A 3D human iPSC-derived multi-cell type neurosphere system to model cellular responses to chronic amyloidosis

Wendt, Stefan; Lin, Ada J.; Ebert, Sarah N.; Brennan, Declan J.; Cai, Wenji; Bai, Yanyang; Kong, Da Young; Sorrentino, Stefano; Groten, Christopher J.; Lee, Christopher; Frew, Jonathan; Choi, Hyun B.; Karamboulas, Konstantina; Delhaye, Mathias; Mackenzie, Ian R.; Kaplan, David R.; Miller, Freda D.; MacVicar, Brian A.; Nygaard, Haakon B.

doi:10.1186/s12974-025-03433-3

Research
Open access
Published: 24 April 2025

A 3D human iPSC-derived multi-cell type neurosphere system to model cellular responses to chronic amyloidosis

Stefan Wendt¹,
Ada J. Lin²,
Sarah N. Ebert^1,4,
Declan J. Brennan²,
Wenji Cai²,
Yanyang Bai¹,
Da Young Kong²,
Stefano Sorrentino²,
Christopher J. Groten¹,
Christopher Lee²,
Jonathan Frew^2,5,
Hyun B. Choi¹,
Konstantina Karamboulas⁶,
Mathias Delhaye¹,
Ian R. Mackenzie⁷,
David R. Kaplan^3,6,8,
Freda D. Miller^3,4,
Brian A. MacVicar¹ &
…
Haakon B. Nygaard²

Journal of Neuroinflammation volume 22, Article number: 119 (2025) Cite this article

1071 Accesses
7 Altmetric
Metrics details

Abstract

Background

Alzheimer’s disease (AD) is characterized by progressive amyloid beta (Aβ) deposition in the brain, with eventual widespread neurodegeneration. While the cell-specific molecular signature of end-stage AD is reasonably well characterized through autopsy material, less is known about the molecular pathways in the human brain involved in the earliest exposure to Aβ. Human model systems that not only replicate the pathological features of AD but also the transcriptional landscape in neurons, astrocytes and microglia are crucial for understanding disease mechanisms and for identifying novel therapeutic targets.

Methods

In this study, we used a human 3D iPSC-derived neurosphere model to explore how resident neurons, microglia and astrocytes and their interplay are modified by chronic amyloidosis induced over 3–5 weeks by supplementing media with synthetic Aβ1 - 42 oligomers. Neurospheres under chronic Aβ exposure were grown with or without microglia to investigate the functional roles of microglia. Neuronal activity and oxidative stress were monitored using genetically encoded indicators, including GCaMP6f and roGFP1, respectively. Single nuclei RNA sequencing (snRNA-seq) was performed to profile Aβ and microglia driven transcriptional changes in neurons and astrocytes, providing a comprehensive analysis of cellular responses.

Results

Microglia efficiently phagocytosed Aβ inside neurospheres and significantly reduced neurotoxicity, mitigating amyloidosis-induced oxidative stress and neurodegeneration following different exposure times to Aβ. The neuroprotective effects conferred by the presence of microglia was associated with unique gene expression profiles in astrocytes and neurons, including several known AD-associated genes such as APOE. These findings reveal how microglia can directly alter the molecular landscape of AD.

Conclusions

Our human 3D neurosphere culture system with chronic Aβ exposure reveals how microglia may be essential for the cellular and transcriptional responses in AD pathogenesis. Microglia are not only neuroprotective in neurospheres but also act as key drivers of Aβ-dependent APOE expression suggesting critical roles for microglia in regulating APOE in the AD brain. This novel, well characterized, functional in vitro platform offers unique opportunities to study the roles and responses of microglia to Aβ modelling key aspects of human AD. This tool will help identify new therapeutic targets, accelerating the transition from discovery to clinical applications.

Highlights

Well-characterized functional human iPSC-derived 3D neurospheres (hiNS) consisting of neurons and astrocytes can be supplemented with microglia/macrophages (hiMG).
Chronic amyloidosis in the presence of hiMG recapitulate key features and gene expression profiles of AD.
hiMG within the model phagocytose Aβ and mitigate Aβ-induced neurotoxicity, reducing oxidative stress and neuronal damage
hiMG are essential for Aβ to upregulate AD-like gene expression signatures in astrocytes.
Immunohistochemical analysis reveals hiMG-dependent colocalization of Aβ and APOE.

Introduction

Alzheimer’s disease (AD) remains inextricably linked to progressive aggregation of Aβ and Tau in the brain [1]. Both are prime targets for therapeutic interventions, with two monoclonal antibodies against Aβ recently approved by the US Food and Drug Administration for use in AD [2, 3]. The mechanisms by which Aβ and Tau cause neurodegeneration are not fully understood. Aβ plaques initiate a significant neuroinflammatory reaction including robust microglial activation. However, the detrimental versus neuroprotective effects of microglia activation remains controversial. For example, a mutation in TREM2, which reduces the efficiency of microglial Aβ phagocytosis confers significant risk for developing sporadic AD, suggesting an important protective role of microglia [4]. Microglial phagocytosis is also the principal mechanism by which monoclonal antibodies against Aβ, such as lecanemab and donanemab, remove brain Aβ, leading to improvement in clinical function and AD biomarkers [5, 6]. However, preclinical models have shown that microglial activation can also lead to excessive synaptic pruning which can accelerate neurodegeneration [7]. While there is broad agreement on the potential importance of resolving how glia contribute to sporadic AD pathophysiology [8,9,10], finding the most relevant preclinical system to best model AD pathophysiology remains a significant challenge. Murine models of AD have been used to investigate cellular mechanisms in AD for several decades, but clinical trials based on results from these models have rarely been successful [11]. The establishment of cerebral organoid and neurosphere 3D human cell culture systems sparked widespread interest in developing 3D cell culture models that more faithfully resemble human AD, to achieve better translation of pre-clinical studies to novel therapies [12,13,14,15]. Most of these approaches rely on the expression of familial AD mutations in APP or PSEN1 [16,17,18,19]. Mimicking the AD-associated transcriptional changes of glial cells in a human 3D sporadic AD model system could help to unravel the impact of neuron-glia-Aβ interactions in AD.

Here we describe the utility of using human induced pluripotent stem cell (hiPSC)-derived 3D neurospheres (hiNS) to determine the impact of chronic Aβ exposure, a characteristic feature of AD, on neuronal and glial function. Introducing hiPSC-derived microglia (hiMG) into hiNS during chronic Aβ treatment triggered striking neuroprotection, by preventing Aβ-induced dysfunction of neuronal activity, oxidative stress and reducing neuronal death. While hiMG displayed a remarkable phagocytic activity towards Aβ, single nuclei RNA sequencing revealed that the addition of hiMG also had significant effects on astrocyte and neuronal transcriptional profiles. For example, many AD-associated genes of interest in astrocytes, such as APOE, CLU, LRP1 and VIM, were upregulated by Aβ only in the presence of hiMG. This study sheds light on the critical timing and multifaceted role of microglia in neuroprotection against Aβ, offering new insights into potential therapeutic strategies for AD.

Results

Formation of hiPSC-derived neurospheres as a 3D tissue culture system includes neurons, astrocytes and microglia

Blood-derived hiPSCs were generated from a healthy individual and subsequently differentiated into neural progenitor cells (NPCs) (Fig. 1A, see Suppl. Table 1 for media compositions). NPCs were plated in an AggreWell™ 800 plate to initiate the formation of hiNS. After 7 days in vitro (DIV), hiNS were transferred to 9 cm petri dishes on an orbital shaker for subsequent differentiation and maturation before plating into 48-well plates (Fig. 1B). While NPCs spontaneously differentiated into neurons and astrocytes, microglia (hiMG) did not differentiate in the 3D neurospheres. Therefore, hiMG were cultured in parallel to hiNS (Fig. 1A) and added to a subset of mature neurospheres with neurons and astrocytes/NPCs to determine the impact of microglia. Neurospheres without added hiMG are termed hiNS(−) whereas neurospheres with hiMG infiltration are termed hiNS(+) (Fig. 1B).

To confirm the identity of cell types in hiNS and their patterns of gene expression, we performed single nuclei RNA sequencing (snRNA-seq). Specifically, we cultured hiNS for 60 days, and hiMG were added and allowed to infiltrate into the tissue for the last 10 days. hiNS(+) or hiNS(−) were collected and flash frozen for subsequent snRNA-seq analysis at DIV 60 (Fig. 1B). Four to six individual hiNS in each group were pooled together for nuclei isolation. Single cell nuclei were isolated and sequenced using the 10× Genomics platform, and transcriptome analysis was performed using our previously described pipeline [20,21,22,23,24]. After filtering the datasets to remove low quality cells with few expressed genes or high mitochondrial proportions and cell doublets, we obtained 4678 and 1892 single cell transcriptomes for the hiNS(+) and hiNS(−), respectively. We then merged the datasets to allow a direct comparison of transcriptomes in the two conditions. Genes with high variance were used to compute principal components as inputs for projecting cells in two-dimensions using t-distributed stochastic neighbor embedding (t-SNE) and clustering, performed using a shared nearest neighbors-cliq (SNN-cliq)-inspired approach built into the Seurat R package at a range of resolutions.

As performed in previous work [25, 26], we corrected for read depth and library size variation using logNormCounts [21]. Once the matrices were normalized they were imported into Seurat, and principal component analysis (PCA) was performed, using the top 2000 highly variable genes. Two-dimensional t-SNE projections were generated using the top principal components (RunTSNE, Seurat). The same principal components were then used with SNN-Cliq-inspired clustering (FindClusters, Seurat). To annotate and visualize cell types and states, we used gene expression overlays on UMAP plots (FeaturePlot, Seurat).

Uniform Manifold Approximation and Projection for Dimension Reduction (UMAP) visualization and analysis of well-defined marker genes identified 16 clusters containing the predicted cell types (Fig. 1C, D). Cell clusters with similar gene expressions were grouped together in a two-dimensional UMAP. These included astrocytes expressing GFAP, AQP4, SLC1 A3 and GABBR2 (cluster 0), and two types of neurons in various stages of maturation: inhibitory interneurons (clusters 4, 7, 9 and 14) expressing GAD1, GAD2 and SYN3, and excitatory neurons (clusters 2, 3, 5, 6, 11, 12 and 15) enriched for GRID2, GRM7, SLC17 A6 and SYN3. The dataset also included precursor cells (clusters 1, 8, 10 and 14), and microglia/macrophages expressing PTPRC, AIF1, CX3 CR1 and CD68 (cluster 13). Note that GFAP expression was not exclusive to astrocytes but also highly expressed in some precursor populations. We did not find evidence for the presence of any mature oligodendrocyte populations in our cell culture system, but found OLIG1 and OLIG2 expression in parts of our NPC populations suggesting that either our hiNS are not aged for long enough to give rise to oligodendrocytes, or factors are missing in our cell culture system to allow for oligodendrocyte differentiation.

Analysis of the dataset of origin (Fig. 1E) confirmed that the microglial cluster included cells only from the hiNS(+) dataset. The other clusters included intermingled cells from both conditions, except for cluster 0 astrocytes, where many of the cells appeared transcriptionally distinct, evident by the visual sub-clustering of hiNS(−) and hiNS(+) astrocytes. To further explore this finding, we subsetted the astrocytes and reanalyzed them separately. This analysis (Fig. 1F, G) defined two large clusters and one smaller cluster. Two of the clusters (0a and 0c) included astrocytes from both conditions, while the other large cluster (0b) was almost completely comprised of hiNS(+) astrocytes. Pearson correlation analysis of average gene expression confirmed that, as suggested by the clustering analysis, hiNS(+) astrocytes were similar but not identical to the hiNS(−) astrocytes, with r = 0.978 (Fig. 1H).

To better understand these similarities and differences, we performed differential gene expression (DEG) analysis. There were 400 DEGs when comparing the two conditions, with 44 enriched in hiNS(+) astrocytes and 356 in hiNS(−) astrocytes (> 1.3-fold average change, p adj < 0.05; Suppl. Tables 2 and 3). Notably, DEG analysis coupled with gene ontology (Fig. 1I, J; Suppl. Table 4) demonstrated that hiNS(+) astrocytes were highly enriched for genes and gene categories associated with oxidative phosphorylation such as MT-CO3, MT-ND3, MT-ATP6, and MT-CYB, suggesting they were more metabolically active. By contrast, hiNS(−) astrocytes gene ontology analysis identified significant gene categories including neurogenesis, post-synapse and synapse organisation (Fig. 1J, Suppl. Table 5).

hiNS are highly functional 3D human tissue cultures

The presence of the various differentiated major cell populations identified in our snRNA-seq analysis were confirmed using immunofluorescence staining for neurons (MAP2), astrocytes/NPCs (GFAP), and microglia/macrophages (IBA1) in fixed hiNS(+) at DIV 60. Neurons and astrocytes with extensive processes were observed throughout the neurospheres in addition to hiMG (Fig. 2A). Additional immunostainings were performed for NeuN, SLC7 A11 and CD45 to confirm the cellular identity of all three cell types (Fig. S1).

hiNS were readily transduced using adeno-associated viruses (AAVs) encoding cell type specific expression with the neuronal-selective promoter hSyn or the astrocyte-selective promoter gfaABC1D at DIV 39–41 of culture. AAV expression can be detected reliably about a week after transduction, and it takes a few more days for expression to be quantifiable. To monitor tissue health of hiNS, we transduced neurospheres with hSyn-roGFP1, a redox-sensitive green fluorescent protein [27] that detects cellular oxidative stress associated with reactive oxygen species production and, typically, necrosis [28]. Neurons in hiNS display a largely reduced cytosol, confirming the absence of necrotic tissue (Fig. 2B). Imaging intracellular calcium concentration transients was used for tracking neuronal activity in hiNS. This was monitored by imaging hSyn-GCaMP6f or hSyn-jRGECO1 expression for live imaging of neuronal activity versus imaging GCaMP8 s in astrocytes induced by gfaABC1D-GCaMP8 s expression. hiNS(−) displayed frequent synchronized and propagating calcium transients. Co-transduction of hSyn-jRGECO1 with gfaABC1D-GCaMP8 s allowed us to image intracellular calcium of neuronal and astrocytic cell populations simultaneously and revealed that calcium elevations rise in tandem in both populations (Fig. 2C). Note that the jRGECO1 signal to noise ratio was inferior to GCaMP6f imaging. We therefore focused on utilizing GCaMP6f for monitoring neural calcium wave activity moving forward. In addition to monitoring [Ca²⁺]_i using jRGECO1 or GCaMP6f we additionally transduced hiNS using hSyn-iGluSnFr, to monitor extracellular glutamate levels and found elevated glutamate signals to coincide with neural calcium wave activity (Fig. 2D). The synchronized large [Ca²⁺]_i transients traversed through hiNS in a wave-like pattern reminiscence of large calcium waves (Fig. 2E) observed during neurodevelopment [29, 30].

Astrocytic uptake of extracellular glutamate is known to regulate neuronal calcium activity [31, 32]. To further characterize the functional properties of hiNS we induced hSyn-iGluSnFr expression in neurons to monitor extracellular glutamate release during calcium wave propagations. Our snRNA-seq data revealed widespread expression of the glutamate transporters EAAT1 (SLC1A3) and EAAT2 (SLC1A2), with high levels of expression in astrocytes (Fig. S2 A). We monitored baseline wave-like activity of extracellular glutamate concentrations using (Fig. S2B) followed by pharmacological block of EAAT1/2 with TFB-TBOA led to a steady rise in extracellular glutamate levels which eventually inhibited wave formation (Fig. S2 C). We performed additional pharmacological experiments to characterize the nature of calcium waves in hiNS. [Ca²⁺]_i waves were suppressed by either the Na⁺ channel blocker TTX, or ionotropic glutamate receptor antagonists AP5/CNQX, or by applying GABA. Picrotoxin application, which antagonizes GABA receptors and chloride channels, significantly increased calcium wave frequencies (Fig. 2F).

Before assessing the impact of hiMG on the functional properties of hiNS, we collected protein samples as well as coverslips of fixed hiMG to perform immunostainings. Immunofluorescence for IBA1 (Fig. S3 A) as well as western blots (Fig. S3B) confirmed robust IBA1 expression. We further tested and confirmed that our hiMG can be stimulated with LPS leading to robust TNFα and Il1β expression (Fig. S3 C). Full length blots for TNFα and Il1β are shown in Fig. S3D as well as all individual blots (Fig. S3E).

We then tested whether infiltrating hiMG alter calcium wave frequencies by comparing GCaMP6f recordings from hiNS(−) (N = 10) with hiNS(+) (N = 12). The presence of hiMG reduced the calcium wave frequency by half, from 0.19 to 0.09 Hz (***p < 0.001) and increased wave duration from 3.2 to 5.0 s (**p < 0.01), indicating that hiMG leads to altered neuronal activity (Fig. 2G).

We next investigated the infiltration behaviour and physiological properties of hiMG during and after tissue infiltration into hiNS. We performed 24 h-continuous confocal live imaging on hiNS expressing hSyn-mCherry and gfaABC1D-EGFP to monitor live-stained hiMG infiltrating into hiNS (N = 4). hiMG were pre-stained with tomato lectin DyLight 649. Tracking hiMG movements in hiNS revealed infiltration occurring 6–12 h after the transfer (see yellow arrows, Fig. 2H). To test if hiMG that infiltrated within hiNS(+) exhibit physiological microglia-like properties, we again stained hiMG with tomato lectin (DyLight 488) 7–9 days post-infiltration into hiNS. hiNS(+) were imaged by 2-photon microscopy to monitor microglial process movement, damage responses and for patch-clamp recordings. We observed tissue surveillance by hiMG processes as well as process outgrowth towards a focal laser lesion mimicking in vivo-like behaviour (N = 3, Fig. 2I). Patch-clamp recordings on hiMG within hiNS(+) confirmed microglial identity, revealing average reversal potentials of − 38 ± 3.9 mV and cell capacitances of 22 ± 2.4 pF (N = 9). A series of de- and hyperpolarizing voltage steps revealed inward rectifying potassium currents typical for microglia [33,34,35] (Fig. 2J). We conclude that infiltrating hiMG display a robust microglial-like phenotype exhibiting several properties that resemble in vivo microglia.

Chronic treatment of hiNS with oligomeric Aβ results in plaque-like aggregates and a series of neurotoxic effects

To model how the human brain responds to Aβ aggregation, we first tested if supplementing hiNS(−) media with Aβ42 results in the formation of insoluble, plaque-like aggregates in the tissue. We followed a well characterized [36,37,38,39,40] protocol for the formation of oligomeric Aβ42 (oAβ) and tested three different concentrations added during regular media changes for a total of 7 days (0.012, 0.048 and 0.24 µg/mL). Only the highest Aβ concentration resulted in substantial accumulation of insoluble Aβ in the tissue (Fig. S4 A, B). As an alternative strategy, we tested whether exposing hiNS(−) to fibrillar Aβ42 (fAβ) for 7 days would have the same effect as that seen with oAβ. However, fAβ caused highly aggregated Aβ deposits that appeared as abnormally large sheets spanning the tissue and were much different from the plaque-like aggregates found after oAβ stimulation (Fig. S4 C). We consequently focused on using oAβ treatment instead of fAβ for tissue culture treatments.

