimer's disease (AD) is the most common age-related neurodegenerative disorder, currently affecting approximately 50 million people worldwide (Crous-Bou et al., 2017). The main clinical feature of AD is dementia, caused by wide-spread degeneration of neurons and synapses in the cerebral cortex and subcortical regions of the brain (Serrano-Pozo et al., 2011). It is a slowly progressive disease and the first pathological changes in the brain are estimated to occur 15-20 years before the first clinical symptoms emerge (Sperling et al., 2011). The major pathological hallmarks of AD include accumulation of extracellular amyloid-beta (Aβ) as plaques and intracellular deposits of hyperphosphorylated tau as neurofibrillary tangles (reviewed in Ingelsson and Hyman, 2002). Although the staging of disease progression has been extensively documented, the actual mechanisms behind the initial protein deposits and the spreading of pathological proteins remain unclear (Braak et al., 2006). Around 95 % of AD cases are sporadic and result from a combination of different genetic and environmental risk factors (Piaceri et al., 2013). To date, the only known substantial genetic risk factor for sporadic AD is the apolipoprotein E gene (APOE) (Corder et al., 1993;Saunders et al., 1993), which encodes for a protein that is mainly expressed by astrocytes in the brain.
According to the amyloid cascade hypothesis, Aβ-accumulation is the causative agent of AD, consequently driving the formation of neurofibrillary tangles, vascular damage, inflammation, and neuronal cell loss. The Aβ-peptides, especially Aβ 42 , aggregate in the human AD brain and form insoluble fibrillary structures that deposit as plaques (Hardy and Higgins, 1992). A major concern with the amyloid cascade hypothesis is that the number of plaques does not correlate with the severity of dementia (Lue et al., 1999;McLean et al., 1999). However, compelling data indicate that smaller, soluble Aβ-aggregates are more toxic and exhibit a higher propensity for seeding (Chen et al., 2017;Kayed et al., 2009). Loss of neuronal cells and neuronal dysfunction is the apparent consequence of AD pathology; hence, most research studies have focused on how neurons cope with Aβ-aggregates. Nevertheless, other brain cells, including astrocytes, have recently emerged as important players in AD progression (Avila-Muñoz and Arias, 2014).
Astrocytes are the most abundant glial cells in the brain and are responsible for preserving homeostasis as well as many other functions, including the support to neurons, modification of synapse signaling, recycling of neurotransmitters, blood-brain barrier (BBB) regulation and glymphatic clearance (Eroglu and Barres, 2010;Sofroniew and Vinters, 2010). In the case of AD and other neurodegenerative disorders, astrocytes respond to the shift in the microenvironment by becoming reactive in a process known as astrogliosis (Li et al., 2019). In post mortem human AD brains, the degree of astrogliosis has been shown to correlate with that individual's cognitive decline (Kashon et al., 2004). Reactive astrocytes differ significantly from their naïve counterparts in morphology and gene expression (Anderson et al., 2014). In the AD brain, reactive astrocytes are mainly found around Aβ-plaques and are known to produce excessive amounts of Aβ, thereby further exacerbating AD pathology (Zhao et al., 2011). They also release proinflammatory cytokines in response to cellular stress that, if not properly regulated, could lead to chronic inflammation and cause additional damage to the surrounding tissue (Liddelow et al., 2017;Sofroniew, 2009). Apart from their possible active role in disease propagation, reactive astrocytes may further accelerate the disease course by not fulfilling their physiological tasks, such as providing metabolic and trophic support to neurons and participating in protein clearance (Thal, 2012). Astrocytes are highly phagocytic and are also involved in the clearance of Aβ from the extracellular space (Akiyama et al., 1996). Our previous data demonstrate that astrocytes, although being very effective at ingesting soluble Aβ-aggregates, have limited capacity for degrading the material, which is instead accumulated as intracellular deposits (Beretta et al., 2020;Rostami et al., 2021;Söllvander et al., 2016). The reason behind why astrocytes are such good phagocytes, but bad degraders, could be, at least partly, explained by their ability to act as antigen-presenting cells (Constantinescu et al., 2005;Gimsa et al., 2004;Rostami et al., 2020). Importantly, partial digestion of Aβ can lead to secretion and spreading of truncated peptides that are toxic to neighboring neurons (Beretta et al., 2020;Söllvander et al., 2016). Although amyloid deposits in the extracellular space have been extensively researched, the role of intracellular Aβ-inclusions remains to be evaluated.
In this study, we have investigated the effect of long-term Aβ-deposits in human induced pluripotent stem cell (hiPSC)-derived astrocytes. Our overall aim was to clarify if the intracellular Aβ-storage is harmful for the astrocytes in a way that could have consequences for surrounding cells. Accordingly, we analyzed if the astrocytes have strategies to cope with the high Aβ-load and whether their reactivity and functions were affected by the intracellular Aβ-aggregates over time.

Aβ-F-exposure of astrocytes
Astrocytes were exposed to 200 nM sonicated Aβ-F in astrocyte complete medium for two days. Control cultures received medium without Aβ-F. At day 2 the cells were thoroughly washed and then cultured in Aβ 42 -free medium, without being passaged, for another 1 or 10 weeks, for the first time point 1 (T1) or second time point 2 (T2), respectively.

Immunocytochemistry
Cells were fixed with 4 % paraformaldehyde (PFA) (#P6148, Merck) in 1× PBS for 15 min at room temperature (RT) and washed 3 times with 1× PBS. Prior to antibody incubation, the cells were permeabilized and blocked with 0.1 % Triton X-100 or saponin in 1× PBS with 5 % normal goat serum (NGS) (#S-1000, Vector Laboratories, CA, USA) for 30 min at RT. Primary antibodies (Table 1) were diluted in 0.1 % Triton X-100 or 1:200 -saponin in 1× PBS with 0.5 % NGS and the coverslips were incubated with the antibody solution for 1-2 h at RT. The coverslips were washed 3 times with 1× PBS and then incubated with secondary antibodies (Table 1) diluted in 0.1 % saponin in 1× PBS with 0.5 % NGS for 45 min at 37 • C. The coverslips were washed 3 times with 1× PBS and mounted on microscope slides using EverBrite hardset medium with DAPI (#230032, Biotium, CA, USA). Images were captured using the fluorescence microscope Zeiss Axio Observer Z1 (ZEISS microscopy, Jena, Germany).

Cell lysis
The culture medium was removed, and the cells were lysed in 150 μl ice-cold lysis buffer (20 mM Tris pH 7.5, 0.5 % Triton X-100, 0.5 % deoxycholic acid, 150 mM NaCl, 10 mM EDTA, 30 mM NaPyro) supplemented with 1× Halt protease inhibitor cocktail (#78430, Thermo Fisher Scientific). The samples were transferred to protein LoBind tubes (#0030108094, Eppendorf, Hamburg, Germany) and incubated for 30 min on ice prior to centrifugation at 10,000 ×g for 10 min at 4 • C. The supernatants were separated from the pellets, transferred to new tubes and stored at − 70 • C until analysis.

Western blot analysis
Total protein concentration of the cell lysates was measured with Pierce BCA protein kit (#23225, Thermo Fisher Scientific), according to the manufacturer's instructions. A total of 10 μg of protein was mixed with Bolt LDS Sample buffer and Sample Reducing agent (both from Thermo Fisher Scientific) and incubated for 5 min at 95 • C to denature the proteins. Samples from both time points and experimental conditions were loaded on the same Bolt 4-12 % Bis-Tris plus gel and run in Bolt MES sodium dodecyl sulfate (SDS) running buffer (both from Thermo Fisher Scientific) for 25 min at 200 V. PageRuler™ Plus Prestained Protein Ladder, 10 to 250 kDa (#26619, Thermo Fisher Scientific) was used for visualization of gel migration, protein size and orientation. Transfer to a PVDF membrane was performed for 1 h at 20 V in Bolt transfer buffer containing 10 % methanol and 0.1 % Bolt antioxidant (Thermo Fisher Scientific). Blocking of the membrane was performed in 5 % Blotting-Grade Blocker (#1706404, Bio-Rad Laboratories, CA, USA) in 0.1 % tris-buffered saline-Tween (TBS-T) for 1 h on shake at RT, prior to overnight incubation with primary antibodies (Table 1) at 4 • C. Following 20-min washes in TBS-T, the membrane was incubated with horseradish peroxidase-conjugated (HRP) secondary goat anti-rabbit (1:1000, Pierce) and goat anti-chicken (1:10,000) antibodies in 5 % Blotting-Grade Blocker in 0.1 % TBS-T for 1 h on shake at RT. Development of the membrane was performed with enhanced chemiluminescence (ECL, GE Healthcare, IL, USA) by using a ChemiDoc XRS with Image Lab Software to visualize the intensity of the immunoreactive bands (Bio-Rad Laboratories).