We next tested the effects of chronic oligomeric Aβ42 (0.24 µg/mL) for up to 35 days, first in hiNS(−), in the absence of hiMG (Fig. 3A). Large quantities of aggregated Aβ were detected with different Aβ antibodies within hiNS(−) (Fig. 3B). Immunostaining for neurons and astrocytes/NPCs revealed degenerated or swollen processes and cell bodies at the end of the treatment period indicating severe toxicity (Fig. 3C, see arrows). To monitor neuronal redox levels, hiNS(−) expressing hSyn-roGFP1 were imaged after treatment, revealing severe oxidative stress evident by a significant increase in roGFP1 redox ratios (Fig. 3D). Continuous (up to 20 h) live roGFP1 imaging in Aβ treated hiNS(−) revealed a significant increase in the frequency of oxidizing cells compared to hiNS(−) not exposed to Aβ, further confirming Aβ-induced oxidative stress (Fig. S5). Neuronal activity was repeatedly monitored by imaging GCaMP6f in hiNS(−) as described above. A significant reduction in calcium wave frequencies was observed at the end of the third week and in the fourth week of Aβ exposure (Fig. 3E).

We then investigated whether chronic Aβ treatment ultimately results in neuronal loss. Immunostaining for cleaved Caspase- 3 (c-Cas- 3), a key mediator of neuronal apoptosis, indicated a ~ 50% increase in apoptotic cells after Aβ treatment in hiNS(−) as well as a significant reduction of nuclei density within the tissue, suggesting cell death (Fig. 3F). Neuronal apoptosis during Aβ exposure was observed by co-staining for c-Cas- 3 and the neuronal marker MAP2 (Fig. 3G). To confirm the increase in apoptotic cell death, we performed terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) to detect fragmented DNA. There was a significant increase in TUNEL fluorescence confirming apoptosis during chronic Aβ exposure (Fig. 3H).

To validate our hiNS differentiation and chronic amyloidosis protocols, we have grown hiNS using the commercial KOLF2.1J cell line and treated them for up to 38 days with oligomeric Aβ. KOLF2.1J hiNS were transduced with the same AAVs to induce roGFP1 and GCaMP6f for monitoring neuronal redox states and calcium wave activity, respectively (Fig. S6 A). Chronic Aβ treatment resulted in similar oxidative stress which gradually increased in severity with treatment time (Fig. S6B). Spontaneous neural calcium wave activity was present, similar to our hiNS derived from our main cell line, and activity decreased significantly with Aβ exposure (Fig. S6 C). Overall, this data suggests that our hiNS differentiation as well as the chronic amyloidosis protocol induced neuropathology can be utilized using alternative and commercially available hiPSCs.

In conclusion, our chronic Aβ stimulation protocol resulted in a sequence of neurotoxic effects including the formation of plaque-like Aβ aggregates within days of exposure, oxidative stress and dysfunctional neuronal activity 3–4 weeks into Aβ treatment, and ultimately Caspase- 3 dependent apoptosis and neuronal loss 4–5 weeks after Aβ exposure.

hiMG dependent phagocytosis and removal of plaque-like Aβ aggregates

The role of microglia in AD pathogenesis has been extensively studied, with both neurotoxic and neuroprotective roles described [41, 42]. We designed two different chronic Aβ treatment protocols to elucidate the impact of microglia on neurotoxicity when added to hiNS at different time points following Aβ exposure. This included six experimental conditions: control (ctrl), 3-week (3w Aβ hiNS) or 5-week (5w Aβ hiNS) exposure each with (+) or without (−) hiMG. hiMG were added to the tissue for the last 10 days of the treatment protocol (Fig. 4A).

We first assessed the impact of microglia on Aβ aggregation. Aβ immunoreactivity was quantified using an anti-Aβ antibody (4G8) for all groups. The amount of Aβ was less in 3w Aβ hiNS(+) compared to 5w Aβ hiNS(−) (Fig. 4B). hiMG addition significantly reduced Aβ levels at 3w (~ 17% Aβ coverage reduced to ~ 3%, ***p < 0.001), but not at 5w, indicating that adding hiMG is more efficient at clearing Aβ aggregates during the shorter Aβ treatment protocol, with a shorter overall exposure to Aβ (Fig. 4B). The more efficient removal of Aβ indicates that neuroprotective properties of hiMG could be more pronounced at 3w than at 5w Aβ exposure in hiNS(+).

Next, we investigated whether hiMG phagocytosis of Aβ contributed to the observed reduction in Aβ deposits. Co-staining microglia (lectin or IBA1) and Aβ (6E10, 4G8 or mOC87) revealed that Aβ was readily found in the cytosol of infiltrated hiMG in Aβ-treated hiNS(+) (Fig. 4C). The strongest intracellular Aβ signal was seen by staining specifically for fibrillar Aβ (mOC87 [43]), while the Aβ antibody 6E10 revealed only a faint signal inside hiMG, indicating that Aβ inside hiMG appears mostly in a highly aggregated form. Immunostaining for CD68 confirmed that phagocytosing hiMG in both 3w and 5w Aβ treated hiNS(+) express CD68, a well-known phagocytosis marker[44]. Intracellular Aβ was identified after both 3w and 5w Aβ treatment (Fig. 4D, white arrows). While we did not find hiMG to significantly reduce overall extracellular Aβ burden after 5w, they nevertheless appear to have cleared significant amounts of Aβ, leaving Aβ cleared tissue in their path. This indicates that in 5w Aβ hiNS(+) microglia had removed aggregates but they either lost the capacity during the 10 day time window of co-culture or the manifestation of newly forming Aβ aggregates is faster than the phagocytosis of them by microglia (Fig. 4E, white arrows) resulting in no net change in Aβ deposits (Fig. 4E, red arrows).

Successful phagocytosis requires the uptake of Aβ into acidic phagocytic compartments. To test that, we conjugated our oligomeric Aβ with pHrodo green which is fluorescent at acidic pH [45]. To verify that pHrodo signal comes from pHrodo bound to Aβ we treated 2D hiMG with either just pHrodo or pHrodo-Aβ. No pHrodo signal was observed in the former group, confirming that all pHrodo signal comes from pHrodo-Aβ (Fig. 4F). hiNS were treated with regular oligomeric Aβ for 18 days followed by pHrodo-Aβ for an additional 7 days (4w Aβ total). Aβ treatment was then stopped prior to hiMG addition. hiNS were fixed 8 days later and imaged to quantify intracellular pHrodo-Aβ levels in the tissue. Co-staining for Aβ (4G8) confirmed that only intracellular pHrodo-Aβ was fluorescent in the tissue post-fixation. We found a strong uptake of pHrodo-Aβ inside IBA1 positive hiMG, whereas only small amounts overlapped with GFAP-positive astrocytes/NPCs, confirming that the main cell types in hiNS(+) facilitating Aβ phagocytosis are microglia/macrophage-like hiMG (Fig. 4G, H).

Time-dependent neuroprotective properties of hiMG

We next sought to elucidate neuroprotective properties of hiMG in response to chronic Aβ. hiNS were immunostained for neurons (MAP2), astrocytes/NPCs (GFAP), and microglia (IBA1) markers in all six experimental conditions. First, we quantified the number of hiMG (manual counting of IBA1 positive cells within the volume of the z-stack) inside hiNS(+) and found that the number of hiMG within 5w Aβ hiNS(+) increased by more than two-fold compared to ctrl hiNS(+), whereas the number of microglia in 3w Aβ hiNS(+) did not increase (Fig. 5A). This may be due to more infiltration of hiMG into hiNS as a result of more severe Aβ stimulus in 5w Aβ hiNS(+). Additionally, we found that GFAP immunofluorescence increased significantly in the presence of hiMG at 3w and 5w, indicating microglial-dependent astrocyte reactivity by Aβ (Fig. 5A). We repeated roGFP1 redox imaging following viral transduction using our hSyn-roGFP1 construct. Live confocal microscopy, performed every few days, allowed us to follow neuronal redox states during the chronic Aβ treatment with or without hiMG present. For hiNS that received 3w Aβ treatment, no significant Aβ induced oxidative stress was detected at DIV 54, but was evident by significantly increased roGFP1 ratios at the end of the treatment window at DIV59, (****p < 0.0001) (Fig. 5B, white arrows). Interestingly, oxidative stress was completely prevented by the presence of hiMG in 3w Aβ hiNS(+), suggesting robust neuroprotective functions of hiMG (Fig. 5B). However, Aβ induced oxidative stress at 5w was severe and not significantly reduced after hiMG infiltration (Fig. 5B, bottom part). These data suggest that hiMG can reduce Aβ-induced oxidative stress if added early enough during chronic Aβ treatment, but fail to recover or reverse severe oxidative stress late during Aβ treatment. Similar to roGFP1 live imaging, we monitored neuronal activity using GCaMP6f. We found that with 3w Aβ treatment, calcium wave frequencies remained similar to ctrl hiNS at DIV 50, but were reduced at the end of the 3w Aβ treatment window in the absence of microglia (**, p < 0.01) at DIV 58 (Fig. 5C, D). Matching the neuroprotective properties of hiMG, ameliorating oxidative stress in 3w Aβ hiNS(+), we found that hiMG fully rescued neuronal activity in 3w Aβ hiNS(+), confirming their neuroprotective ability preventing the decline of neural activity. At 5w Aβ, calcium waves were completely absent without ameliorating effects of hiMG, suggesting that functional recovery is only possible if hiMG are present early during chronic Aβ treatment (Fig. 5D).

Quantification of calcium wave frequencies for all groups at DIV 50 and 58 are shown in Fig. 5E. At both time points, we observe a reduction in frequency with the addition of hiMG in ctrl hiNS(+), as shown in Fig. 2G. At DIV 58, neuronal activity in 3w Aβ hiNS(−) is significantly reduced while in 3w Aβ hiNS(+) it remains at frequencies close to ctrl hiNS(+) (Fig. 5E). Neuronal activity in ctrl hiNS and 3w Aβ hiNS(+) remained consistent throughout DIV 50 to DIV 58 while in 3w Aβ hiNS(−) it declined (Fig. 5F).The functional recovery by hiMG in hiNS exposed to chronic Aβ correlates with their high capacity to remove plaque-like aggregates in 3w Aβ hiNS(+) (Fig. 4B) suggesting a potential link between Aβ phagocytosis and neuroprotection by hiMG.

We then determined whether hiMG prevented neuronal death by staining for c-Cas- 3 and quantifying nuclei in hiNS. While we found only a moderate increase in c-Cas- 3 immunoreactivity at 3w, it was significantly reduced in 3w Aβ hiNS(+) (*p < 0.05) with an even larger difference at 5w Aβ hiNS (***p < 0.001). Note that hiMG were identified by lectin staining, and their nuclei count was subtracted from whole-tissue nuclei quantification. Cell density in 5w Aβ hiNS(−) was dramatically reduced, and partially but highly significantly rescued in 5w Aβ hiNS(+) (***p < 0.001) (Fig. 5G).

Aβ aggregation and accumulation in the AD brain are believed to precede and trigger subsequent Tau hyperphosphorylation [46]. A well characterized phosphorylation site is threonine 181 (pTau181) which we immunostained for in tissue from all six experimental groups (Fig. S7 A), together with a total Tau co-staining. The ratio of pTau181/total Tau was calculated to test whether Aβ or microglia impact Tau hyperphosphorylation in our model. While we detected small but significant changes in ratios due to the presence of hiMG, we did not detect any significant increase due to Aβ with or without hiMG present (Fig. S7B). The lack of Aβ induced Tau hyperphosphorylation in our model was further confirmed by ELISA on supernatant of hiNS (Fig. S7 C) suggesting that the severe Aβ-induced pathology we observed does not result in Tau pathology in our embryonic model system.

Taken together, these data indicate that the neuroprotective potential of hiMG during chronic Aβ exposure is highly time-dependent: At 3w Aβ, added hiMG preserved neuronal function, prevented oxidative stress, and reduced Caspase- 3 cleavage, while at 5w Aβ, hiMG did not rescue neuronal function or reduce oxidative stress but hiMG did significantly reduce apoptosis and loss of cell nuclei, demonstrating limited neuroprotection by hiMG at late stages of amyloidosis. Visual inspection revealed hiMG attached to and accumulated around hiNS prior to infiltration, and 5w Aβ treated hiNS displayed morphological changes indicating tissue degeneration (Fig. S8) matching our immunofluorescence staining which revealed substantial tissue deterioration in 5w Aβ hiNS (Fig. 5A).

We further investigated the electrophysiological properties of hiMG in Aβ treated hiNS. Membrane capacitances were significantly increased at 5w Aβ (Fig. S9), but no other changes were significant.

hiMG alter astrocyte gene expression profiles in hiNS in response to Aβ

To elucidate potential mechanisms underlying the protective effects of microglia in hiNS chronically exposed to Aβ, we performed snRNA-seq on all six hiNS culture conditions (ctrl, 3w Aβ, 5w Aβ with or without hiMG). We isolated and sequenced single cell nuclei using the 10X Genomics platform and put the resultant transcriptomes through our pipeline [20,21,22,23,24]. After filtering the datasets for low quality cells, we merged these transcriptomes with those of the previously analyzed control hiNS (as in Fig. 1C, E).

UMAP visualization and analysis of the datasets of origin showed that the datasets merged well (Fig. 6A, B). Marker gene analysis defined all of the same cell types that were seen in the ctrl hiNS cultures including neural and glial precursor cells, astrocytes, inhibitory and excitatory neuron lineages, and microglia/macrophages. However, analysis of the datasets of origin identified differences in cellular composition that were most obvious in the 5w Aβ hiNS. Specifically, in the 5w Aβ hiNS(−), the only surviving cells were astrocytes and inhibitory neurons, while excitatory neurons were almost entirely absent which is consistent with the dramatic reduction of nuclei numbers in this group. The number of nuclei per cell type in each treatment group is outlined in Fig. 6C. This observation corroborates the known finding that excitatory neurons are more vulnerable in AD than inhibitory neurons [47, 48]. The 5w Aβ hiNS(+) contained more excitatory and inhibitory neurons, but the overall cell density was reduced relative to the 3w Aβ cultures and ctrl groups (Fig. 6B, C). In addition, relative to the ctrl and 3w Aβ hiNS(+), there were proportionately more hiMG in 5w Aβ hiNS(+), likely reflecting increased proliferation or infiltration of hiMG matching our quantification of IBA1 positive microglia/macrophage-like cells (Fig. 5A). By contrast, the distribution of cell types in the 3w Aβ hiNS cultures were similar to that seen in the ctrl groups.

We first compared astrocytes in these different culture conditions by subsetting and reanalyzing transcriptomes from the astrocyte cluster (cluster 0). UMAP cluster and dataset of origin visualizations (Fig. 6D, E) identified four clusters containing variable proportions of astrocytes from the different culture conditions. Most astrocytes that came from cultures without added microglia were localized to cluster 0b regardless of whether they were exposed to control peptide or Aβ. Conversely, most astrocytes from cultures with hiMG were localized to cluster 0a. The exceptions to this intermingling were cluster 0 d, which was almost completely comprised of transcriptomes from the 5w Aβ hiNS(+) cultures, and cluster 0c, which was mostly comprised of astrocytes from the control hiNS(+) cultures (Fig. 6E, F).

These data suggest that astrocytes in the different cultures were transcriptionally distinct, forming noticeable sub-clusters. We therefore assessed how Aβ altered astrocytes in the presence of hiMG by performing DEG analysis comparing the 5w Aβ hiNS(+) cells with ctrl hiNS(+). This analysis identified 288 DEGs when comparing the two conditions, with 209 enriched in 5w Aβ hiNS + cells and 159 in ctrl hiNS(+) without Aβ (> 1.3-fold average change, p adj < 0.05; Suppl. Tables 6 and 7). Single-cell gene expression heatmaps (Fig. 6G) showed that only a few of these mRNAs were upregulated in the 3w Aβ hiNS(+) astrocytes. Notably, gene expression overlays showed that some of the most highly upregulated genes following 5w Aβ exposure have previously been associated with the astrocyte response in AD, including APOE, CLU, SPP1, TLR4, VIM, S100 A11 and LRP1 (Fig. 6H). Gene ontology comparing mRNAs upregulated in 5w Aβ hiNS(+) relative to ctrl hiNS(+) astrocytes (Suppl. Table 8) showed that the top-ranked term in the molecular function category was Aβ binding (Fig. 6I). By contrast, genes that were significantly increased in ctrl hiNS(+) astrocytes were highly associated with gene ontology terms like neurogenesis and postsynaptic membrane (Suppl. Table 9), suggesting that the presence of Aβ interferes with normal astrocyte biology and cellular interactions.

Next, we compared astrocytes expression profiles, with or without hiMG addition, at 3w as well as 5w, following chronic Aβ exposure, and plotted the results in volcano plots (Fig. 6J, K, Suppl Tables 10 and 11). The expression of AD-associated astrocyte genes depends on the presence of hiMG during Aβ exposure, including genes such as GFAP, APOE, and VIM. Interestingly, we found DEGs linked to oxidative stress handling (OXR1, CCL2 and CHL1) to be upregulated only in 3w Aβ hiNS(+), the experimental group in which we observed a significant reduction in Aβ induced oxidative stress (Fig. 5B). Thus, in hiNS(+) exposed to Aβ for 3 weeks, hiMG increase the expression of subsets of astrocyte marker genes associated with AD as well as the expression of genes handling oxidative stress.

hiMG drive unique expression profiles in neurons, including APOE

Similar to the subsetted astrocyte snRNA-seq analysis (Fig. 6), we subsetted the neuronal populations (both excitatory and inhibitory) for all six experimental groups (Fig. S10 A). Total neuronal nuclei count reflected the severe cell death caused by 5w Aβ exposure, particularly in the absence of hiMG (Fig. S10B). The very small population of neurons in 5w Aβ hiNS(−) makes it challenging to perform any direct comparisons to this treatment group.