Quantification of western blot membranes
The intensity of the detected immunoreactive bands was measured using the Image Lab software. Each band was normalized to the total protein of the respective lane, using the No-Stain™ protein labeling reagent (#A44717, Thermo Fisher Scientific). Briefly, after the proteins were transferred to the PVDF membranes, the membranes were washed in ultrapure water and incubated for 10 min in No-Stain™ protein labeling solution. After a second washing step, the membranes were imaged using a ChemiDoc XRS and the acquired images were used for normalization of each immunoreactive band, detected after immunoblotting. The full western blot membranes are included in the Supplementary material.
The cell cultures were fixed in 2.5 % glutaraldehyde and 1 % paraformaldehyde. The cells were then rinsed with 0.15 M sodium cacodylate (pH 7.2-7.4) for 10 min and incubated in fresh 1 % osmium tetraoxide in 0.1 M sodium cacodylate for 1 h at RT. After incubation, the sodium cacodylate was rinsed away to dehydrate the dishes with 70 % ethanol for 30 min, 95 % ethanol for 30 min and >99 % ethanol for 1 h. A thin plastic layer (#AGR1031, Agar 100 resin kit, Agar Scientific Ltd., Essex, UK) was added to the dishes and incubated for 1 h. The plastic was poured off and a new plastic layer was added onto the dishes for incubation overnight in a desiccator. Next, the plastic was heated to enable its removal after which a new thicker plastic layer was added before another incubation for 1 h in a desiccator. Cells were covered with 3 mm plastic and polymerized in the oven at 60 • C for 48 h. Embedded cells were sectioned by using a Leica Ultracut UTC Ultramicrotome (Rowaco AB, Sweden) and visualized with a Tecnai G2 transmission electron microscope (FEI company) with an ORIUS SC200 CCD camera and Gatan Digital Micrograph software (both from Gatan Inc., CA, USA).

Cytokine assay
Medium samples from control and Aβ-F-treated astrocytes at T1 and T2 were collected and analyzed. The cytokine array (#ARY005B, R&D systems, MN, USA) was performed according to the manufacturer's protocol. Briefly, the membranes were blocked with array buffer 4 at the same time as the samples were incubated with the detection antibody cocktail for 1 h, on shake at RT. Next, the antibody-sample mixture was added to the membranes which were incubated overnight at 4 • C. The following morning, the membranes were washed 3 times for 10 min with washing buffer and incubated with HRP (1:1000) for 30 min before repeating the wash step. The membranes were then incubated with a mixture of ECL reagents 1 and 2 (1:1) for 15 min and developed in the chemiluminescence detection machine for 2 min. Quantification of cytokine levels was performed in images acquired at the optimal intensity of each cytokine. Briefly, the images were analyzed in ImageJ by creating a region of interest (ROI) around the dot signal and measuring the mean intensity value of each sample. The background was subtracted based on measurements from two different blank areas.

Timelapse microscopy
HiPSC-derived astrocytes were cultured on glass bottom culture dishes and recorded using a time-lapse microscopy (Nikon Biostation IM Cell Recorder). Images were captured every 90 or 180 s at 20× magnification. Time lapse movies were exported at a rate of 10 frames/s.

Image analysis
Quantification of Aβ-load was calculated from the Z-stack sum slice projection (constructed with ImageJ) of the intensity of Aβ 42 -555 labelled inclusions at each captured location. Each projection consisted of a 6.86 μm deep Z-section (15 slices). For each time point, 10 of those stacks were captured from each coverslip in a total of three separate cover slips (30 projections per time point). These projections were then used to measure intracellular Aβ-F-aggregates. Briefly, cytoskeletal markers (vimentin) were used to create and outline a ROI. The Aβ-Famount was calculated as the mean integrated density (IntDen) of the Aβ-F-signal within this ROI and normalized to the total cell number in that same image. Quantification of the LAMP1-signal was conducted in the same way but with wide field images.

Statistics
Statistical analyses were conducted in GraphPad Prism v9.3.1. To determine the appropriate type of statistical test we performed a Shapiro-Wilk and a Brown-Forsythe test to assess normal data distribution and the variance of standard deviation in each group, respectively. Welch's t-test and Mann-Whitney test were used for comparing two groups while one-way ANOVA with Bonferroni test as post hoc analysis was used for comparing three or more groups. The level of significance for all the graphs is: * = p < 0.05, ** = p < 0.01, *** = p < 0.001 and **** = p < 0.0001. Results are presented as mean ± SEM.

Human iPSC-derived astrocytes store ingested Aβ-F for an extensive time
Mature human iPSC-derived astrocytes, expressing the markers ALDH1L1, vimentin, Cx43, GLAST and EAAT2 (Fig. S1) (Lundin et al., 2018), were exposed to fluorescently labelled, sonicated Aβ-F for 2 days. The cells were then washed and cultured for another 1 week (T1) or 10 weeks (T2) in the absence of Aβ (Fig. 1A). Parallel control cultures were left untreated. Transmission electron microscopy was performed as a quality check of the fibrils before and after sonication (Fig. 1B). The two time points were chosen to enable a direct comparison between the short and long-term effects of Aβ-inclusions in astrocytes. At T1 and T2, both control and Aβ-exposed astrocytes displayed characteristic starshaped morphology with branching processes.
Data from our previous work suggest that astrocytes can effectively engulf Aβ-F, but have a tendency to store rather than degrade the ingested material (Söllvander et al., 2016). In line with these data, distinct Aβ-inclusions were present in the central part of the cell body of all treated cells (Figs. 1C-D, S2, and Video S1). Interestingly, fluorescent images from T2 revealed significantly increased amounts of Aβ in the treated cells, demonstrating that the astrocytes not only store the ingested Aβ-F for an extensive time, but also accumulate more (Fig. 1E). However, the total number of cells was significantly decreased at T2, suggesting that the extra accumulated Aβ originated from undigested fibrils of engulfed dying/dead astrocytes (Fig. 1F). In support of this hypothesis, immunostainings revealed the presence of dead cells with Aβ-deposits inside live astrocytes at T2 (Fig. 1G). Moreover, time lapse microscopy confirmed that dying astrocytes were instantly taken up by neighboring cells (Fig. 1H, Videos S2 and S3).
The extensive presence of intracellular Aβ-deposits at T2 may be explained by an active containment mechanism, by which the astrocytes try to prevent spreading of Aβ to surrounding cells. Another possibility is that the intracellular Aβ-storage is simply a consequence of failures in the degradation machinery. To sort out these different scenarios, we first analyzed the distribution pattern of cytoplasmic lysosomal associated membrane protein (LAMP1) organelles in relation to the Aβ-inclusions. In both controls and Aβ-exposed T1 astrocytes, LAMP1 was detected mainly around the cell nucleus with only some organelles located further away and closer to the processes (Figs. 2A, S3, S4). In particular, Aβ-inclusions were surrounded by LAMP1-positive compartments, demonstrating that the deposits are contained within lysosomal organelles. Control and Aβ-exposed T2 astrocytes showed the same LAMP1localization pattern around the nucleus as the T1 cells. Further z-stack analysis of the precise localization and distribution patterns of Aβ and LAMP1 confirmed the presence of Aβ within LAMP1-positive organelles (Figs. 2B, S5). Both control and Aβ treated astrocytes appeared to have a stronger LAMP1-signal at the latter time point (Figs. 2A, C, S3).
We have previously shown that following ingestion, protein aggregates are relocated inside the astrocytes to end up in "storage dumps" around the nucleus (Rostami et al., 2017;Söllvander et al., 2016;Streubel-Gallasch et al., 2021). In order to evaluate whether the increased intensity of LAMP1 in the immunostainings was a result of an overall increased LAMP1-expression or a re-distribution of LAMP1positive vesicles, we analyzed cell lysates from control and treated T1 and T2 astrocytes with western blot (WB) (Fig. 2D). The WB results indicated a slight increase of LAMP1-expression at T2, compared to T1 (albeit without reaching statistical significance), suggesting that the elevated LAMP1 intensity in the stained astrocytes may be due to redistribution as well as increased expression.
To further expand our analysis, we quantified the expression of Rab5, a protein characteristic to early endosomes that are vital during the processing of ingested material through the endosomal pathway. Although there was no difference in the levels of Rab5 between control and treated astrocytes at T1, there was a trend towards increased levels in treated astrocytes at T2 (Figs. 2E, F, S6). Taken together, these data demonstrate that the Aβ-deposits remain in lysosomal compartments over time, which may affect the degradation capacity of the cells.
Of note, Aβ-exposed astrocytes displayed an increased tendency to form cellular connections, including tunneling nanotubes (TNTs). Although we could not detect any TNT-mediated cell-to-cell transfer of the Aβ-aggregates within the time-range of the analysis, we did observe frequent exchange of LAMP1-positive organelles between astrocytes with Aβ-inclusions (Figs. 2G, S7).