Interestingly, we found that the presence of hiMG during 5w Aβ exposure significantly increased neuronal APOE, SPP1 and FTL expression (Fig. S10 C,D, Suppl. Table 12), similar to the hiMG effect on astrocytes (Fig. 6). Since we only observed functional neuroprotection by hiMG during 3w Aβ exposure (Fig. 5), we isolated neuronal DEGs in 3w Aβ hiNS(+) (Suppl. Table 13) and 5w Aβ hiNS(+) (Suppl. Table 12) to potentially identify pathways linked to neuroprotection and found 228 and 711 upregulated genes respectively. Gene ontology analysis for 3w Aβ hiNS(+) (Suppl. Table 14) suggests biological processes linked to maintenance of cell–cell and synaptic signaling matching the preserved neuronal activity we observed in the presence of hiMG while 5w Aβ hiNS(+) revealed significant upregulation of processes linked to extracellular exosomes, response to stress, regulation of programmed cell death and amyloid-beta binding (Fig. S10E, Suppl. Table 15). Despite the fewer DEGs in 3w Aβ hiNS(+), we performed several direct comparisons between treatment groups, similar to our astrocyte gene expression analysis (Fig. 6J, K), to identify more subtle genes of interests and graphed them in volcano plots (Fig. S10 F, Suppl. Tables 16–18). Genes were only plotted if minimum expression percentage exceeded 10% and the upper 10 top genes judging by fold change and/or adjusted pvalue were labeled by name. At 3w Aβ exposure, the timepoint at which the addition of hiMG reverses Aβ-induced oxidative stress and neuronal network dysfunction, is associated with unique gene expression profiles in the combined inhibitory and excitatory neurons. When directly comparing 3w Aβ hiNS(−) with 3w Aβ hiNS(+) we found that a number of AD-linked genes appeared upregulated in 3w Aβ hiNS(+). Several of the highest expressed genes have previously been associated with AD, including CCL2, VIM and STAT3 (Fig. S10 F). Another gene of interest in the direct comparison of 3w Aβ hiNS(+) with ctrl hiNS(+) is BDNF, which has been implicated in AD pathogenesis [49].

Our data indicate that hiMG drive several genes of interest following different exposure times to Aβ. These genes, including APOE, are of high interest for their possible direct role in Aβ-mediated neuronal dysfunction on our hiNS model, particularly as they might relate to differential vulnerability of excitatory and inhibitory neuronal populations in AD.

Considering our astrocytic and neuronal gene expression analysis together, we found larger changes in gene expression in 5w Aβ hiNS(+) astrocytes and neurons while we observed functional neuroprotection only at 3w Aβ hiNS(+) indicating that AD-related reactive astrocyte and neuronal gene expression changes in the surviving population correlate with a more severe pathology at 5w Aβ hiNS(+). This implicates that more subtle gene expression changes at 3w Aβ hiNS(+), both in neurons and astrocytes, should be explored in the future to identify signaling pathways of neuroprotection.

Heterogenous gene expression profiles in hiMG

After characterizing the shift in astrocyte and neuronal gene expression by hiMG during chronic Aβ exposure, we subsetted the microglia/macrophages (cluster 13, Fig. 6A) of the 5w Aβ hiNS(+) experimental group to investigate gene expression profiles in hiMG to better understand their reactive state. Due to less infiltration/proliferation of microglia in the ctrl and 3w Aβ hiNS(+) treatment groups, microglia quantities in those groups were not sufficient to perform comparative transcriptomic analysis.

In the 5w Aβ hiNS(+) treatment group, we identified 3 distinct clusters within the microglia/macrophage group which we identified further by DEG analysis (Fig. 7A, Suppl. Table 19). Cluster 13a showed similarities to a more macrophage-like population characterized by upregulation of CD36, TLR2, SYK, SLC11 A1 and PTPRC. Cluster 13c mimics gene expression of homeostatic microglia-like cells including upregulation of ITGAM, CX3 CR1 and P2RY12. Cluster 13b gene expression was similar to a disease associated microglia (DAM) [50]-like state, sharing a large number of DEGs including APOE, TREM2, CD9, LPL, CD9 and B2M, indicating that chronic Aβ exposure induces a similar activation state in a subset of hiMG in 5w Aβ hiNS(+) (Fig. 7B). We conclude that the microglia/macrophage population in Aβ treated hiNS(+) is heterogeneous and that each subpopulation could play distinct pathophysiological roles. We performed gene ontology analysis on all three clusters highlighting distinct biological processes of each individual group (Fig. 7C, Suppl. Table 20). Volcano plots highlighting DEGs genes (Suppl Table 21) of interest for each sub-cluster displayed in Fig. 7D (see heatmap with top 10 most significant DEGs in Fig. S11).

hiMG cytokine and ligand expression indicating signaling to neurons and astrocytes

To better understand the phenotypical changes of hiMG during chronic Aβ treatment, we analyzed cell culture supernatant of the above hiNS conditions for proinflammatory cytokine levels using a multiplex ELISA due to their high sensitivity and specificity. Supernatant was collected on the last day of the treatment protocol. Overall, cytokine levels were enriched in media of all hiNS(+) samples but not in hiNS(−), indicating that hiMG produce the listed cytokines independent of Aβ exposure. Note that IL- 8 data was excluded from the panel due to saturated readings for hiNS(+) groups, preventing accurate quantification. Interestingly, only IL- 6 and IL- 12 levels increased significantly (** p < 0.01) with 5w Aβ treatment, indicating an Aβ induced shift in hiMG cytokine production (Fig. 7E).

Our data indicate a significant neuroprotective impact of hiMG in our chronic amyloidosis model by reducing amyloid deposits, oxidative stress and preserving neuronal activity inside hiNS (Figs. 4 and 5). hiMG induce significant transcriptomic changes, particularly in astrocytes, mimicking human AD-astrocyte gene expression (Fig. 6), as well as neurons (Fig. S10). To identify potential pathways by which hiMG exert neuroprotection we performed ligand-receptor interaction analysis based on our 5w Aβ hiNS(+) snRNA-seq data set. DEG lists were generated to identify microglia/macrophage specific ligands (cluster 13), astrocyte specific receptors (cluster 0) and neuronal specific receptors (all neuronal clusters pooled Suppl. Table 22). Ligand-receptor DEG pairs were identified with each having at least 10% mean expression in the respective cell population (Fig. 7F, Suppl. Table 23). We identified 44 pairings between hiMG and astrocytes but only 11 between hiMG and neurons, matching our observations that hiMG dramatically change the transcriptional landscape of astrocytes (Fig. 6), with fewer changes observed on neuronal populations (Fig. S10), with specific pathways of interest highlighted.

APOE expression in hiNS is hiMG-dependent

APOE is well known for being the most important genetic risk factor for sporadic AD and its role in amyloid clearance [51]. The iPSC-line used in this study carries the common homozygous ε3 allele thought to be ‘neutral’ without increasing or decreasing AD risk. We found astrocytic and neuronal APOE expression to be upregulated in 5w Aβ hiNS(+) (Fig. 6G, H and Fig. S10 C,D) as well as in a subset of hiMG (Fig. 7B,D) from the same treatment group, suggesting widespread APOE expression across multiple cell types in 5w Aβ hiNS(+).

We next immunostained for APOE in fixed slices (80 µm thickness) of hiNS of all six experimental groups and found significant increases in immunoreactivity exclusively in 3w/5w Aβ hiNS(+), indicating that high expression levels of APOE appear only in Aβ-treated hiNS that also contain hiMG (Fig. 8A, B). Next, we investigated if APOE is present intracellularly in hiMG and/or astrocytes by co-staining for IBA1 and GFAP. We observed both APOE-positive hiMG and APOE-positive astrocytes with the majority of APOE apparently extracellular (Fig. 8B). We then tested if soluble APOE was detectable by western blot in supernatant from hiNS media of all six groups, and found APOE in all groups containing hiMG, while APOE levels in 5w Aβ hiNS(+) appeared significantly lower compared to ctrl and 3w Aβ hiNS(+). Analysis of soluble Aβ levels using ELISA, showed a reduction in 5w Aβ hiNS(+) compared to 5w Aβ hiNS(−), suggesting potential aggregation of Aβ-APOE complexes in this specific condition resulting in the large quantities of extracellular APOE in 5w Aβ hiNS(+) (Fig. 8C), and consequently reduced soluble levels. Additional co-staining for APOE and Aβ (4G8) of 5w Aβ hiNS(±) confirmed that extracellular APOE aggregates co-localized with extracellular Aβ at the outer tissue volumes of hiNS in a hiMG-dependent manner (Fig. 8D).

Our snRNA-seq analysis revealed APOE expression in microglia, astrocytes/NPCs as well as neurons in 5w Aβ hiNS(+). To compare expression levels between cell types we subsetted all neuronal, astrocytic and microglia/macrophage clusters from all three experimental groups containing hiMG (ctrl, 3w Aβ, 5w Aβ hiNS(+)). Plotting all expression levels side by side confirmed the specific increase in astrocytes and neurons in 5w Aβ hiNS(+) but also indicate that microglia/macrophages express most APOE, despite driving APOE expression in the other cell types (Fig. 8E).

To validate our APOE expression data in our AD neurosphere model, we examined APOE immunoreactivity patterns in human post-mortem AD brain sections to confirm known AD APOE expression [52]. To remove background and autofluorescence of paraffin-embedded human cryosections, we cleared and expanded the tissue, based on a recently developed protocol [53] prior to immunofluorescence, and stained sections from two familial AD as well as one sporadic AD patient. In all three patients, we found APOE co-localized with dense-core Aβ plaques, while co-staining for GFAP or IBA1 revealed APOE-positive astrocytes and microglia respectively in AD patients (Fig. 8F).

In summary, our 5w Aβ hiNS(+) model mimics APOE expression patterns in AD and revealed that microglia, while expressing the highest levels of APOE, also drive significant APOE expression in astrocytes and neurons. This tool offers unique opportunities to study cell type and APOE isoform specific functions in the future.

Discussion

Functional characteristics of hiNS demonstrate healthy brain physiology

Our protocol for the generation of hiNS relies on the differentiation of human iPSC-derived NPCs into neuronal and astrocyte-like cell populations into a spherical 3D neural tissue culture model (Figs. 1, 2). Our snRNA-seq analysis does not suggest the presence of mature oligodendrocytes in our cell culture system. Given the late differentiation of oligodendrocytes during neurodevelopment [54] and the known challenges in the field to obtain mature oligodendrocytes in 3D brain tissue culture models [55] this was an expected outcome and limitation of the model. We extensively characterized the physiological properties of hiNS and showed that neurons in hiNS maintain a healthy reduced redox status with high levels of synchronized wave-like neuronal calcium activity, matching previous reports of calcium waves in both brain development [56, 57] and other neurosphere cultures [58, 59]. In addition, we found that extracellular glutamate increases coincide with wave activity and that glutamate transporters EAAT1/2 are required to maintain glutamate balance implicating a direct role of astrocytes since pharmacological blockage of EAAT1/2 resulted in extracellular glutamate overload blocking wave activity (Fig. S2C). Our pharmacological data including application of TTX, AP5/CNQX, GABA, and picrotoxin suggest that both excitatory and inhibitory neurons contribute to calcium wave formation and dynamics matching other studies [59]. During brain development, not only do neurons and astrocytes play key roles, but infiltrating microglial cells into the brain are vital in shaping the neuronal network [60]. Few other studies [17, 61] have incorporated microglia into neurospheres using different protocols suggesting robust tissue infiltrating properties of iPSC-derived microglia/macrophages. hiMG readily infiltrate hiNS attached to Matrigel-coated surfaces, and we demonstrate that hiMG display microglia-like electrophysiological properties similar to mature in situ ramified mouse or human microglia [33, 35, 62]. Moreover, we show that the presence of microglia shapes neuronal calcium wave activity kinetics and also alters the transcriptional landscape of other cell types in hiNS (Fig. 1). Our gene expression analysis of astrocytes in hiNS with vs. without hiMG revealed an increase of gene expression linked to mitochondrial function and oxidative phosphorylation, indicating increased metabolic function in the presence of microglia. We believe that our hiNS offer unique opportunities to investigate neuron-glia interaction in human brain development.

Chronic Aβ exposure models plaque-like aggregation and neurotoxic effects

Instead of utilizing familial AD mutations in APP or PS1 to drive Aβ pathology, we modeled chronic Aβ exposure by supplementing hiNS media with high concentrations (compared to endogenous Aβ levels in the CSF [63] of oligomeric Aβ ~ 0.24 µg/mL) for 3 to 5 weeks to induce progressive neurotoxicity (Fig. 3 and Fig. S6). We found that oligomeric rather than fibrillar Aβ was more suitable for inducing plaque-like aggregates inside neurosphere tissue within 7 days of media supplementation. Accumulation of Aβ in the brain of AD patients is believed to be one of the earliest pathological events, occurring decades before clinical symptoms [64]. We did not observe any detectable pathological impact within the first 2 weeks of Aβ exposure in hiNS, suggesting that chronic exposure is required to trigger neurotoxicity, mimicking Aβ-driven neurodegeneration in the human brain.

At the end of 3-week long exposure to Aβ we found that calcium wave frequency decreased and that oxidative stress became evident (Fig. 3). Oxidative stress is a known early hallmark of AD believed to precede the loss of neurons [65, 66]. It can be observed by immunofluorescence staining of post-mortem AD brain tissue [67]. Here we demonstrate how chronic Aβ exposure over time leads to the build-up of oxidative stress in neurons, which precedes the severe cell death observed after 4–5 weeks of Aβ treatment. Similarly, network dysfunction and synaptic dysregulation [68] are early known features of AD pathogenesis and are believed to precede neurodegeneration and coincide with the formation of Aβ induced oxidative stress in hiNS. Our immunofluorescence staining of hiNS after long Aβ exposure indicates an increase in cleaved Casp- 3 immunoreactivity and TUNEL fluorescence, implicating apoptosis as a key cell death pathway. Although caspase-dependent cell death has been proposed as a neuronal death pathway in AD [69, 70], multiple other pathways have more recently been proposed to play a role in later stages of disease progression [71]. More research is needed to decipher the exact cell death pathways in our Aβ treated hiNS, but our model offers a slow-progressing pathology to study key AD disease mechanisms in a human 3D tissue environment.

Impact of hiMG on Aβ clearance and neuroprotection

While microglia have gained substantial attention in the field over the past 10 years, their precise role in AD pathogenesis remains elusive[72,73,74]. During the initial stages of AD, microglia are generally believed to release anti-inflammatory cytokines and neurotrophic factors, and are efficient in phagocytosing Aβ later in AD pathology microglia switch to producing pro-inflammatory cytokines, actively contribute to neurodegeneration and have a reduced capacity to clear Aβ from the brain [75,76,77]. In our chronic amyloidosis model system, we found that hiMG markedly reduced Aβ levels in the tissue of 3w Aβ hiNS(+) and were found to effectively phagocytose Aβ (Fig. 4), further highlighted by positive CD68 immunoreactivity, a marker for phagocytic microglia [44]. However, we found that only in 3w but not 5w Aβ hiNS(+) they significantly reduced Aβ levels in the tissue indicating that microglial phagocytosis reached saturation and their capacity to effectively reduce aggregated Aβ at 5w Aβ hiNS(+) is impaired. Other groups have reported that human iPSC-derived microglia are capable of phagocytosing Aβ [78, 79] matching our findings.

In AD, microglial phagocytic function is generally believed to be dysfunctional due to either senescence [80] or an exhaustion due to chronic long-term Aβ stimulation [81]. Both are aspects we cannot mimic with the here presented approach due to the limited time of hiMG in the co-culture with hiNS as well as the developmental phenotype of both neurons and microglia due to their iPSC-derived origin. Our findings therefore shed light on the initial response of naive microglia to different severities of Aβ deposition.

We have demonstrated that hiMG, within a 10-day time window, prevent oxidative stress and sustain neuronal calcium wave activity early during chronic Aβ treatment. However, when hiMG were introduced at a later stage, loss of neuronal activity or severe oxidative stress did not recover, although overall cell death in hiNS was markedly reduced, highlighting that robust neuroprotection could persist (Fig. 5B–G). Quantifying levels of a series of pro-inflammatory cytokines from the supernatant of all treatment groups revealed that all measured cytokines were produced in all hiNS(+) groups. However, typical pro-inflammatory cytokines, like IL1β and TNFα, were not elevated in 3w/5w Aβ hiNS(+) when compared to ctrl hiNS(+). Indeed, we saw a robust release of these cytokines (Fig. S3 C) when hiMG were stimulated with LPS, which is known to induce pro-inflammatory cytokine production [82]. We conclude that the production of cytokines in ctrl hiNS(+) marks the baseline levels without indicating a pro-inflammatory phenotype.

The only two cytokines significantly increased with Aβ treatment were IL- 6 and IL- 12 (Fig. 7E) and their role in neuroinflammation has been context depended suggesting both neuroprotective and neurotoxic properties depending on conditions. IL- 6 has been implicated in neurogenesis [83] and shown to be protective during ischemia [84] and was further linked to a proinflammatory response in AD [85] suggesting a more detrimental role. IL- 12 has been shown to reduce neuroinflammation and facilitate neuroprotection [86] and was implicated in slower cognitive decline in older adults [87]. Its reduction in AD mouse models were shown to ameliorate AD pathology [88] even though it is generally characterized as a proinflammatory cytokine [89].

Defining the definite roles of these cytokines should be addressed in future studies to shed light on which cytokines have a clear neuroprotective function in our human AD model system.

We conclude that hiMG acquired an overall neuroprotective phenotype in our chronic Aβ model system. In future studies, actively priming hiMG to a pro-inflammatory state, for instance by LPS, prior to adding them to hiNS would be an interesting avenue to pursue.

A comparable cell culture approach to induce Aβ toxicity in a human 3D neural tissue was presented by Takata et al. [90]. Utilizing a modified, highly cytotoxic modified Aβ (at similar concentrations to ours), the authors induced neurotoxicity which was rescued with the addition of microglial cells. However, the impact of Aβ on neural function or transcriptomic changes in the tissue was not a focus of the study.

Recently, Jorfi et al. [17] have demonstrated in a similar 3D cell culture model that the addition of iPSC-derived primitive macrophages during Aβ treatment resulted in less Aβ aggregation and better survival of neurons, indicating that this activation state of microglia-like cells is consistent between similar model systems. Modelling interaction with additional infiltrating peripheral immune cells is a possible requirement to trigger robust pro-inflammatory phenotypes. Blood–brain-barrier (BBB) breakdown leads to infiltration of soluble factors and cells from the periphery and is a known hallmark of AD [91], this process is not modelled in the current study. Most recently, in an amyloid-overproducing 3D tissue culture model comprised of neurons, astrocytes and microglia, the addition of peripheral immune cells, particularly CD8⁺ T cells, greatly facilitated Aβ induced neurodegeneration; although the addition of microglia alone appeared to have neurotoxic effects [17], matching previous observations using the same model system [61]. It is therefore reasonable to suggest that our hiMG could be polarized to a more neurotoxic function by further modelling aspects of BBB breakdown or interaction with peripheral immune cells.

AD-related astrogliosis and astrocyte activation states depend on hiMG

Cytokines produced by microglia are known to shape astrocyte activation states [92, 93]. We found that the addition of hiMG to hiNS increases the soluble level of a range of different cytokines in the cell culture media (Fig. 7E) and we found hiMG to be able to change the astrocytic transcriptomic state under control conditions (Fig. 1). We hypothesize that microglial cytokines are the key driver in shaping astrocyte function in our neurosphere model system but further experiments are needed to identify the mechanism by which hiMG alter the transcriptional profile during chronic amyloidosis.