Reactivity markers are sustained in astrocytes with Aβ-deposits
Astrogliosis is a key phenomenon in the pathophysiology of AD. Reactive astrocytes are commonly located around Aβ-plaques in the human AD brain, but their exact role in Aβ-clearance or spreading is still under evaluation. Upregulation of several key markers has been associated with the reactive state of astrocytes, including vimentin, glial fibrillary acidic protein (GFAP), aquaporin 4 (AQP4) and S100 calciumbinding protein β (S100β). To evaluate if long-term storage affected the astrocytic reactivity, we analyzed the expression of those markers in the hiPSC-derived astrocytes. Immunocytochemistry (ICC) staining of control and Aβ-F-treated astrocytes revealed clear expression of all proteins. In particular, the cytoskeleton protein vimentin was highly expressed in all cells (Fig. 3A). In the same images, AQP4, an important waterchannel membrane protein, was detected around the astrocytic processes (Figs. 3A, S8). Neither vimentin, nor AQP4 showed a different expression pattern in the Aβ-exposed astrocytes compared to the control astrocytes. Also, the expression of GFAP, the most commonly used astrocytic reactivity marker, was evident in both control and Aβ-Ftreated astrocytes (Figs. 3A, S8).
Furthermore, we sought to investigate potential differences in expression between T1 and T2 astrocytes. By analyzing cell lysates from control and Aβ-F-treated astrocytes we found that all proteins had comparable levels at T1 (Fig. 3B). However, at T2, control astrocytes displayed reduced or tendency of reduced expression of all markers, compared to Aβ-F-treated cells, suggesting a prolonged reactive state after Aβ-treatment.

Secretion of pro-inflammatory cytokines is upregulated in long-term astrocytes with Aβ-deposits
Astrocytes are highly secretory cells and therefore, analysis of the type of cytokines and chemokines they release, could provide valuable information regarding their inflammatory state. We investigated the secretion profiles of T1 and T2 astrocytes with or without Aβ-inclusions (Fig. 4). Control and Aβ-F-treated T1 astrocytes displayed similar expression values for the chemo attractants CCL2/MCP-1, MIF and Serpin E1/PAI-1 in the media (Fig. 4A). We could also detect small amounts of IL-18/IL-1F4 in both conditions. However, control T2 astrocytes appeared to have lower CCL2/MCP-1 levels compared to (C) 3D rendering of an astrocyte with Aβ-inclusions from two different viewing angles. (D) Fluorescent images of control and Aβ-F exposed astrocytes at T1 and T2. Phalloidin labels the actin skeleton (white). Treated cells display inclusions of Aβ (red) (Scale bar = 50 μm). (E) The Aβ-inclusions within astrocytes are increased at T2 compared to T1 (F), while the total cell number per image is decreased, suggesting a mechanism of Aβ-accumulation via phagocytosis of neighboring cells (3 replicates for each condition, 10 images per replicate; Welch's t-test and Mann-Whitney test were used for graphs (E) and (F) respectively; * = p < 0.05, **** = p < 0.0001; results are presented as mean ± SEM). (G) Representative image of a T2 astrocyte following ingestion of a dead cell with condensed nucleus, surrounded by Aβ-deposits. (H) Representative image series of a time lapse experiment that further confirms the uptake of apoptotic cells by neighboring astrocytes at T2 (the apoptotic and engulfing cells are outlined and marked with a black and yellow asterisk, respectively). The complete time lapse movies are provided as supplementary video files (Scale bar = 50 μm).

Aβ-accumulation induces ER swelling and formation of pathological lipid structures in astrocytes
The exact mechanism of how exposure to Aβ-F interferes with compartments. (C) Quantification of the LAMP1-immunostaining. (D-E). Quantification of western blots from cell lysates from T1 and T2 astrocytes using antibodies against LAMP1 and Rab5 (all the compared samples were run on the same gel, T1 = 3 replicates for each condition; T2 = 2 replicates for each condition; one-way ANOVA with Bonferroni test as post hoc analysis was used for both graphs; **** = p < 0.0001; results are presented as mean ± SEM). (F) Representative images of Rab5-expression in T1 and T2 astrocytes (Scale bar = 50 μm). (G) Phalloidin staining indicates formation of TNTs between neighboring astrocytes and cell-to-cell exchange of LAMP1-positive organelles (Scale bar = 10 μm). (A) Representative images of astrocytes expressing GFAP (green), vimentin (red), S100β (green) and AQP4 (green) (Scale bar = 50 μm). (B) Quantification of western blots from cell lysates from T1 and T2 astrocytes using antibodies against vimentin, GFAP, S100β and AQP4 (all the compared samples were run on the same gel, T1 = 3 replicates for each condition; T2 = 2 replicates for each condition; one-way ANOVA with Bonferroni test as post hoc analysis was used for both graphs; * = p < 0.05, ** = p < 0.01, **** = p < 0.0001; results are presented as mean ± SEM). cellular homeostatic functions remains unclear. To that end, we evaluated cellular structures via electron microscopy. We observed normal, healthy mitochondrial profiles and lysosomes around the nucleus of control T1 astrocytes (Fig. 5A). In contrast, analysis of Aβ-F-treated T1 astrocytes revealed altered mitochondrial profiles and lysosomes around the nucleus (Fig. 5A). Enlarged lysosomes were detected in both control and Aβ-F-treated T2 astrocytes (Fig. 5B). Furthermore, we observed abnormalities in several other organelles in Aβ-F-treated astrocytes at both timepoints, but also in control T2 astrocytes (Fig. 5C). More specifically, we detected abnormal swollen endoplasmic reticulum (ER) ( Figure 5C1), organelles with double membranes encapsulating smaller inclusions that resemble autophagosomes ( Figure 5C2) and circular organelles that were identified as lipid droplets ( Figure 5C3). Lipid droplets are characteristic lipid-rich organelles that are used by various cell types for lipid storage. None of these pathological structures were observed in control astrocytes at T1, suggesting potential dysregulation in response to astrocytic Aβ-accumulation and/or extended culturing time. Disturbed energy supply and lipid droplet formation have been linked to brain aging and neurodegenerative disorders, including AD (Mecocci et al., 2018;Ralhan et al., 2021). However, whether these alterations constitute a cause or a consequence warrants further investigation.