Our immunofluorescence staining revealed a dramatic increase in reactive microglia and GFAP immunoreactivity in 5w Aβ hiNS(+), suggesting gliosis triggered by Aβ exposure (Fig. 5A). Reactive astrocytes are known to play both beneficial and detrimental roles in AD pathology [94]. To understand the reactive state of astrocytes in our AD model system we performed snRNA-seq analysis on Aβ treated hiNS with or without hiMG, comparing gene expression levels to control hiNS.

We found that the presence of hiMG shifted the transcriptional profile of the astrocyte cluster in Aβ treated groups. Interestingly, 3w and 5w Aβ hiNS(+) astrocytes appeared to have unique transcriptional profiles (Fig. 6D, E). DEG analysis comparing astrocytes of 5w Aβ vs. ctrl hiNS(+), to examine chronic Aβ-induced transcriptional changes exclusively in the presence of hiMG, highlighted significant upregulation of a number of genes associated with reactive astrocytes in human AD: 1) LRP1 is a receptor for APOE and has been associated with neuroprotection in animal models of AD [95, 96], 2) CLU is an extracellular chaperone linked to neuroprotection via Aβ aggregation and clearance by astrocytes [97, 98], and 3) VIM was reported to be upregulated in AD although its role in pathology remains elusive [99,100,101]. Significant gene ontology processes in 5w Aβ hiNS(+) astrocytes included Aβ binding, phagocytosis and endocytosis, indicating that, in the presence of microglia, astrocytes become reactive and could participate in Aβ clearance.

Next, we analyzed astrocytes comparing DEGs of hiNS(−) vs hiNS(+) after 3w/5w Aβ treatment separately to identify more microglia-dependent astrocyte genes. Many DEGs were re-identified to the previous 5w Aβ hiNS(+) vs ctrl hiNS(+) comparison such as VIM, SPP1, LRP1 and S100A11, highlighting the importance of hiMG in driving astrocyte gene expression (Fig. 6J, K). Interestingly, GFAP was significantly upregulated in hiNS(+) astrocytes compared to their hiNS(−) counterpart in both 3w and 5w Aβ comparisons. Note that GFAP was not differentially expressed when comparing 5w Aβ hiNS(+) to ctrl hiNS(+) astrocytes earlier, although our immunofluorescence analysis strongly suggests an increase in GFAP immunoreactivity indicating astrocyte activation (Fig. 5A). Our snRNA-seq quantification of astrocyte numbers (Fig. 6C) indicates that 5w Aβ hiNS(+) harbors the largest population of astrocytes compared to all other groups. This however could be in parts because of a bias towards collecting astrocyte nuclei rather than neuronal nuclei due to the evident neuronal loss. The increased GFAP immunoreactivity could be a result of either or both an increase in astrocyte numbers or increase in GFAP expression. This highlights possible limitations of snRNA-seq, as it only captures mRNA still inside the nuclei at the time of tissue collection, not the full transcriptome [102]. Upregulation of GFAP in AD-associated astrocytes is a well-known biomarker for AD [100, 103, 104] and is believed to be associated with a neuroprotective reactive phenotype [105, 106].

A recent study described a loss of homeostatic gene expression in AD astrocytes, including downregulation of NRXN1 and CADM2 [107]. Both genes were significantly downregulated in astrocytes of 3w/5w Aβ hiNS(+) compared to hiNS(−), indicating hiMG dependence. Another gene of interest that was downregulated is SLC7A11. Since its upregulation was recently associated with neurotoxic functions of astrocytes [108], this further supports a neuroprotective astrocyte phenotype in the presence of hiMG.

The presence of hiMG in 3w Aβ hiNS dramatically ameliorated Aβ induced oxidative stress and prevented decline of neural activity (Fig. 5B, C). Interestingly, genes linked to oxidative stress response in astrocytes (OXR1, CCL2, CHL1) were significantly upregulated with hiMG, indicating astrocytes could play a role in reducing neuronal oxidative stress independent of Aβ clearance. OXR1 increases oxidative stress resistance [108]. A recent study found upregulation of OXR1 following omaveloxolone treatment in a mouse model of AD linked to markedly reduced AD pathology [109]. OXR1 therefore could be an interesting target for future studies.

Gene expression changes in surviving neuronal populations after Aβ treatment

Transcriptional changes in the presence of hiMG during Aβ exposure was not limited to astrocytes but extended to neuronal populations (Fig. S10). Most strikingly, we found APOE, SPP1 and FTL to be highly upregulated in the surviving 5w Aβ hiNS(+) neuronal populations (Fig. S10 C,D). APOE will be discussed separately below. SPP1 expression is more commonly observed in immune cells, and elevated FTL levels have been observed in the AD brain [110]. Due to the almost complete death of all excitatory neurons in 5w Aβ hiNS(−) we restrained from attempting any direct 5w Aβ hiNS(+) vs hiNS(−) comparisons in neuronal gene expression patterns. Since we observed functional neuroprotection at 3w Aβ hiNS(+) we investigated DEGs in this experimental group and GO analysis revealed significant hits for terms associated with cell–cell and synaptic signaling, suggesting that the presence of hiMG preserves functional synaptic activity despite the presence of Aβ. When performing GO analysis on the surviving neurons of 5w Aβ hiNS(+) instead, we identified significant terms associated with Aβ binding or the response to stress and programmed cell death (Fig. S10E). While we did identify more DEGs in neurons of 5w Aβ hiNS(+) than in 3w Aβ hiNS(+), we performed direct group comparisons of ctrl/3w Aβ hiNS with or without hiMG present to identify potential genes of interests and neuroprotective pathways during amyloidosis. When directly comparing 3w Aβ hiNS(−) with 3w Aβ hiNS(+) we identified genes of interest in AD to be upregulated in the presence of hiMG: CCL2, as already mentioned above, is implicated in AD but its precise role in terms of neuroprotection or neurotoxicity remains elusive [111, 112]. STAT3 is a transcription factor which was shown to be activated by Aβ and is implicated in inflammation driving apoptosis in the AD brain [113, 114]. We speculate that the presence of hiMG, or a subpopulation of them, drives the STAT3 pathway in neurons mimicking aspects of neuronal AD gene expression. VIM, as discussed above, is known to be expressed in astrocytes in AD, but has been reported to be expressed in neurons early in the AD brain as part of a damage response mechanism [115].

In conclusion, we believe that further research into astrocytic and neuronal gene expression changes in 3w Aβ hiNS(+) could uncover novel underlying mechanisms of neuroprotection in AD.

hiMG drive astrocytic and neuronal APOE expression during chronic amyloidosis

One of the most interesting DEGs we identified, and discussed here separately from the others mentioned above, is APOE. Our data indicate that microglia could be required to induce astrocyte AD gene expression profiles and speculate that hiMG cytokine production could be the trigger for APOE expression in astrocytes since a crosstalk of cytokines and APOE expression has been reported in various tissues and model systems [116].

One of the most significantly upregulated genes in 5w Aβ hiNS(+) astrocytes was APOE, which is known to be the largest genetic risk factor for sporadic AD and is expressed predominantly by astrocytes, microglia as well as neurons under stress in the human AD brain [51, 117]. Surprisingly, our results show that chronic Aβ exclusively after 5w Aβ exposure, and not 3w, induced upregulation of APOE depends on the presence of hiMG. Much in the same way, hiMG also increased APOE expression in neuronal populations, indicating that specifically during long chronic Aβ exposure microglia are key drivers in upregulating APOE expression in other cell types in the tissue. APOE is closely involved in amyloid clearance, and its alleles ε2 and ε4 are extensively studied for their protective and deleterious roles, respectively [118]. The iPSC-line used in this study carries homozygous ε3 alleles not associated with any changes in AD risk. When analyzing the transcriptional profile of hiMG, we found that hiMG in 5w Aβ hiMG(+) sub-cluster into three distinct populations (Fig. 7) including one group resembling a unique transcriptional profile of disease associated microglia [50] which includes the expression of APOE and TREM2, two genes extensively studied in AD research.

Next, we compared APOE expression levels between the three cell types in all hiNS(+) groups and found that the population of microglia/macrophage-like cells showed higher APOE expression levels compared to astrocytes or neurons, which were similar to each other in expression levels (Fig. 8E). Interestingly, we did identify obvious extracellular APOE aggregates in the tissue of 3w Aβ hiNS(+) (Fig. 8A) although APOE gene expression in this group was not elevated in astrocytes or neurons suggesting that the soluble levels of APOE were produced by hiMG and required the presence of Aβ to manifest as aggregates in the tissue. It is generally believed that astrocytes express most APOE in the healthy brain, while gradually microglial APOE expression becomes more dominant in AD matching our expression pattern [119].

hiMG are required for APOE-Aβ co-localization

APOE remains the most important risk gene for sporadic AD [51, 120, 121]. Despite this fact, the mechanisms by which each APOE expressing cell in the brain impacts AD pathogenesis is poorly understood. APOE is a secreted protein which directly binds Aβ and has been implicated in both its aggregation and clearance, depending on the genetic isoform of APOE and stage of AD pathology [122,123,124]. Besides its role in clearance and aggregation of Aβ, it was established over 20 years ago that amyloid plaques co-localize with extracellular APOE in the human AD brain marking a well-known disease hallmark [52] and was later confirmed in AD mouse models [125] as well. We have confirmed APOE-Aβ co-localization in plaques of AD human postmortem tissue (Fig. 8). Interestingly, in our here presented 3D human amyloidosis model we only detected APOE-Aβ interaction if hiMG were present during the Aβ treatment, particularly 5w and to a lesser extend after 3w long exposure, mimicking the colocalization found in the AD brain. Note that soluble APOE and Aβ levels were equal or higher in 3w Aβ hiNS(+) compared to 5w Aβ hiNS(+) (Fig. 8C, Fig. S12) but the number of hiMG was substantially higher in 5w compared to 3w Aβ hiNS(+) (Fig. 5A) further supporting the importance of microglial/macrophage presence for APOE-Aβ to aggregate in the tissue. The functional significance of this co-localization is not well understood but has gained more attention in the recent literature. The significance of APOE-Aβ co-aggregation in the pathogenesis of AD was recently illuminated utilizing single molecule imaging suggesting that co-aggregation impacts the amount of Aβ clearance by microglia [126]. To our knowledge, the only evidence for the requirement of microglia for the co-localization of Aβ and APOE at plaques come from studies utilizing mouse models from the Green lab, suggesting that the depletion of microglia cells reduce APOE levels in the CSF and reduce co-localization with Aβ plaques [127, 128].

Amyloidosis triggered AD-like pathology independent of Tau hyperphosphorylation

During AD pathogenesis, it is generally believed that Aβ accumulation and aggregation is one of the early hallmarks of AD and upstream of hyperphosphorylation and aggregation of Tau [1, 129, 130]. We did not find evidence of an Aβ driven increase in Tau hyperphosphorylation, neither with or without the addition of hiMG (Fig. S7), by either measuring pTau levels in the tissue or the supernatant. Moreover, we did not find any significant changes in expression of kinases linked to Tau hyperphosphorylation, such as GSK- 3b, CDK5, PKA or MAPK, in our snRNA-seq data sets, supporting this finding. Our immunofluorescence staining for pTau181 generally indicated bright immunoreactivity in all hiNS, including ctrl hiNS. The fact that iPSC-derived neurons are young neurons in early developmental stages and that Tau phosphorylation is a known neurodevelopmental phenotype [131], it is possible that any Aβ induced hyperphosphorylation of Tau is difficult or impossible to detect in our model system. We therefore conclude that our tissue culture model reflects heavy amyloidosis driven features of AD pathology, independent of Tau hyperphosphorylation. Other human 3D tissue culture models in vitro have reported a Tau pathology but utilized cells derived from fAD or sAD patients to drive AD pathologies and not a chronic synthetic Aβ stimulation paradigm we use here [132,133,134,135]. We conclude that synthetic Aβ supplementation is not sufficient to drive additional Tau pathology in our model.

Pathway analysis of hiMG ligands matching astrocytic and neuronal receptors

To identify potential pathways of how hiMG affected astrocytic and neuronal function in our 5w Aβ hiNS(+) group, we performed ligand-receptor analysis on our snRNA-seq data (Fig. 7F). To highlight the most specific ligands and receptors, we exclusively analysed ligand/receptor DEG pairs with a minimum mean expression of > 10%. While determining the exact signaling mechanisms with pharmacological experiments or genetical knockdowns of specific candidates exceed the scope of this manuscript, we here highlight and discuss the most promising pathways in respect to their reported relevance to AD pathogenesis.

We first focused on APOE (pathway highlighted in red, Fig. 7F). With microglia/macrophages displaying the highest APOE expression in 5w Aβ hiNS(+) (Fig. 8E), we were not surprised that it was identified as a DEG, despite its expression in astrocytes and neurons. Two receptors, LRP2 (with a relatively low expression rate) and SORL1 (with a moderate expression rate) were identified on astrocytes but none on neurons, suggesting that soluble APOE directly acts on astrocytes. SORL1 is a known AD risk gene and its signalling pathways are more characterized in neurons, rather than astrocytes [136, 137]. In a most recent preprint article, APOE-SORL1 was also shown in LR analysis using spatial transcriptomics on human AD autopsy material [138]. It was further suggested that SORL1-TREM2-APOE signalling is strongly linked to Aβ clearance indicating a functional role of astrocytes in Aβ clearance pathways in our model system [139]. LRP2 expression in astrocytes has also been linked to Aβ clearance [140] and supports this assumption.

Next, we focused on SPP1 (pathway highlighted in dark green, Fig. 7F), expressed by microglia/macrophages, with the matching receptor genes CD44 and ITGAV on astrocytes and ITGA9 on neurons. SPP1 encodes for the soluble protein osteopontin (OPN) which we found to be upregulated in astrocytes in the presence of hiMG, presumably independent of Aβ in ctrl (Fig. 1), 3w and 5w Aβ hiNS(+) (Fig. 6). Note that SPP1 was also differentially expressed in the DAM-like microglia/macrophage population (Fig. 7D, cluster 13b volcano plot), the same cluster expressing APOE and TREM2. OPN has been recently suggested as an early AD biomarker [141] and its deletion or reduction through anti-OPN antibodies was reported to reduce plaque burden in AD mice suggesting an overall neurotoxic function during AD pathogenesis [142]. One of its well described receptors is CD44 and highly expressed in astrocytes of AD patients [143], which we found to be upregulated by the presence of hiMG with and without Aβ exposure. A second OPN receptor we found to be expressed on astrocytes is Integrin-α5 (encoded by ITGAV), which was shown to exert neuroprotective functions in a mouse model of AD [144]. Another OPN receptor Integrin-α5 (encoded by ITGA9), part of the integrin family, was found to be expressed on neurons suggesting that OPN signaling is not limited to microglia/macrophage crosstalk to astrocytes and involves multiple members of the integrin receptor family.

Another microglia/macrophage ligand we identified, with different receptor genes on astrocytes (EGFR) and neurons (ERBB4), is HBEGF (pathways highlighted in light green, Fig. 7F). HBEGF encodes for heparin-binding epidermal growth factor (HB-EGF) and was suggested to be an endogenous neuroprotective molecule in the brain [145] and inhibition of HB-EGF was reported to facilitate Aβ expression in rats [146]. The role of its receptors we identified in our data set is more documented to be involved in AD. Inhibiting EGFR was shown to exert neuroprotection in AD [147] highlighting it as a novel potential target for AD [148] overall suggesting that EGFR plays a neurotoxic role in AD pathogenesis [149]. On the neuronal side, we identified ERBB4 (encoding for ErbB) which was shown to be expressed in neurons of AD mice [150] and was also associated with neurotoxicity [151]. The conflicting literature on HB-EGF, as a neuroprotective factor, and the neurotoxicity associated with the receptors EGFR/ErbB further stresses the need of novel model systems to investigate these pathways in more depths and their therapeutic potential for AD treatment.

We further identified IL1B (encoding for IL1β, (pathway highlighted in blue, Fig. 7F)) as a differentially expressed microglia/macrophage ligand matching the genes IL1RAP (on astrocytes) and IL12RB2 (on neurons). The role of IL1β signaling in AD is well documented as a pathological factor [152] and reported to be upregulated in AD, driving astrocyte activation and neuroinflammation overall suggesting a neurotoxic contribution in AD [153, 154], potentially acting through astrocytes and neurons in our experimental model.

The role of Insulin-Like Growth Factor − 1 (IGF- 1) (encoded by IGF1, identified ligand expressed in our microglia/macrophage population in low levels, (pathway highlighted in brown, Fig. 7F)) signaling in neuroinflammation and AD is controversial and context dependent. While an anti-inflammatory role promoting neuronal survival through its receptor, IGF- 1R (encoded by IGF1R, identified ligand expressed in our astrocyte population in moderate levels) on neurons and astrocytes is well supported [155] another study reports that blocking IGF- 1R results in lowered Aβ levels and reduced microglia activation in a mouse model of AD [156].

Another potential crosstalk implicated in AD we identified is the ligand gene S100A9 and TLR4 (pathway highlighted in purple, Fig. 7F) expressed on astrocytes. While TLR4 expression is well documented for microglia, its role in astrocytes, particularly in AD has been suggested [157]. The agonist S100A9 was reported to be expressed by activated microglia surrounding Aβ plaques implicated in phagocytosis, Aβ aggregation and neurotoxicity [157, 158]. To what extend and where S100A9 and TLR4 are expressed in our 5w Aβ hiNS(+) model system would be worth pursuing in the future.

In summary, we found more ligand-receptor matches between microglia/macrophages with receptors on astrocytes rather than neurons, suggesting a higher level of glia-glia rather than microglia/macrophage-neuron interaction. We identified several pathways of interests, both linked to neuroprotection or neurotoxicity in AD pathology indicating that our model partially mimics both aspects of AD allowing the investigations of these pathways and their relevance to Aβ pathology in future studies.

Limitations

The here presented cell culture protocol of human neurospheres requires the addition of microglia to form a co-culture system with functional neurons, astrocytes and microglia-like cells. The microglia must be differentiated in parallel to the neurospheres as described. We would like to highlight that the neurospheres are plated into coated 48-well plates, in which they attach. The microglia addition requires the neurospheres to be attached and we have observed that added microglia infiltrate readily but show a tendency to migrate back out of the neurospheres after around 2 weeks of co-culture. For this reason, we have restrained ourselves to 7–10 days of co-cultures to test the impact of microglia on neurospheres.

The plating of neurospheres also limits subsequent cell culture time as we have experienced neurospheres to slowly flatten out over time and we suggest to not maintain plated neurospheres for more than 3 or 4 weeks after plating.