Discussion
Alzheimer's disease is the leading cause of dementia worldwide and poses a tremendous societal and economic burden (Tahami Monfared et al., 2022). It is a progressive disease, where dementia symptoms gradually worsen over a number of years. In late-stage AD, individuals lose the ability to carry on a conversation and respond to their environment, making it one of the most devastating diagnoses that patients and their families can receive.
Knowledge about the cellular mechanisms behind the initiation and propagation of AD is still very limited. Up until now, most research has been focused on neuronal abnormalities, but accumulating evidence indicates that equal focus must be allocated to identify non-neuronal therapeutic targets. Astrocytes are a major component of the central nervous system and their involvement in neurodegenerative diseases, including AD, has received much attention during the last decade (Ding et al., 2021). It has been suggested that mouse astrocytes have the capacity to clear Aβ. (Wyss-Coray et al., 2003) In line with that, we have previously shown that astrocytes have the ability to rapidly engulf large amounts of Aβ-aggregates. However, in contrast to microglia, neither mouse (Söllvander et al., 2018(Söllvander et al., , 2016 nor human astrocytes (Rostami et al., 2021) fully degrade the ingested aggregates. Instead, they partially digest the Aβ, and the remaining material is stored as large, compact intracellular deposits (Söllvander et al., 2016). Hence, Aβ aggregates that are present in the brain are probably degraded if they are phagocytosed by microglia, but not if they are ingested by astrocytes. We have noticed that the Aβ-deposits in astrocytes are difficult to detect with commercially available antibodies, presumably because the astrocytes have modified the proteins so much that the antibodies do not recognize them. By using fluorescently labelled human Aβ, it is possible to circumvent this problem. The intracellular Aβ-accumulation is very stressful for the astrocytes (Söllvander et al., 2016) and long-term Aβ-deposits are probably detrimental to their functionality. If the astrocytes cope with the Aβ-storage, that could serve as a confinement strategy in the long run, thereby minimizing the spread of potentially pathogenic Aβ-species to neighboring cells. In murine astrocytes, Aβ-accumulation is known to result in secretion of N-truncated Aβ-peptides in the extracellular medium that are toxic to neuronal cultures (Beretta et al., 2020;Söllvander et al., 2016). In our current study, we aimed to evaluate how long-term presence of Aβ-inclusions in hiPSC-derived astrocytes affects the cellular functions, focusing on the lysosomal pathway and markers of astrocytic reactivity.
To date, a number of studies have shown that astrocytes effectively degrade ingested monomeric Aβ-peptides (Basak et al., 2012;Koistinaho et al., 2004;Söllvander et al., 2016). However, aggregated Aβ-forms are apparently not processed in the same manner and end up as large intracellular inclusions that are not easily cleared (Nielsen et al., 2010;Söllvander et al., 2016). These astrocytic inclusions have also been reported to be present in the cortex of post mortem human AD brains and their size correlates with the pathology in the respective region (Nagele et al., 2003;Thal et al., 2000). The data presented in this study demonstrate that after 10 weeks of culturing (in Aβ-free medium), hiPSC-derived astrocytes still retain the ingested Aβ-deposits. Interestingly, they also maintain their capacity to absorb aggregated Aβ from their surroundings (by engulfing dead cells with Aβ-inclusions). Hence, the astrocytes are either not able to adequately degrade the fibrillary Aβ or actively choose to store it. Although extracellular Aβ-plaques are considered the classical hallmark of AD pathology, understanding the role and processing of intracellular deposits provides valuable insight into disease spreading and neuronal dysfunction.
Astrocytic Aβ-inclusions have been reported to co-localize with LAMP1-positive organelles, suggesting a direct interplay of lysosomes in the storage and degradation of Aβ-aggregates (Basak et al., 2012;Söllvander et al., 2016). Consistent with those reports, we found that Aβ-F are contained within LAMP1-positive compartments in both the early (T1) and long-term (T2) cultures, suggesting that the lysosomal machinery fails to degrade the material. Although the intensity of the LAMP1-ICC staining was significantly increased over time, quantification of the protein in cell lysates only showed a trend of higher LAMP1levels at T2 compared to T1. Presumably, the more intense LAMP1immunostaining could be due to reorganization of the lysosomes (with more LAMP1-positive puncta gathering), in addition to an upregulation of "free" LAMP1-protein. Another important protein in the endosomallysosomal pathway, Rab5, remained elevated in cell lysates from Aβ-Ftreated T2 astrocytes, but was decreased in control T2 astrocytes, hinting at a demand for additional degradation machinery in the presence of Aβ. Consequently, the endosomal pathway that participates in engulfment and degradation of ingested material, appeared affected in the long-term cultures, indicating that astrocytes continuously try to get rid of the Aβ-deposits via the lysosomal pathway, albeit with no success. It is known that the aged brain has a reduced degradation capacity compared to the young brain, probably because the glial cells have a reduced lysosomal function (Hanslik et al., 2021). Further studies are required to fully understand the underlying mechanisms of how the altered LAMP1/ Rab5-levels affect the degradation capacity of the astrocytes over time.
Cell-to-cell transfer is a broad term that includes direct membrane fusion, as well as intercellular connections via TNTs or thicker protrusions. We detected frequent exchange of LAMP1-positive organelles between neighboring cells via TNTs. Tunneling nanotubes are known to not only enable transfer of cellular components, but also to participate in the spreading of pathogenic proteins, and viruses (Mittal et al., 2019;Rostami et al., 2017;Sisakhtnezhad and Khosravi, 2015). We have previously demonstrated that α-synuclein accumulation in cultured human astrocytes results in severe cellular stress, indicated by lysosomal, mitochondrial, and ER deficiencies. The stressed α-synuclein exposed astrocytes respond by sending out TNTs, enabling intercellular transfer of α-synuclein inclusions and mitochondria to nearby cells (Rostami et al., 2017). In this study, we found similar stress patterns in Aβ-exposed astrocytes, regarding ER and mitochondrial swelling. Notably, at T2, we could detect altered lysosomes in control astrocytes that could further point towards a stressful environment. In contrast to our previous studies on α-synuclein, we were not able to detect any exchange of Aβ via TNTs at T1 or T2 within the two days of live cell analysis. Hence, our data suggest that inclusions of Aβ and α-synuclein are, at least to some extent, handled in different ways by the astrocytes.
Reactive astrocytes are associated with neurodegenerative diseases in general, and AD in particular . Their response to pathological conditions in the brain has been proposed to be either protective or detrimental to the surrounding cells (Acioglu et al., 2021;Anderson et al., 2014). The reactive state of astrocytes is characterized by changes in cell morphology as well as increased expression of specific markers and secretion of inflammatory cytokines. By analyzing the expression levels of reactivity associated markers in cell lysates of control and Aβ-F-treated astrocytes at the early and late time point, we noted a sustained reactivity in Aβ-F-treated astrocyte cultures. Interestingly, both control and Aβ-F-treated astrocytes displayed the same levels of vimentin, GFAP, AQP4 and S100β at T1, suggesting a potential effect of the culture conditions on cellular phenotype, rather than a representation of the Aβ-F-treatment at this time point. Astrocytes that were cultured for 9 more weeks (T2) displayed clear time-related changes in both the Aβ-treated and untreated conditions. Analysis of cytokines secreted in the medium revealed the persistent expression of CCL2/MCP-1 in Aβ-exposed T2 astrocytes. Although its role is not fully understood, CCL2/MCP-1 has been shown to be increased in patients with HIV-1-associated dementia and other neuropathological conditions and is known to participate in the recruitment of macrophages to the brain, as a defense mechanism (Lawrence et al., 2006). It is possible that a more complex culture system is needed to obtain all aspects of astrocyte reactivity, as the intercellular crosstalk with infiltrating inflammatory cells, as well as with other brain cells are crucial for their activity status.
Aging of astrocytes has been associated with changes in gene expression, cellular structure, and function. Accumulating data indicate that the alterations are multifaceted, region-dependent and may differ between species (Preininger and Kaufer, 2022). Astrocytes isolated from aged mice showed impaired ability to carry out essential functions over time (Clarke et al., 2018;Ishii et al., 2017). Traditionally, aged astrocytes have been synonymous with an increase in reactivity markers. However, contradictory data show a drop in GFAP expression in aged rodents (Orre et al., 2014). The effect of in vitro aging may also differ, depending on the astrocyte source. Studies on murine astrocytes report increased reactivity following long-term culturing (Revuelta et al., 2021), while our results on human astrocytes indicate a reduction in reactivity markers over time. In line with our data, a previous study demonstrated that oxidative stress-induced senescence results in reduced expression of GFAP and S100β in human astrocytes (Crowe et al., 2016). In order to reduce the impact of those in-vitro aging alterations in our reported results, we included control and Aβ-F-treated cells from both timepoints and performed the relevant comparisons between the different conditions.