For our chronic amyloidosis treatment paradigm using oligomeric Aβ1–42 we have used a scrambled Aβ 1–42 solution as a control group. While ideal control groups are challenging to design when using rapidly oligomerizing and aggregating Aβ1–42 we believe that scrambled Aβ1–42 is one of the most appropriate control to use. However, we highlight limitations of our approach and the experimental design of the here presented protocol: A: The scrambled Aβ1–42 treatment window was limited to 3-weeks while oligomeric Aβ1–42 treatment windows were 3 or 5 weeks. B: The scrambled Aβ1–42 concentration (0.36 µg/mL) was higher compared to the oligomeric Aβ 1–42 concentration (0.24 µg/mL).

We further wish to highlight that our chronic amyloidosis cell culture paradigm did not induce an evident Tau pathology in neurospheres and that the here presented model system and the reported AD-like pathology appears to be Aβ specific with the lack of Tau pathology.

We also want to highlight that neurosphere development between the 3 and 5 week treatment paradigm differs by two weeks between the two groups which could impact vulnerability towards Aβ exposure in the longer treatment paradigm.

Insufficient nuclei numbers in some cell types for certain treatment groups prevented us from thoroughly analyzing their transcriptomic data. Since we observed fewer hiMG in ctrl and 3w Aβ hiNS(+) compared to 5w Aβ hiNS(+), there were not enough microglia in ctrl and 3w Aβ hiNS(+) for comparative transcriptomic analysis. We focused on characterizing the transcriptomic profile of microglia only from the 5w Aβ hiNS(+) group as a result. Additionally, due to significant Aβ induced neuronal death in 5w Aβ hiNS(−), there were not enough nuclei to do any comparative analysis with this group.

Conclusions

This study introduces a human iPSC-derived neurosphere model, including neural progenitor cells, neurons, astrocytes and microglia residing in a healthy brain-like 3D environment. We have performed extensive functional characterizations of our neurospheres demonstrating synchronized neural activity, the absence of necrotic tissue in the core, and characterized functional properties of microglia inside the tissue, demonstrating surveillance behaviour and damage response properties.

Using this tool, we developed a novel chronic amyloidosis model by supplementing media with Aβ to gradually induce an AD-like pathology that effectively recapitulates key hallmarks of AD, providing a robust platform for studying the interplay of neurons, astrocytes, and microglia in chronic amyloidosis. Through this system, we demonstrated that microglia exhibit critical neuroprotective properties by mitigating Aβ induced oxidative stress, preserving neuronal activity, and reducing apoptosis. The timing of microglial infiltration during chronic amyloidosis determined the extend of their neuroprotective properties. The flexibility of this model to introduce microglia at a defined time during chronic amyloidosis driven neurodegeneration offers a unique platform to study their impact on AD pathology.

Importantly, we identified microglia as pivotal drivers of unique expression profiles in neurons and astrocytes, with several transcriptional changes resembling that found in human AD. This microglial-dependent modulation of astrocytic and neuronal responses underscores the intricate cell–cell signaling that governs disease progression. Although our model successfully mimics Aβ-driven neurotoxicity and transcriptional shifts, it does not recapitulate Tau hyperphosphorylation, highlighting its specificity for investigating amyloid centered mechanisms.

By bridging cellular, molecular, and transcriptional dynamics in a human context, this system provides a scalable and translationally relevant platform for preclinical research. Future studies leveraging this model could elucidate the pathogenic effects of identified gene expression profiles, including APOE isoforms, to explore the earliest, cell-specific pathophysiologic changes in the human brain in response to Aβ. Our cell culture system would also be a suitable tool to study monoclonal Aβ antibody mechanisms and the specific role of microglia in their ability to clear Aβ from 3D human neural tissue.

Methods

Generation of human iPSCs and induction of NPCs

The presented work on hiPSCs was approved by the University of British Columbia’s Clinical Research Ethics board. If not stated otherwise, a single cell line was used, obtained from a healthy female individual carrying the APOE alleles ε3/ε3. Peripheral blood was drawn and hiPSCs were generated from erythroid progenitor cells. The cell line has been used and characterized by our laboratory in the past [159, 160]. To validate our cell culture protocol, some experiments were performed using commercially available hiPSCs from the Jackson Laboratory (line KOLF2.1J).

hiPSCs were grown on Matrigel (BD Biosciences, Franklin Lakes, USA) coated 6-well plates in mTeSR™ Plus (STEMCELL Technologies, Vancouver, Canada, 100–0276). Media was supplemented with 10 µM Y- 27632 (EMD Millipore, Burlington, USA, 688000 - 5MG) for 24 h after plating. Cells were fed every 2–3 days and used for both the induction of neural progenitor cells (NPCs) and microglial differentiation. For storage, cells were frozen in media, supplemented with Y- 27632 and 10% DMSO (EMD Millipore) in liquid nitrogen (LN).

hiPSCs to NPC differentiation was achieved by a dual SMAD inhibition protocol as described before [161]. In short, hiPSCs were plated in 6-well plates in Basal Neural Maintenance media (BNMM) (Suppl. Table 1) supplemented with 10 µM SB- 431542 (Tocris, Bristol, United Kingdom, 1614) and 0.5 µM LDN- 193189 (Tocris, 6053) for 7 days, followed by a 3-week differentiation in Neural Stem Cell Media (see Suppl. Table 1) with media changes every 2–3 days. NPCs were then expanded and passaged twice before freezing them down in LN (BNMM supplemented with Y- 27632 and 10% DMSO) at P2.

Neurosphere generation

To generate human iPSC-derived neurospheres (hiNS), P2 NPCs were thawed and plated in Matrigel-coated 6-well plates using Neurosphere Expansion Media (NEX) (Suppl. Table 1) supplemented with 10 µM Y- 27632 for 24 h. Cells were fed every 2–3 days and allowed to recover for at least 5 days until the cell density was sufficient to start hiNS generation. To start hiNS formation, NPCs were lifted using Accutase (STEMCELL Technologies, 07920) and plated in single wells (treated with anti-adherence rinsing solution (STEMCELL Technologies, 07010)) of AggreWell^TM800 (STEMCELL Technologies, 34815) plates at a density of 1.2 * 10⁶ NPCs per well in NEX supplemented with 10 µM Y- 27632 for 24 h and centrifuged for 2 min at 100 g. hiNS were fed daily by careful partial media changes until day 7 (days in vitro, DIV 7). At DIV 7 hiNS were dislodged using wide bore tips and 37 µm reversible strainers (STEMCELL Technologies, 27250) and transferred to standard 9 cm cell culture petri dishes (VWR International, #10861 - 592, Edmonton, Canada) in Neurosphere Differentiation Media (NDM) (Suppl. Table 1) and immediately placed on an orbital shaker (Celltron, Infors HT, Bottmingen, Switzerland) in the cell culture incubator at 65 rpm. hiNS were fed every 2–3 days by full media change and switched to Neurosphere Maintenance Media (NMM) (Suppl. Table 1) on DIV 28. hiNS were kept in the petri dishes until DIV 39–41 when they were transferred to Matrigel coated (1% overnight) single wells of 48-well plates for future experiments either on regular cell culture plates, 8 mm coverslips or glass bottom live imaging plates (MatTek, Ashland, USA, P48G-1.5–6-F). For precise positioning on glass surfaces a tissue dissection microscope (Zeiss Stereo Discovery V8) was used inside a biosafety cabinet. Plated hiNS readily attach to surfaces allowing for easy confocal live imaging.

AAV transduction of hiNS

All AAVs used in this study were obtained from Neurophotonics Centre in Quebec, Canada. To induce the expression of genetically encoded fluorescent proteins, hiNS were transduced with 0.5–0.75 µL of custom AAVs (AAV2/PHP.eB) on the day of plating. Promoters used were gfaABC1D (for EGFP) and hSyn (for mCherry, roGFP1, GCaMP6f and iGluSnFr) and all AAVs titers ranged from 1–2 * 10¹³ gene copies per mL. After plating and transduction, hiNS media was partially changed every 2–3 days. Fluorescence as a result of AAV transduction can generally be observed after 7 days. Bright and easily quantifiable fluorescence was detected one and a half to two weeks after transduction.

Differentiation of human iPSC-derived microglia

Human iPSC-derived microglia (hiMG) were differentiated based on a protocol by the Cowley lab with minor adjustments [162] as follows. hiPSCs were collected from 6-well plates using Accutase and 1.5 * 10⁶ cells were added to a single well (treated with anti-adherence rinsing solution (STEMCELL Technologies, 07010)) of an AggreWell^TM800 plate in Embryoid Body Media (Suppl. Table 1) and centrifuged at 100g for 2 min. Media was changed daily for 4 days. On day 5, EBs were collected using a wide bore tip and a 37 µm reversible strainer and resuspended in 10 mL Hematopoietic Differentiation Media (HDM) (Suppl. Table 1) in a standard T75 Cell Culture flask. The flask was kept in the incubator for 7 days without feeding or moving for EBs to attach to the surface. Afterwards, EBs were fed with 5 mL of HDM twice a week. As early as day 35, accumulating macrophage precursors were harvested for subsequent microglia differentiation. Collected HDM enriched with macrophage precursors was centrifuged at 400g for 5 min and cells were resuspended in Microglial Differentiation Media (MDM) (Suppl. Table 1) and plated in 6-well glass bottom plates at densities of 0.5–1 * 10⁶ cells/well. Partial media changes were conducted every 2–3 days with MDM and hiMG were allowed to differentiate for 10–14 days.

hiMG transfer into neurospheres

Mature hiMG after 10–14 days of differentiation were collected using TrypLE Express (Thermo Scientific, Waltham, USA, 12,604,013) and resuspended in NMM supplemented with 100 ng/mL IL- 34 (BioLegend, San Diego, USA, 577906) and 25 ng/mL TGFβ1 (PeproTech, Cranbury, USA, 100 - 21) and added to well-attached hiNS in 48-well plates at a density of 35,000 cells/well between DIV 49–60. hiNS(+) were fed with partial media changes every 2–3 days with NMM supplemented with IL- 34/TGFβ1. Note that IL- 34/TGFβ1 supplementation was performed for all treatment groups, including hiNS(−) throughout the experiment.

ELISA

For multiplex ELISA assays of hiNS media, the supernatant of 6 individual hiNS (across all treatment groups) was pooled together and frozen at − 80 °C. Before analysis, samples were thawed on ice and concentrated using Amicon Ultra 3 kDa centrifugal filters (EMD Millipore, UFC500324) by centrifuging 500 µL at 14,000g for 15 min. This step was repeated twice until all media was concentrated. Concentrated media was then collected after flipping the filter and centrifuging again at 1000g for 2 min. The resulting total volume for each pooled sample of concentrated media was ~ 80 µL. For soluble Aβ and pTau181 quantification the V-PLEX Aβ Peptide Panel 1 (4G8) and S-PLEX Human Tau (pT181) kit (Meso Scale Diagnostics, Rockville, USA, K15199E- 2 and K151 AGMS- 1 respectively) were used respectively for electro chemiluminescent detection of Aβ38, Aβ40, Aβ42 and pTau181 in concentrated supernatant. Samples were diluted 1/5,000 in Diluent 35 (provided in the kit). All samples and calibrators were run in duplicate following manufacturers’ instructions. To quantify cytokine expression in hiNS media, the same samples were used for the Proinflammatory Panel 1 Human Kit (Meso Scale Diagnostics, K15049D- 1) for the quantification of ten cytokines: IFNγ, IL1β, IL- 2, IL- 4, IL- 6, IL- 8, IL- 10, IL- 12, IL- 13 and TNFα. All concentrated samples were diluted 15 × and all samples/calibrators were run in duplicate following the manufacturer's instructions. MESO QuickPlex SQ 120MM was used for plate reading and protein concentrations were determined using the Discovery WorkBench Software provided by Meso Scale. For cytokine quantification, concentrated supernatant protein concentration was detected via BCA for normalization of cytokines between samples. Note that the IL- 8 reading was saturated in all hiNS(+) treatment groups and excluded from analysis.

Oligomeric Aβ preparation

Oligomeric Aβ preparation was performed based following well-established and characterized protocol [36,37,38,39,40]: Synthetic lyophilised Aβ1–42 (ERI Amyloid Laboratory, New Haven, USA) was mixed thoroughly and dissolved in 40 μL of DMSO for 20 min at room temperature (RT) at a concentration of ~ 2 mM. The mixture was then separated into two 20 μL aliquots and 1 mL of F12 (Thermo Scientific, 11765054) media was added to each aliquot. The dissolved Aβ was incubated at RT for 24 h to induce oligomerization. Next, 500 μL of the oligomeric Aβ1–42 (Aβ) was centrifuged at 14,000g for 15 min using Amicon Ultra 3 kDa centrifugal filters. This was repeated 3 times to concentrate the Aβ. The Aβ was then washed twice with 500 μL of PBS and centrifuged at 14,000g for 15 min. The filters were then flipped onto collection tubes and centrifuged at 1000g for 2 min to collect the Aβ. The final Aβ was stored at − 80 °C. To determine the final Aβ concentration we used a multiplex Aβ (4G8) ELISA panel (see above) on four different batches of Aβ and averaged the results. The final stock concentration of Aβ was 0.24 ± 0.034 mg/mL and was used throughout the experiment in a 1:1000 dilution, unless stated otherwise.

To conjugate Aβ with pHrodo red/green (Thermo Scientific, P35372 and P35373), 2 μL were added into each 1 mL of the F12/Aβ mix prior to centrifugation and incubated at 4 °C in the dark for 3 h. Centrifugation steps were continued afterwards as described above with one additional PBS washing step and the conjugated pHrodo-Aβ was stored in the dark at − 80 °C.

Fibrillar Aβ preparation

For fibrillar Aβ− 1–42 (fAβ) generation (adapted from [163]) (see Fig. S4), the same lyophilised Aβ1–42 powder was used as described above. After dissolving the powder in 40 μL of DMSO for 20 min at room temperature (RT), and splitting the solution in two, 980 μL of 10 mM HCl was added each. Tubes were then stored at 37 °C overnight. Afterwards, the same filtration steps were taken to concentrate the fAβ.

Chronic Aβ treatment paradigm

To model chronic amyloidosis in our 3D neurosphere cultures, we supplemented NDM/NMM with Aβ in a 1:1000 dilution resulting in a final Aβ concentration in the media to be 0.24 μg/mL. Aβ was added during partial media changes (150 μL from a total of 400 μL per well) calculated for the total volume of the wells resulting in a gradual increase of aggregated Aβ over time. The duration of Aβ treatment was either 3 or 5 weeks. For the control groups (ctrl hiNS(−/+)), media was supplemented with 0.36 μg/mL scrambled Aβ1–42 for a total of 3 weeks.

Confocal microscopy of neurospheres

For live neurosphere imaging, hiNS were plated and transduced with AAVs in 48-well glass bottom plates as described above. hiNS were imaged 7–10 days post transfection using a confocal microscope equipped with a live imaging incubator system (LSM 880, Zeiss, Oberkochen, Germany). For hSyn-GCaMP6f, hSyn-iGluSnFr and gfaABC1D-GCaMP8 s 3–5 min time series were recorded of green emission excited using a 488 nm argon laser. hiNS display synchronized wave-like neuronal and calcium activity leading to rises in intracellular calcium levels (quantifying GCaMP6f or GCaMP8 s fluorescence) or a rise in extracellular glutamate concentrations (quantifying iGluSnFr) (Fig. 2). To simultaneously image hSyn-iGluSnFr and gfaABC1D-GCaMP8 s with neuronal calcium levels, we co-transduced to express hSyn-jRGECO1, a red fluorescent calcium indicator [164]. AQuA [165] was used for calcium wave quantification as follows. For each hiNS, calcium activity was recorded at one z-plane for 300 frames at a frame rate of 0.931 s per frame, and the resolution was 0.692 μm per pixel. Prior to analysis, the resolution was downsized to 1.384 μm/pixel in FIJI to allow for smoother processing. In AQuA, an ROI was drawn around each neurosphere. For event detection, AQuA uses the parameters intensity threshold, smoothing, and minimum pixel size, which were adjusted so that signals that spanned the entire neurosphere were counted as a single event, and other background single cell activity was classified as noise and ignored.

For pharmacological characterization of neuronal calcium waves we used the following drugs and concentrations: 1 μM Tetrodotoxin (TTX, Cayman Chemical Co., Michigan, USA, 14964 - 1), 100 μM D-AP5 (Tocris, 0106), 20 μM CNQX (Cayman Chemical Co., 14618 - 50), 300 μM GABA (Sigma Aldrich, A2129 - 100G) and 100 µM picrotoxin (PTX, Sigma Aldrich, P1675 - 1G).

To quantify neuronal redox states in hiNS we employed roGFP1 [27], a genetically encoded redox indicator we have used in the past [166]. roGFP1 is a ratiometric indicator in which upon oxidation, the excitation spectrum of roGFP1 shifts to lower wavelengths (~ 400 nm) while fully reduced roGFP1 has a maximum excitation peak at ~ 490 nm. The green fluorescence of roGFP1 on the other hand is not affected by its redox states. Therefore, roGFP1 was imaged on our live imaging confocal microscope alternately with either a 405 nm laser (maximum excitation for oxidized roGFP1) and a 488 nm laser (maximum excitation for reduced roGFP1) to quantify neuronal redox states across experimental conditions. Laser settings were kept constant and green emission for both excitation wavelengths were captured and averaged across a 30 μm z-stack. The ‘roGFP1 ratio’ (Fig. 2) was defined as green emission acquired exciting at 405 nm and divided by green emission acquired exciting at 488 nm. For one data set (Fig. 3D) hiNS expressing roGFP1 were fixed in 4% PFA substituted with 20 mM N-Ethylmaleimide (Sigma Aldrich, 04260 - 25G-F) to preserve the redox state of roGFP1 allowing post-fixation redox imaging [166].

To quantify hiMG infiltration into hiNS and simultaneously image neurons and astrocytes, we pre-stained differentiating hiMG in their respective 6-well glass bottom plate for 30 min with tomato-lectin- 649 (Thermo Scientific) in a 1:1000 dilution and subsequently added to hiNS co-transduced with AAV.php.eb.-hSyn-mCherry and AAV.php.eb.-hSyn-EGFP. hiNS and hiMG were then live imaged for 24 h (see Fig. 2H).

Human formalin-fixed paraffin-embedded (FFPE) magnified analysis of proteome (MAP) tissue clearing and expansion

The FFPE MAP procedure was adapted from the original MAP procedure [167] with slight modifications [53]. Three FFPE brain sections with 4 μm thickness were used, two familial AD patients and one sporadic AD patient.

Deparaffinization and rehydration

High precision #1.5 coverslips were cut and glued directly on the original slides to which the sections were attached on each side of the sections. These served as spacers for the gel embedding (see below). Then sections were deparaffinized and rehydrated by washing them with xylene three times for 5 min each and then two times for 5 min each in the following solutions: 100% ethanol 95% ethanol, 70% ethanol, and 50% ethanol. Finally, the samples were rinsed two times with deionized water for 5 min each. From this step onward, samples were protected from light.

Post-fixation

Sections were kept in fixative solution (4% PFA in PBS) for 2 h at 37 °C then washed three times in washing solution (PBS with 0.02% of NaN3) from 30 min each at 37 °C with gentle shaking.

AA-integration

Sections were incubated in low-AA solution (4% acrylamide (Sigma Aldrich, A3553), 4% PFA in PBS) at 4 °C with agitation overnight, followed by a 2 h incubation at 37 °C with gentle shaking.

Inactivation

Brain sections were washed with washing solution three times at 37 °C for 30 min each. Sections were then incubated in Inactivation solution (1% acetamide (Sigma Aldrich, A0500), 1% glycine (Sigma Aldrich, G7126), 0.02% NaN3 (Sigma Aldrich, S2002) with the pH titrated to 9.0 with NaOH) at 37 °C for 3 h with gentle shaking.

Monomer incubation

Brain sections were washed with washing solution three times at 37 °C for 30 min each. MAP solution was made by freshly adding V- 50 initiator (Sigma Aldrich; 440914) to a solution of 30% acrylamide, 0.1% bisacrylamide (Sigma Aldrich, M7279), and 10% sodium acrylate (Sigma Aldrich, 408220) in PBS; the brain sections were incubated in MAP solution overnight at 4 °C with gentle shaking.

Mounting and gel embedding

The gelation was done directly on the original slides to which the sections were attached. 100 μL of MAP solution with freshly added V- 50 initiator was placed directly on each section and topped with a coverslip to form gelation chambers. Gelation chambers were placed in a humid gelation box which was sealed, and the air purged with nitrogen gas. The gelation box was put in the oven at 45 °C for 2 h 30 min, following which the hybrid gel–tissue was removed and excess gel around the tissue cut away.

Denaturation, clearing, and expansion

The tissue–gel hybrid sections still attached to the slides were immersed in denaturation solution (in mM, 200 SDS, 200 NaCl, 50 Tris, pH 9) and incubated at 65 °C for 3 days with gentle shaking using an EasyClear device (LifeCanvas Technologies). The FFPE-MAP tissue–gel hybrids generally floated free of the slides after denaturation. Sections were washed 4 times for 1 h each with 0.001 × PBS and gentle shaking and then incubated in 0.001 × PBS at RT overnight with gentle shaking. To prepare the sections for immunostaining, the 0.001 × PBS was replaced with PBST (0.1% Triton X- 100 in PBS) for 1 h before staining. The side that had been facing the original slide, thus which had the tissue at the surface, was mounted next to the coverslip for imaging.

Immunofluorescence

For immunofluorescence staining, hiNS were plated on 8 mm glass coverslips and fixed in 4% paraformaldehyde for 1 h. Fixed hiNS were washed and kept in PBS at 4 °C before staining. Immunofluorescence stainings were conducted as described before [166]. hiNS were either stained in full or sliced with a vibratome into 80 µm thin sections (VT1200, Leica, Nussloch, Germany) before immunostainings. In brief, hiNS (full or slices) were blocked with 10% normal donkey serum (NDS) (Jackson ImmunoResearch, West Grove, USA, 017 - 000- 121) in 0.01 M PBS containing 2% Triton X- 100 and 20% DMSO for 1 h. Afterwards, samples were washed once with PBS before incubating in primary antibody solution containing 0.1 M PBS, 2.5% NDS, 2%Triton X- 100 and 20% DMSO. Primary antibodies used were Aβ (6E10), 1:500 (BioLegend, San Diego, USA, 803001); Aβ (4G8), 1:500 (BioLegend, 800708); Aβ (mOC87), 1:500 (abcam, Cambridge, United Kingdom, ab201062); APOE (D7I9 N), 1:1000 (New England Biolabs, Ipswich, USA, 13366S); GFAP, 1:1500 (Thermo Scientific, 13–0300); MAP2, 1:1500 (abcam ab5392); IBA1, 1:500 (FUJIFILM Wako Chemicals, Richmond, USA, 019–19741); Cleaved Caspase- 3, 1:500 (Cell Signaling Technologies, Danvers, USA, 9661S); CD68 (KP1), 1:200 (Bio-rad, Hercules, USA, MCA5709); Tau (HT7), 1:500 (Thermo Scientific, MN10000); pTau181 (D9 F4G), 1:500 (Cell Signaling Technologies, 12885); NeuN, 1:500 (abcam, ab279297); CD45, 1:500 (abcam, ab30470); SLC7 A11, 1:500 (Abova, Taipei City, Taiwan, PAB7138). Afterwards, samples were thoroughly washed with PBS 3–4 times before incubating in secondary antibody solution containing 0.1 M PBS, 2.5% NDS, 2%Triton X- 100 and 20% DMSO and secondary antibodies targeted against the host species of primary antibodies conjugated with either Alexa Fluor 405, 488, 568 or 647 (Thermo Scientific) and incubated for 1 h. In some stainings, DAPI (Thermo Scientific, D1306) was added in a 1:5000 dilution for 15 min, after the secondary antibody staining. For quantification of apoptosis, TUNEL (Click-iT™ Plus, Invitrogen, C10617) was used following the manufacturer guidelines after immunostainings were performed. After 3–4 washing steps in PBS, stained hiNS were imaged on the LSM 980 confocal microscope (Zeiss). Quantification of fluorescence was performed using FIJI (ImageJ 1.53 t). The immunostaining protocol for FFPE MAP human sections was performed the same way except for using a 0.1% Triton-X- 100 PBS solution for blocking and staining buffers instead of the 2% Triton-X- 100/20% DMSO mix.

Western Blot

Western blot analysis was performed on differentiating hiMG harvested at different time points. Cell lysates were prepared using RIPA lysis buffer supplemented with phosphatase and protease inhibitor cocktail (Roche, Basel, Switzerland, 11697498001), followed by centrifugation at 13,000g for 10 min at 4 °C. The resulting supernatant was diluted with 4 × Laemmli sample buffer and boiled for 5 min before loading onto Mini-Protean TGX gels (Bio-Rad, Hercules, USA, 4561096). Subsequently, SDS-PAGE was conducted, and the separated proteins were transferred onto PVDF membranes using a Trans-Blot Turbo semi-dry transfer system (Bio-Rad, 1704156). Membranes were blocked using 3% BSA in TBS with 0.1% Tween20 (TBST), followed by overnight incubation at 4 °C in TBST with primary antibodies against IBA1 (1:500, GeneTex, Irvine, USA, GTX635400), IL1β, 1:1000 (abcam, ab283818), TNFα, 1:300 (Cell Signaling, 6945S) (APOE (D7I9 N), 1:1000 (New England Biolabs, 13366S), β-actin, 1:1000 (Thermo Scientific, MA5 - 15739) or GAPDH, 1:1000 (Thermo Scientific, PA1988). After washing, membranes were probed with HRP-conjugated secondary antibodies (Jackson ImmunoResearch) for 2 h at room temperature. Protein bands were visualized using Bio-Rad Clarity Western ECL substrate and imaged with a ChemiDoc MP imaging system. Protein band size and intensity were quantified using ImageJ software.

For soluble APOE quantifications we normalized the gel loading for the individual groups to the protein content and raw intensities were compared as arbitrary units as housekeeping proteins were not detectable in the hiNS supernatant solutions.

Electrophysiology and 2-photon imaging

Patch-clamp recordings on microglial cells were performed on hiNS(+) either ctrl or Aβ (3w/5w) treated after plating them on coverslips, which allowed moving them to a 2-photon microscope for patch-clamp and image recordings (Zeiss 510Meta microscope with a Ti:Sapphire Chameleon laser). hiMG were stained with 2 µL tomato-lectin- 488 (for patch-clamp) or 594 (for process motility quantification) (Thermo Scientific, L32470 and L32471, respectively) per well for 30 min to allow for the identification of microglia in neurospheres using the 2-photon microscope. hiNS were supplied with a continuous perfusion of carbogen (95% O₂/5% CO₂) infused BrainPhys Imaging Optimized media (STEMCELL Technologies, 05796) and the following intracellular solution was used for microglial patch-clamp recordings, as described before (in mM) [33, 34]: KCl,130; MgCl₂, 2;CaCl₂, 0.5; Na-ATP, 2; EGTA, 5; HEPES, 10 and sulforhodamine 101, 0.01 (Sigma Aldrich, S7635 - 100MG) at an osmolarity of ~ 370 mOsm and a pH of 7.3. Borosilicate patch pipettes were pulled with a final pipette resistance of 3–5 MΩ. hiMG were targeted using the lectin fluorescence and clamped at − 40 mV after reaching whole-cell configuration. A series of de- and hyperpolarizing voltage steps were applied ranging from − 150 to + 60 mV with 10 mV increments and patched hiMG with abnormally small capacitances (< 10 pF) or a series resistance of > 65 MΩ were discarded. To test if hiMG perform surveillance functions and respond to tissue damage, ctrl hiNS(+) (N = 3) were stained with lectin- 594 and z-stack (27 µm thick) time series were acquired using our 2-photon microscope at 850 nm. After recording a baseline to monitor process movement, a laser lesion was burned into the centre of the neurosphere by tuning the laser to 800 nm at 100% power for 15 s. hiMG process movement towards the lesion was then imaged again using 850 nm excitation wavelengths in a z-stack time series recording.

Single nuclei RNA sequencing

For tissue collection, hiNS were pooled as follows: 4 each for ctrl hiNS(−), ctrl hiNS(+), 3w Aβ hiNS(−), and 3w Aβ hiNS(+); 6 each for 5w Aβ hiNS(−) and 5w Aβ hiNS(+). For Nuclei dissociation and preparation, we used the 10X Nuclei Isolation Kit from 10 × Genomics. The protocol defined here: CG000505_Chromium_Nuclei_Isolation_Kit, had a few modifications to ensure enough nuclei for processing:

Following Chapter 1: Single Cell Gene Expression & Chromium Fixed RNA Profiling f–h) Added 500 µL of lysis buffer into tube containing hiNS and started a 5 min (rather than 10 min) timer. Once the hiNS were in solution, transferred the 500 µL into a 2 mL glass douncer (Dounce Tissue Grinder Set). Used 10 strokes with pestle A and 10 strokes with pestle B to break apart the hiNS (Note: can add less or more strokes depending on how the tissue breaks apart). m) After lysis, the nuclei were viewed under the microscope using AOPI and determined that debris removal was not necessary. Step m was removed and proceeded directly to wash steps (o). Note: Debris removal causes loss of at least 50% of the nuclei and was skipped because the hiNS sample was clean and we wanted to prevent the extreme nuclei loss. o) Washed the nuclei in 500 µL wash buffer q) Did not complete a second wash since the sample was clean and we wanted to prevent nuclei loss.

Single nucleus capture was performed at the Princess Margaret Genomics Centre (Toronto, ON) using the relevant Genomics Chromium system protocol. Libraries were sequenced on the Illumina NovaseqX Plus instrument. FASTQ reads were aligned to the wild-type human genome (hg38) using the relevant 10 × Genomics Cell Ranger software. Output data was analyzed using a previously described computational pipeline [20,21,22,23,24]. In brief, cells with low UMI counts,, cells with high mitochondrial DNA content, and putative doublets were removed. Genes detected in fewer than three cells were removed and cell transcriptomes were normalized using the scran R package to account for read depth and library size [21]. The Seurat package (v4.2.0) [20] was then used to process the normalized expression matrices. Following this, principal component analysis (PCA) was performed using the top 2000 highly variable genes. Next, the top principal components were used to project the dataset into two-dimensional space as Uniform Manifold Approximation and Projection (UMAP) embeddings using the RunUMAP function in Seurat. Afterwards, the top principal components were also used to iteratively carry out SNN-Cliq-inspired clustering using the FindClusters function in Seurat at resolutions ranging from 0.2 to 2.4. Each dataset was analyzed by choosing the most conservative resolution; resolution was only increased to interrogate cell heterogeneity or if two cell populations of known identity, based on the expression of canonical markers, were not separating into distinct clusters as expected. To merge or subset datasets, barcodes corresponding to the cells or cell types of interest were used to extract relevant gene expression information from filtered gene expression matrices. For the merging of datasets, the gene expression information from each respective dataset was then combined. The cells were then once again run through the pipeline as described above.

For the hiNS(−/+) merged dataset (as in Fig. 1C), clusters were assigned at a resolution of 0.4 (16 clusters identified with at least 107 DEGs between most similar clusters, FDR < 0.01). To generate the subsetted astrocyte dataset (as in Fig. 1F), cells from cluster 0 (res. 0.4) were extracted and run through the pipeline. Clusters were assigned at a resolution of 0.2 (4 clusters identified with at least 112 DEGs between most similar clusters, FDR < 0.01). For the hiNS(−/+) in control, 3w and 5w Aβ merged dataset (as in Fig. 6A), clusters were assigned at a resolution of 0.4 (16 clusters identified with at least 204 DEGs between most similar clusters, FDR < 0.01). To generate the subsetted astrocyte dataset (as in Fig. 6D), cells from cluster 0 (res 0.4) were extracted and run through the pipeline. Clusters were assigned at a resolution of 0.2 (4 clusters identified with at least 112 DEGs between most similar clusters, FDR < 0.01). For subsequent subsetted data sets we used resolutions of 0.4 for microglia/macrophages (as in Fig. 7A), for neurons 0.2 (as in Fig. S10 A) and 0.1 for the hiNS(+) only merge of all six experimental groups (as in Fig. 8E).

DEG statistical analysis

DEG analysis of snRNA-seq data was performed using the FindMarkers function in Seurat. Expression levels were tested using a Wilcoxon Rank Sum Test. A Bonferroni-adjusted p-value smaller than 0.05 (padj < 0.05), coupled with an average fold-change of greater than 1.3 (FC > 1.3), was considered statistically significant.

Gene ontology analysis

g:Profiler (version e109_eg56_p17_1 d3191 d, https://biit.cs.ut.ee/gprofiler/) [168] was used to determine which gene ontology (GO) terms were significantly differentially overrepresented in cell populations following gene expression (fold-change > 1.3, adjusted p-value < 0.05) comparisons. If multiple Ensembl IDs were detected for a particular gene, the ID with the most GO annotations was used. In cases with multiple IDs with the same number of GO annotations, the first ID was used. Some genes were excluded from analysis if Ensembl IDs could not be found. The g:SCS multiple testing correction algorithm was used for p-value correction with a significance threshold of 0.05.

Ligand-receptor interaction analysis

For ligand-receptor interaction analysis between differentially expressed ligands with differentially expressed receptors on astrocytes and neurons we used our 5w Aβ hiNS(+) snRNA-seq group and subsetted from the merged Seurat including all treatment groups, and clustered with 0.4 resolution. The clusters that represented microglia/macrophages (13), astrocytes (0), and excitatory and inhibitory neurons (5, 4, 2, 7, 12, 10, 14, 3, 9, 8) were isolated for ligand-receptor analysis. Genes for microglia ligands were matched with their astrocyte and neuron receptor genes using the LR network database from Zenodo (https://zenodo.org/record/7074291/files/lr_network_human_21122021.rds). For each ligand or receptor, to determine if the gene was specific to its cluster, differential gene analysis was done with the Seurat function FindAllMarkers. Any genes with a Bonferroni-adjusted p-value smaller than 0.05 (padj < 0.05) and an average logfold-change of greater than 1.3 (log2 FC > 1.3) were considered cluster specific. Only ligand-receptor pairs where both genes were cluster specific and had a mean percent expression greater than 10% were plotted with cc_circos in the CCPlotR package.

Pearson and single-cell correlation analyses

Pearson correlation analysis (Fig. 1H) was performed by averaging the expression of a given transcript within each cluster/cell type of interest and correlating the expression values between clusters/cell types using the cor.test function in R.

Statistical analysis

Single individual hiNS were defined as N of one for the purpose of statistical analysis throughout the manuscript. For the comparison of calcium wave frequencies (quantified by GCaMP6f imaging), paired student-t-tests were used for pharmacological experiments and comparing two different time points. Unpaired student t-tests were used when comparing baseline calcium wave frequencies, Aβ phagocytosis, immunoreactivity and nuclei density of control hiNS vs. Aβ treated hiNS. For comparing microglial densities and GFAP immunoreactivity between control, 3w Aβ and 5w Aβ hiNS(+) ordinary one-way ANOVA followed by Holm-Sidak post-hoc tests were used. Logarithmic transformation was performed on all data sets comparing control, 3w Aβ and 5w Aβ hiNS with or without hiMG prior to using two-way ANOVAs, testing for the effect of hiMG, followed by Holm-Sidak post-hoc tests. Statistical significance was indicated as follows: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Data availability

Our snRNA-seq has been uploaded to Gene Expression Omnibus (GEO) and can be found under the accession number GSE272186. Other data are available from the corresponding authors upon reasonable request.