Limitations
A major challenge with in vitro-models of AD is to recapitulate the physiological aging process. An important parameter in this context is the origin and maturity of the cell culture system. In our study, we addressed this issue by culturing fully differentiated hiPSC-derived astrocytes with or without Aβ-inclusions for a period of 10 weeks. During the long culture time, there will inevitably be changes in the astrocytes gene expression, as well as alterations in cellular structure and function. We primarily explored the differences between cells treated with Aβ and their control counterparts, without performing comparisons between the early and late time points. However, investigating how the long-term culturing itself affects the astrocytes would also provide valuable information to the field. Another limitation of the study is the number of cell lines and replicates included in the analysis. Although, heterogeneity between hiPSC lines may affect the results, our previous data demonstrate that murine cells, as well as human astrocytes of another origin ingest and accumulate aggregated proteins in a similar fashion. Finally, the cytokine assay used in this study is qualitative and additional studies with more sensitive methods are needed to obtain quantitative data.

Conclusions
Many of the pathological changes in the AD brain are estimated to occur decades before the first clinical symptoms emerge. According to the amyloid cascade hypothesis, Aβ-accumulation is the causative agent of AD, consequently driving the formation of neurofibrillary tangles, inflammation, and finally neuronal cell loss. However, the exact cellular and molecular mechanisms by which Aβ induce these processes remain unclear. In conclusion, this study contributes with knowledge of how long-term accumulation of Aβ affects astrocytes, which is of importance to uncovering the role of glial cells in AD progression and identify novel treatment strategies.

Declaration of competing interest
The authors declare no competing interest.

Data availability
No data were used for the research described in the article.