References

Bloom GS. Amyloid-beta and tau: the trigger and bullet in Alzheimer disease pathogenesis. JAMA Neurol. 2014;71:505–8.
Article PubMed Google Scholar
Sims JR, Zimmer JA, Evans CD, Lu M, Ardayfio P, Sparks J, Wessels AM, Shcherbinin S, Wang H, Monkul Nery ES, et al. Donanemab in early symptomatic alzheimer disease: the TRAILBLAZER-ALZ 2 randomized clinical trial. JAMA. 2023;330:512–27.
Article CAS PubMed PubMed Central Google Scholar
van Dyck CH, Swanson CJ, Aisen P, Bateman RJ, Chen C, Gee M, Kanekiyo M, Li D, Reyderman L, Cohen S, et al. Lecanemab in early Alzheimer’s disease. N Engl J Med. 2023;388:9–21.
Article PubMed Google Scholar
Guerreiro R, Wojtas A, Bras J, Carrasquillo M, Rogaeva E, Majounie E, Cruchaga C, Sassi C, Kauwe JS, Younkin S, et al. TREM2 variants in Alzheimer’s disease. N Engl J Med. 2013;368:117–27.
Article CAS PubMed Google Scholar
Zhang Y, Chen H, Li R, Sterling K, Song W. Amyloid beta-based therapy for Alzheimer’s disease: challenges, successes and future. Signal Transduct Target Ther. 2023;8:248.
Article CAS PubMed PubMed Central Google Scholar
Leisher S, Bohorquez A, Gay M, Garcia V, Jones R, Baldaranov D, Rafii MS. Amyloid-lowering monoclonal antibodies for the treatment of early Alzheimer’s disease. CNS Drugs. 2023;37:671–7.
Article PubMed PubMed Central Google Scholar
Piccioni G, Mango D, Saidi A, Corbo M, Nistico R. Targeting microglia-synapse interactions in Alzheimer's disease. Int J Mol Sci. 2021; 22.
Grubman A, Chew G, Ouyang JF, Sun G, Choo XY, McLean C, Simmons RK, Buckberry S, Vargas-Landin DB, Poppe D, et al. A single-cell atlas of entorhinal cortex from individuals with Alzheimer’s disease reveals cell-type-specific gene expression regulation. Nat Neurosci. 2019;22:2087–97.
Article CAS PubMed Google Scholar
Sadick JS, O’Dea MR, Hasel P, Dykstra T, Faustin A, Liddelow SA. Astrocytes and oligodendrocytes undergo subtype-specific transcriptional changes in Alzheimer’s disease. Neuron. 2022;110(1788–1805): e1710.
Google Scholar
Gao C, Shen X, Tan Y, Chen S. Pathogenesis, therapeutic strategies and biomarker development based on “omics” analysis related to microglia in Alzheimer’s disease. J Neuroinflammation. 2022;19:215.
Article PubMed PubMed Central Google Scholar
Drummond E, Wisniewski T. Alzheimer’s disease: experimental models and reality. Acta Neuropathol. 2017;133:155–75.
Article CAS PubMed Google Scholar
Barak M, Fedorova V, Pospisilova V, Raska J, Vochyanova S, Sedmik J, Hribkova H, Klimova H, Vanova T, Bohaciakova D. Human iPSC-derived neural models for studying Alzheimer’s disease: from neural stem cells to cerebral organoids. Stem Cell Rev Rep. 2022;18:792–820.
Article PubMed PubMed Central Google Scholar
Papaspyropoulos A, Tsolaki M, Foroglou N, Pantazaki AA. Modeling and targeting Alzheimer’s disease with organoids. Front Pharmacol. 2020;11:396.
Article CAS PubMed PubMed Central Google Scholar
Sreenivasamurthy S, Laul M, Zhao N, Kim T, Zhu D. Current progress of cerebral organoids for modeling Alzheimer’s disease origins and mechanisms. Bioeng Transl Med. 2023;8: e10378.
Article CAS PubMed Google Scholar
Choi SH, Kim YH, Quinti L, Tanzi RE, Kim DY. 3D culture models of Alzheimer’s disease: a road map to a “cure-in-a-dish.” Mol Neurodegener. 2016;11:75.
Article PubMed PubMed Central Google Scholar
Hernandez-Sapiens MA, Reza-Zaldivar EE, Cevallos RR, Marquez-Aguirre AL, Gazarian K, Canales-Aguirre AA. A three-dimensional alzheimer’s disease cell culture model using iPSC-derived neurons carrying A246E mutation in PSEN1. Front Cell Neurosci. 2020;14:151.
Article CAS PubMed PubMed Central Google Scholar
Jorfi M, Park J, Hall CK, Lin CJ, Chen M, von Maydell D, Kruskop JM, Kang B, Choi Y, Prokopenko D, et al. Infiltrating CD8(+) T cells exacerbate Alzheimer’s disease pathology in a 3D human neuroimmune axis model. Nat Neurosci. 2023;26:1489–504.
Article CAS PubMed PubMed Central Google Scholar
Kim YH, Choi SH, D’Avanzo C, Hebisch M, Sliwinski C, Bylykbashi E, Washicosky KJ, Klee JB, Brustle O, Tanzi RE, Kim DY. A 3D human neural cell culture system for modeling Alzheimer’s disease. Nat Protoc. 2015;10:985–1006.
Article CAS PubMed PubMed Central Google Scholar
Kwak SS, Washicosky KJ, Brand E, von Maydell D, Aronson J, Kim S, Capen DE, Cetinbas M, Sadreyev R, Ning S, et al. Amyloid-beta42/40 ratio drives tau pathology in 3D human neural cell culture models of Alzheimer’s disease. Nat Commun. 2020;11:1377.
Article CAS PubMed PubMed Central Google Scholar
Hao Y, Hao S, Andersen-Nissen E, Mauck WM 3rd, Zheng S, Butler A, Lee MJ, Wilk AJ, Darby C, Zager M, et al. Integrated analysis of multimodal single-cell data. Cell. 2021;184(3573–3587): e3529.
Google Scholar
Lun AT, McCarthy DJ, Marioni JC. A step-by-step workflow for low-level analysis of single-cell RNA-seq data with bioconductor. F1000Res. 2016;5:2122.
PubMed PubMed Central Google Scholar
Yuzwa SA, Borrett MJ, Innes BT, Voronova A, Ketela T, Kaplan DR, Bader GD, Miller FD. Developmental emergence of adult neural stem cells as revealed by single-cell transcriptional profiling. Cell Rep. 2017;21:3970–86.
Article CAS PubMed Google Scholar
Borrett MJ, Innes BT, Tahmasian N, Bader GD, Kaplan DR, Miller FD. A shared transcriptional identity for forebrain and dentate gyrus neural stem cells from embryogenesis to adulthood. eNeuro 2022; 9.
Borrett MJ, Innes BT, Jeong D, Tahmasian N, Storer MA, Bader GD, Kaplan DR, Miller FD. Single-cell profiling shows murine forebrain neural stem cells reacquire a developmental state when activated for adult neurogenesis. Cell Rep. 2020;32: 108022.
Article CAS PubMed Google Scholar
Willis A, Jeong D, Liu Y, Lithopoulos MA, Yuzwa SA, Frankland PW, Kaplan DR, Miller FD. Single cell approaches define neural stem cell niches and identify microglial ligands that can enhance precursor-mediated oligodendrogenesis. Cell Rep. 2025;44: 115194.
Article CAS PubMed Google Scholar
Dennis DJ, Wang BS, Karamboulas K, Kaplan DR, Miller FD. Single-cell approaches define two groups of mammalian oligodendrocyte precursor cells and their evolution over developmental time. Stem Cell Reports. 2024;19:654–72.
Article CAS PubMed PubMed Central Google Scholar
Hanson GT, Aggeler R, Oglesbee D, Cannon M, Capaldi RA, Tsien RY, Remington SJ. Investigating mitochondrial redox potential with redox-sensitive green fluorescent protein indicators. J Biol Chem. 2004;279:13044–53.
Article CAS PubMed Google Scholar
Poli D, Magliaro C, Ahluwalia A. Experimental and computational methods for the study of cerebral organoids: a review. Front Neurosci. 2019;13:162.
Article PubMed PubMed Central Google Scholar
Arjun McKinney A, Petrova R, Panagiotakos G. Calcium and activity-dependent signaling in the developing cerebral cortex. Development. 2022; 149.
Namiki S, Norimoto H, Kobayashi C, Nakatani K, Matsuki N, Ikegaya Y. Layer III neurons control synchronized waves in the immature cerebral cortex. J Neurosci. 2013;33:987–1001.
Article CAS PubMed PubMed Central Google Scholar
Uwechue NM, Marx MC, Chevy Q, Billups B. Activation of glutamate transport evokes rapid glutamine release from perisynaptic astrocytes. J Physiol. 2012;590:2317–31.
Article CAS PubMed PubMed Central Google Scholar
Fellin T, Pascual O, Gobbo S, Pozzan T, Haydon PG, Carmignoto G. Neuronal synchrony mediated by astrocytic glutamate through activation of extrasynaptic NMDA receptors. Neuron. 2004;43:729–43.
Article CAS PubMed Google Scholar
Boucsein C, Zacharias R, Farber K, Pavlovic S, Hanisch UK, Kettenmann H. Purinergic receptors on microglial cells: functional expression in acute brain slices and modulation of microglial activation in vitro. Eur J Neurosci. 2003;17:2267–76.
Article PubMed Google Scholar
Wendt S, Maricos M, Vana N, Meyer N, Guneykaya D, Semtner M, Kettenmann H. Changes in phagocytosis and potassium channel activity in microglia of 5xFAD mice indicate alterations in purinergic signaling in a mouse model of Alzheimer’s disease. Neurobiol Aging. 2017;58:41–53.
Article CAS PubMed Google Scholar
Wendt S, Wogram E, Korvers L, Kettenmann H. Experimental cortical spreading depression induces NMDA receptor dependent potassium currents in microglia. J Neurosci. 2016;36:6165–74.
Article CAS PubMed PubMed Central Google Scholar
Haas LT, Kostylev MA, Strittmatter SM. Therapeutic molecules and endogenous ligands regulate the interaction between brain cellular prion protein (PrPC) and metabotropic glutamate receptor 5 (mGluR5). J Biol Chem. 2014;289:28460–77.
Article CAS PubMed PubMed Central Google Scholar
Um JW, Kaufman AC, Kostylev M, Heiss JK, Stagi M, Takahashi H, Kerrisk ME, Vortmeyer A, Wisniewski T, Koleske AJ, et al. Metabotropic glutamate receptor 5 is a coreceptor for Alzheimer abeta oligomer bound to cellular prion protein. Neuron. 2013;79:887–902.
Article CAS PubMed PubMed Central Google Scholar
Um JW, Nygaard HB, Heiss JK, Kostylev MA, Stagi M, Vortmeyer A, Wisniewski T, Gunther EC, Strittmatter SM. Alzheimer amyloid-beta oligomer bound to postsynaptic prion protein activates Fyn to impair neurons. Nat Neurosci. 2012;15:1227–35.
Article CAS PubMed PubMed Central Google Scholar
Lauren J, Gimbel DA, Nygaard HB, Gilbert JW, Strittmatter SM. Cellular prion protein mediates impairment of synaptic plasticity by amyloid-beta oligomers. Nature. 2009;457:1128–32.
Article CAS PubMed PubMed Central Google Scholar
Kaufman AC, Salazar SV, Haas LT, Yang J, Kostylev MA, Jeng AT, Robinson SA, Gunther EC, van Dyck CH, Nygaard HB, Strittmatter SM. Fyn inhibition rescues established memory and synapse loss in Alzheimer mice. Ann Neurol. 2015;77:953–71.
Article CAS PubMed PubMed Central Google Scholar
Zhang G, Wang Z, Hu H, Zhao M, Sun L. Microglia in Alzheimer’s disease: a target for therapeutic intervention. Front Cell Neurosci. 2021;15: 749587.
Article CAS PubMed PubMed Central Google Scholar
Gao C, Jiang J, Tan Y, Chen S. Microglia in neurodegenerative diseases: mechanism and potential therapeutic targets. Signal Transduct Target Ther. 2023;8:359.
Article PubMed PubMed Central Google Scholar
Hatami A, Albay R 3rd, Monjazeb S, Milton S, Glabe C. Monoclonal antibodies against Abeta42 fibrils distinguish multiple aggregation state polymorphisms in vitro and in Alzheimer disease brain. J Biol Chem. 2014;289:32131–43.
Article CAS PubMed PubMed Central Google Scholar
Jurga AM, Paleczna M, Kuter KZ. Overview of general and discriminating markers of differential microglia phenotypes. Front Cell Neurosci. 2020;14:198.
Article PubMed PubMed Central Google Scholar
Jung H, Lee SY, Lim S, Choi HR, Choi Y, Kim M, Kim S, Lee Y, Han KH, Chung WS, Kim CH. Anti-inflammatory clearance of amyloid-beta by a chimeric Gas6 fusion protein. Nat Med. 2022;28:1802–12.
Article CAS PubMed Google Scholar
Busche MA, Hyman BT. Synergy between amyloid-beta and tau in Alzheimer’s disease. Nat Neurosci. 2020;23:1183–93.
Article CAS PubMed PubMed Central Google Scholar
Mathys H, Boix CA, Akay LA, Xia Z, Davila-Velderrain J, Ng AP, Jiang X, Abdelhady G, Galani K, Mantero J, et al. Single-cell multiregion dissection of Alzheimer’s disease. Nature. 2024;632:858–68.
Article CAS PubMed PubMed Central Google Scholar
DeChellis-Marks MR, Wei Y, Ding Y, Wolfe CM, Krivinko JM, MacDonald ML, Lopez OL, Sweet RA, Kofler J. Psychosis in Alzheimer’s disease is associated with increased excitatory neuron vulnerability and post-transcriptional mechanisms altering synaptic protein levels. Front Neurol. 2022;13: 778419.
Article PubMed PubMed Central Google Scholar
Gao L, Zhang Y, Sterling K, Song W. Brain-derived neurotrophic factor in Alzheimer’s disease and its pharmaceutical potential. Transl Neurodegener. 2022;11:4.
Article CAS PubMed PubMed Central Google Scholar
Keren-Shaul H, Spinrad A, Weiner A, Matcovitch-Natan O, Dvir-Szternfeld R, Ulland TK, David E, Baruch K, Lara-Astaiso D, Toth B, et al. A unique microglia type associated with restricting development of Alzheimer’s disease. Cell. 2017;169(1276–1290): e1217.
Google Scholar
Raulin AC, Doss SV, Trottier ZA, Ikezu TC, Bu G, Liu CC. ApoE in Alzheimer’s disease: pathophysiology and therapeutic strategies. Mol Neurodegener. 2022;17:72.
Article CAS PubMed PubMed Central Google Scholar
Strittmatter WJ, Saunders AM, Schmechel D, Pericak-Vance M, Enghild J, Salvesen GS, Roses AD. Apolipoprotein E: high-avidity binding to beta-amyloid and increased frequency of type 4 allele in late-onset familial Alzheimer disease. Proc Natl Acad Sci USA. 1993;90:1977–81.
Article CAS PubMed PubMed Central Google Scholar
Delhaye M, LeDue J, Robinson K, Xu Q, Zhang Q, Oku S, Zhang P, Craig AM. Adaptation of magnified analysis of the proteome for excitatory synaptic proteins in varied samples and evaluation of cell type-specific distributions. J Neurosci. 2024, 44.
Barres BA, Lazar MA, Raff MC. A novel role for thyroid hormone, glucocorticoids and retinoic acid in timing oligodendrocyte development. Development. 1994;120:1097–108.
Article CAS PubMed Google Scholar
Madhavan M, Nevin ZS, Shick HE, Garrison E, Clarkson-Paredes C, Karl M, Clayton BLL, Factor DC, Allan KC, Barbar L, et al. Induction of myelinating oligodendrocytes in human cortical spheroids. Nat Methods. 2018;15:700–6.
Article CAS PubMed PubMed Central Google Scholar
Kuga N, Sasaki T, Takahara Y, Matsuki N, Ikegaya Y. Large-scale calcium waves traveling through astrocytic networks in vivo. J Neurosci. 2011;31:2607–14.
Article CAS PubMed PubMed Central Google Scholar
Yuryev M, Pellegrino C, Jokinen V, Andriichuk L, Khirug S, Khiroug L, Rivera C. In vivo calcium imaging of evoked calcium waves in the embryonic cortex. Front Cell Neurosci. 2015;9:500.
PubMed Google Scholar
Xiang Y, Tanaka Y, Patterson B, Kang YJ, Govindaiah G, Roselaar N, Cakir B, Kim KY, Lombroso AP, Hwang SM, et al. Fusion of regionally specified hPSC-derived organoids models human brain development and interneuron migration. Cell Stem Cell. 2017;21(383–398): e387.
Google Scholar
Woodruff G, Phillips N, Carromeu C, Guicherit O, White A, Johnson M, Zanella F, Anson B, Lovenberg T, Bonaventure P, Harrington AW. Screening for modulators of neural network activity in 3D human iPSC-derived cortical spheroids. PLoS ONE. 2020;15: e0240991.
Article CAS PubMed PubMed Central Google Scholar
Michell-Robinson MA, Touil H, Healy LM, Owen DR, Durafourt BA, Bar-Or A, Antel JP, Moore CS. Roles of microglia in brain development, tissue maintenance and repair. Brain. 2015;138:1138–59.
Article PubMed PubMed Central Google Scholar
Park J, Wetzel I, Marriott I, Dreau D, D’Avanzo C, Kim DY, Tanzi RE, Cho H. A 3D human triculture system modeling neurodegeneration and neuroinflammation in Alzheimer’s disease. Nat Neurosci. 2018;21:941–51.
Article CAS PubMed PubMed Central Google Scholar
Rifat A, Ossola B, Burli RW, Dawson LA, Brice NL, Rowland A, Lizio M, Xu X, Page K, Fidzinski P, et al. Differential contribution of THIK-1 K(+) channels and P2X7 receptors to ATP-mediated neuroinflammation by human microglia. J Neuroinflammation. 2024;21:58.
Article CAS PubMed PubMed Central Google Scholar
Andreasen N, Hesse C, Davidsson P, Minthon L, Wallin A, Winblad B, Vanderstichele H, Vanmechelen E, Blennow K. Cerebrospinal fluid beta-amyloid(1–42) in Alzheimer disease: differences between early- and late-onset Alzheimer disease and stability during the course of disease. Arch Neurol. 1999;56:673–80.
Article CAS PubMed Google Scholar
Golde TE. Alzheimer’s disease—the journey of a healthy brain into organ failure. Mol Neurodegener. 2022;17:18.
Article CAS PubMed PubMed Central Google Scholar
Misrani A, Tabassum S, Yang L. Mitochondrial dysfunction and oxidative stress in Alzheimer’s disease. Front Aging Neurosci. 2021;13: 617588.
Article CAS PubMed PubMed Central Google Scholar
Huang WJ, Zhang X, Chen WW. Role of oxidative stress in Alzheimer’s disease. Biomed Rep. 2016;4:519–22.
Article CAS PubMed PubMed Central Google Scholar
Fracassi A, Marcatti M, Zolochevska O, Tabor N, Woltjer R, Moreno S, Taglialatela G. Oxidative damage and antioxidant response in frontal cortex of demented and nondemented individuals with Alzheimer’s neuropathology. J Neurosci. 2021;41:538–54.
Article CAS PubMed PubMed Central Google Scholar
Palop JJ, Mucke L. Amyloid-beta-induced neuronal dysfunction in Alzheimer’s disease: from synapses toward neural networks. Nat Neurosci. 2010;13:812–8.
Article CAS PubMed PubMed Central Google Scholar
Louneva N, Cohen JW, Han LY, Talbot K, Wilson RS, Bennett DA, Trojanowski JQ, Arnold SE. Caspase-3 is enriched in postsynaptic densities and increased in Alzheimer’s disease. Am J Pathol. 2008;173:1488–95.
Article CAS PubMed PubMed Central Google Scholar
Stadelmann C, Deckwerth TL, Srinivasan A, Bancher C, Bruck W, Jellinger K, Lassmann H. Activation of caspase-3 in single neurons and autophagic granules of granulovacuolar degeneration in Alzheimer’s disease. Evidence for apoptotic cell death. Am J Pathol. 1999;155:1459–66.
Article CAS PubMed PubMed Central Google Scholar
Goel P, Chakrabarti S, Goel K, Bhutani K, Chopra T, Bali S. Neuronal cell death mechanisms in Alzheimer’s disease: an insight. Front Mol Neurosci. 2022;15: 937133.
Article CAS PubMed PubMed Central Google Scholar
Hansen DV, Hanson JE, Sheng M. Microglia in Alzheimer’s disease. J Cell Biol. 2018;217:459–72.
Article CAS PubMed PubMed Central Google Scholar
Leng F, Edison P. Neuroinflammation and microglial activation in Alzheimer disease: where do we go from here? Nat Rev Neurol. 2021;17:157–72.
Article PubMed Google Scholar
Fan Z, Brooks DJ, Okello A, Edison P. An early and late peak in microglial activation in Alzheimer’s disease trajectory. Brain. 2017;140:792–803.
PubMed PubMed Central Google Scholar
Solito E, Sastre M. Microglia function in Alzheimer’s disease. Front Pharmacol. 2012;3:14.
Article PubMed PubMed Central Google Scholar
Wyss-Coray T. Inflammation in Alzheimer disease: driving force, bystander or beneficial response? Nat Med. 2006;12:1005–15.
CAS PubMed Google Scholar
Long HZ, Zhou ZW, Cheng Y, Luo HY, Li FJ, Xu SG, Gao LC. The role of microglia in Alzheimer’s disease from the perspective of immune inflammation and iron metabolism. Front Aging Neurosci. 2022;14: 888989.
Article CAS PubMed PubMed Central Google Scholar
Dolan MJ, Therrien M, Jereb S, Kamath T, Gazestani V, Atkeson T, Marsh SE, Goeva A, Lojek NM, Murphy S, et al. Exposure of iPSC-derived human microglia to brain substrates enables the generation and manipulation of diverse transcriptional states in vitro. Nat Immunol. 2023;24:1382–90.
Article CAS PubMed PubMed Central Google Scholar
Abud EM, Ramirez RN, Martinez ES, Healy LM, Nguyen CHH, Newman SA, Yeromin AV, Scarfone VM, Marsh SE, Fimbres C, et al. iPSC-derived human microglia-like cells to study neurological diseases. Neuron. 2017;94(278–293): e279.
Google Scholar
Thomas AL, Lehn MA, Janssen EM, Hildeman DA, Chougnet CA. Naturally-aged microglia exhibit phagocytic dysfunction accompanied by gene expression changes reflective of underlying neurologic disease. Sci Rep. 2022;12:19471.
Article CAS PubMed PubMed Central Google Scholar
Koenigsknecht-Talboo J, Landreth GE. Microglial phagocytosis induced by fibrillar beta-amyloid and IgGs are differentially regulated by proinflammatory cytokines. J Neurosci. 2005;25:8240–9.
Article CAS PubMed PubMed Central Google Scholar
Rossol M, Heine H, Meusch U, Quandt D, Klein C, Sweet MJ, Hauschildt S. LPS-induced cytokine production in human monocytes and macrophages. Crit Rev Immunol. 2011;31:379–446.
Article CAS PubMed Google Scholar
Erta M, Quintana A, Hidalgo J. Interleukin-6, a major cytokine in the central nervous system. Int J Biol Sci. 2012;8:1254–66.
Article CAS PubMed PubMed Central Google Scholar
Feng Q, Wang YI, Yang Y. Neuroprotective effect of interleukin-6 in a rat model of cerebral ischemia. Exp Ther Med. 2015;9:1695–701.
Article PubMed PubMed Central Google Scholar
Lyra ESNM, Goncalves RA, Pascoal TA, Lima-Filho RAS, Resende EPF, Vieira ELM, Teixeira AL, de Souza LC, Peny JA, Fortuna JTS, et al. Pro-inflammatory interleukin-6 signaling links cognitive impairments and peripheral metabolic alterations in Alzheimer’s disease. Transl Psychiatry. 2021;11:251.
Article Google Scholar
Andreadou M, Ingelfinger F, De Feo D, Cramer TLM, Tuzlak S, Friebel E, Schreiner B, Eede P, Schneeberger S, Geesdorf M, et al. IL-12 sensing in neurons induces neuroprotective CNS tissue adaptation and attenuates neuroinflammation in mice. Nat Neurosci. 2023;26:1701–12.
Article CAS PubMed PubMed Central Google Scholar
Yang HS, Zhang C, Carlyle BC, Zhen SY, Trombetta BA, Schultz AP, Pruzin JJ, Fitzpatrick CD, Yau WW, Kirn DR, et al. Plasma IL-12/IFN-gamma axis predicts cognitive trajectories in cognitively unimpaired older adults. Alzheimers Dement. 2022;18:645–53.
Article CAS PubMed Google Scholar
Vom Berg J, Prokop S, Miller KR, Obst J, Kalin RE, Lopategui-Cabezas I, Wegner A, Mair F, Schipke CG, Peters O, et al. Inhibition of IL-12/IL-23 signaling reduces Alzheimer’s disease-like pathology and cognitive decline. Nat Med. 2012;18:1812–9.
Article Google Scholar
Trinchieri G. Interleukin-12: a proinflammatory cytokine with immunoregulatory functions that bridge innate resistance and antigen-specific adaptive immunity. Annu Rev Immunol. 1995;13:251–76.
Article CAS PubMed Google Scholar
Takata M, Nishimura K, Harada K, Iwasaki R, Ando M, Yamada S, Ginhoux F, Takata K. Analysis of Abeta-induced neurotoxicity and microglial responses in simple two- and three-dimensional human iPSC-derived cortical culture systems. Tissue Cell. 2023;81: 102023.
Article CAS PubMed Google Scholar
Alkhalifa AE, Al-Ghraiybah NF, Odum J, Shunnarah JG, Austin N, Kaddoumi A. Blood–brain barrier breakdown in Alzheimer’s disease: mechanisms and targeted strategies. Int J Mol Sci. 2023;24:16288.
Article CAS PubMed PubMed Central Google Scholar
Sun M, You H, Hu X, Luo Y, Zhang Z, Song Y, An J, Lu H. Microglia–astrocyte interaction in neural development and neural pathogenesis. Cells. 2023; 12.
Matejuk A, Ransohoff RM. Crosstalk between astrocytes and microglia: an overview. Front Immunol. 2020;11:1416.
Article CAS PubMed PubMed Central Google Scholar
Chun H, Lee CJ. Reactive astrocytes in Alzheimer’s disease: a double-edged sword. Neurosci Res. 2018;126:44–52.
Article CAS PubMed Google Scholar
Shinohara M, Tachibana M, Kanekiyo T, Bu G. Role of LRP1 in the pathogenesis of Alzheimer’s disease: evidence from clinical and preclinical studies. J Lipid Res. 2017;58:1267–81.
Article CAS PubMed PubMed Central Google Scholar
Liu CC, Hu J, Zhao N, Wang J, Wang N, Cirrito JR, Kanekiyo T, Holtzman DM, Bu G. Astrocytic LRP1 mediates brain abeta clearance and impacts amyloid deposition. J Neurosci. 2017;37:4023–31.
Article CAS PubMed PubMed Central Google Scholar
Habib N, McCabe C, Medina S, Varshavsky M, Kitsberg D, Dvir-Szternfeld R, Green G, Dionne D, Nguyen L, Marshall JL, et al. Disease-associated astrocytes in Alzheimer’s disease and aging. Nat Neurosci. 2020;23:701–6.
Article CAS PubMed PubMed Central Google Scholar
Foster EM, Dangla-Valls A, Lovestone S, Ribe EM, Buckley NJ. Clusterin in Alzheimer’s disease: mechanisms, genetics, and lessons from other pathologies. Front Neurosci. 2019;13:164.
Article PubMed PubMed Central Google Scholar
Kamphuis W, Kooijman L, Orre M, Stassen O, Pekny M, Hol EM. GFAP and vimentin deficiency alters gene expression in astrocytes and microglia in wild-type mice and changes the transcriptional response of reactive glia in mouse model for Alzheimer’s disease. Glia. 2015;63:1036–56.
Article PubMed Google Scholar
Drummond E, Nayak S, Faustin A, Pires G, Hickman RA, Askenazi M, Cohen M, Haldiman T, Kim C, Han X, et al. Proteomic differences in amyloid plaques in rapidly progressive and sporadic Alzheimer’s disease. Acta Neuropathol. 2017;133:933–54.
Article CAS PubMed PubMed Central Google Scholar
Kim J, Yoo ID, Lim J, Moon JS. Pathological phenotypes of astrocytes in Alzheimer’s disease. Exp Mol Med. 2024;56:95–9.
Article CAS PubMed PubMed Central Google Scholar
Bakken TE, Hodge RD, Miller JA, Yao Z, Nguyen TN, Aevermann B, Barkan E, Bertagnolli D, Casper T, Dee N, et al. Single-nucleus and single-cell transcriptomes compared in matched cortical cell types. PLoS ONE. 2018;13: e0209648.
Article PubMed PubMed Central Google Scholar
Mathys H, Davila-Velderrain J, Peng Z, Gao F, Mohammadi S, Young JZ, Menon M, He L, Abdurrob F, Jiang X, et al. Single-cell transcriptomic analysis of Alzheimer’s disease. Nature. 2019;570:332–7.
Article CAS PubMed PubMed Central Google Scholar
Kamphuis W, Middeldorp J, Kooijman L, Sluijs JA, Kooi EJ, Moeton M, Freriks M, Mizee MR, Hol EM. Glial fibrillary acidic protein isoform expression in plaque related astrogliosis in Alzheimer’s disease. Neurobiol Aging. 2014;35:492–510.
Article CAS PubMed Google Scholar
Jiwaji Z, Tiwari SS, Aviles-Reyes RX, Hooley M, Hampton D, Torvell M, Johnson DA, McQueen J, Baxter P, Sabari-Sankar K, et al. Reactive astrocytes acquire neuroprotective as well as deleterious signatures in response to Tau and Ass pathology. Nat Commun. 2022;13:135.
Article CAS PubMed PubMed Central Google Scholar
Ganne A, Balasubramaniam M, Griffin WST, Shmookler Reis RJ, Ayyadevara S. Glial fibrillary acidic protein: a biomarker and drug target for Alzheimer’s disease. Pharmaceutics. 2022;14:1354.
Article CAS PubMed PubMed Central Google Scholar
Dai DL, Li M, Lee EB. Human Alzheimer’s disease reactive astrocytes exhibit a loss of homeostastic gene expression. Acta Neuropathol Commun. 2023;11:127.
Article CAS PubMed PubMed Central Google Scholar
D’Ezio V, Colasanti M, Persichini T. Amyloid-beta 25–35 induces neurotoxicity through the up-regulation of astrocytic system X(c)(). Antioxidants (Basel). 2021;10:1685.
Article CAS PubMed Google Scholar
Cui X, Zong S, Song W, Wang C, Liu Y, Zhang L, Xia P, Wang X, Zhao H, Wang L, Lu Z. Omaveloxolone ameliorates cognitive dysfunction in APP/PS1 mice by stabilizing the STAT3 pathway. Life Sci. 2023;335: 122261.
Article CAS PubMed Google Scholar
Soni P, Ammal Kaidery N, Sharma SM, Gazaryan I, Nikulin SV, Hushpulian DM, Thomas B. A critical appraisal of ferroptosis in Alzheimer’s and Parkinson’s disease: new insights into emerging mechanisms and therapeutic targets. Front Pharmacol. 2024;15:1390798.
Article CAS PubMed PubMed Central Google Scholar
Joly-Amado A, Hunter J, Quadri Z, Zamudio F, Rocha-Rangel PV, Chan D, Kesarwani A, Nash K, Lee DC, Morgan D, et al. CCL2 overexpression in the brain promotes glial activation and accelerates tau pathology in a mouse model of tauopathy. Front Immunol. 2020;11:997.
Article CAS PubMed PubMed Central Google Scholar
Bose S, Cho J. Role of chemokine CCL2 and its receptor CCR2 in neurodegenerative diseases. Arch Pharm Res. 2013;36:1039–50.
Article CAS PubMed Google Scholar
Wan J, Fu AK, Ip FC, Ng HK, Hugon J, Page G, Wang JH, Lai KO, Wu Z, Ip NY. Tyk2/STAT3 signaling mediates beta-amyloid-induced neuronal cell death: implications in Alzheimer’s disease. J Neurosci. 2010;30:6873–81.
Article CAS PubMed PubMed Central Google Scholar
Toral-Rios D, Patino-Lopez G, Gomez-Lira G, Gutierrez R, Becerril-Perez F, Rosales-Cordova A, Leon-Contreras JC, Hernandez-Pando R, Leon-Rivera I, Soto-Cruz I, et al. Activation of STAT3 regulates reactive astrogliosis and neuronal death induced by abetao neurotoxicity. Int J Mol Sci. 2020;21.
Levin EC, Acharya NK, Sedeyn JC, Venkataraman V, D’Andrea MR, Wang HY, Nagele RG. Neuronal expression of vimentin in the Alzheimer’s disease brain may be part of a generalized dendritic damage-response mechanism. Brain Res. 2009;1298:194–207.
Article CAS PubMed Google Scholar
Zhang H, Wu LM, Wu J. Cross-talk between apolipoprotein E and cytokines. Mediators Inflamm. 2011;2011: 949072.
Article PubMed PubMed Central Google Scholar
Yamazaki Y, Zhao N, Caulfield TR, Liu CC, Bu G. Apolipoprotein E and Alzheimer disease: pathobiology and targeting strategies. Nat Rev Neurol. 2019;15:501–18.
Article CAS PubMed PubMed Central Google Scholar
Chen Y, Strickland MR, Soranno A, Holtzman DM. Apolipoprotein E: structural insights and links to Alzheimer disease pathogenesis. Neuron. 2021;109:205–21.
Article CAS PubMed Google Scholar
Zhou Y, Song WM, Andhey PS, Swain A, Levy T, Miller KR, Poliani PL, Cominelli M, Grover S, Gilfillan S, et al. Human and mouse single-nucleus transcriptomics reveal TREM2-dependent and TREM2-independent cellular responses in Alzheimer’s disease. Nat Med. 2020;26:131–42.
Article CAS PubMed PubMed Central Google Scholar
Fernandez-Calle R, Konings SC, Frontinan-Rubio J, Garcia-Revilla J, Camprubi-Ferrer L, Svensson M, Martinson I, Boza-Serrano A, Venero JL, Nielsen HM, et al. APOE in the bullseye of neurodegenerative diseases: impact of the APOE genotype in Alzheimer’s disease pathology and brain diseases. Mol Neurodegener. 2022;17:62.
Article CAS PubMed PubMed Central Google Scholar
Jackson RJ, Hyman BT, Serrano-Pozo A. Multifaceted roles of APOE in Alzheimer disease. Nat Rev Neurol. 2024;20:457–74.
Article CAS PubMed Google Scholar
Verghese PB, Castellano JM, Holtzman DM. Apolipoprotein E in Alzheimer’s disease and other neurological disorders. Lancet Neurol. 2011;10:241–52.
Article CAS PubMed PubMed Central Google Scholar
Verghese PB, Castellano JM, Garai K, Wang Y, Jiang H, Shah A, Bu G, Frieden C, Holtzman DM. ApoE influences amyloid-beta (Abeta) clearance despite minimal apoE/Abeta association in physiological conditions. Proc Natl Acad Sci USA. 2013;110:E1807-1816.
Article CAS PubMed PubMed Central Google Scholar
Castellano JM, Kim J, Stewart FR, Jiang H, DeMattos RB, Patterson BW, Fagan AM, Morris JC, Mawuenyega KG, Cruchaga C, et al. Human apoE isoforms differentially regulate brain amyloid-beta peptide clearance. Sci Transl Med. 2011;3:8957.
Article Google Scholar
Holtzman DM, Bales KR, Wu S, Bhat P, Parsadanian M, Fagan AM, Chang LK, Sun Y, Paul SM. Expression of human apolipoprotein E reduces amyloid-beta deposition in a mouse model of Alzheimer’s disease. J Clin Invest. 1999;103:R15–21.
Article CAS PubMed PubMed Central Google Scholar
Xia Z, Prescott EE, Urbanek A, Wareing HE, King MC, Olerinyova A, Dakin H, Leah T, Barnes KA, Matuszyk MM, et al. Co-aggregation with Apolipoprotein E modulates the function of Amyloid-beta in Alzheimer’s disease. Nat Commun. 2024;15:4695.
Article CAS PubMed PubMed Central Google Scholar
Spangenberg E, Severson PL, Hohsfield LA, Crapser J, Zhang J, Burton EA, Zhang Y, Spevak W, Lin J, Phan NY, et al. Sustained microglial depletion with CSF1R inhibitor impairs parenchymal plaque development in an Alzheimer’s disease model. Nat Commun. 2019;10:3758.
Article PubMed PubMed Central Google Scholar
Spangenberg EE, Lee RJ, Najafi AR, Rice RA, Elmore MR, Blurton-Jones M, West BL, Green KN. Eliminating microglia in Alzheimer’s mice prevents neuronal loss without modulating amyloid-beta pathology. Brain. 2016;139:1265–81.
Article PubMed PubMed Central Google Scholar
Musiek ES, Holtzman DM. Three dimensions of the amyloid hypothesis: time, space and “wingmen.” Nat Neurosci. 2015;18:800–6.
Article CAS PubMed PubMed Central Google Scholar
Rajmohan R, Reddy PH. Amyloid-beta and phosphorylated tau accumulations cause abnormalities at synapses of Alzheimer’s disease neurons. J Alzheimers Dis. 2017;57:975–99.
Article CAS PubMed PubMed Central Google Scholar
Hefti MM, Kim S, Bell AJ, Betters RK, Fiock KL, Iida MA, Smalley ME, Farrell K, Fowkes ME, Crary JF. Tau phosphorylation and aggregation in the developing human brain. J Neuropathol Exp Neurol. 2019;78:930–8.
Article CAS PubMed PubMed Central Google Scholar
Oakley DH, Chung M, Abrha S, Hyman BT, Frosch MP. beta-Amyloid species production and tau phosphorylation in iPSC-neurons with reference to neuropathologically characterized matched donor brains. J Neuropathol Exp Neurol. 2024;83:772–82.
Article CAS PubMed PubMed Central Google Scholar
Ochalek A, Mihalik B, Avci HX, Chandrasekaran A, Teglasi A, Bock I, Giudice ML, Tancos Z, Molnar K, Laszlo L, et al. Neurons derived from sporadic Alzheimer’s disease iPSCs reveal elevated TAU hyperphosphorylation, increased amyloid levels, and GSK3B activation. Alzheimers Res Ther. 2017;9:90.
Article PubMed PubMed Central Google Scholar
Choi SH, Kim YH, Hebisch M, Sliwinski C, Lee S, D’Avanzo C, Chen H, Hooli B, Asselin C, Muffat J, et al. A three-dimensional human neural cell culture model of Alzheimer’s disease. Nature. 2014;515:274–8.
Article CAS PubMed PubMed Central Google Scholar
Gonzalez C, Armijo E, Bravo-Alegria J, Becerra-Calixto A, Mays CE, Soto C. Modeling amyloid beta and tau pathology in human cerebral organoids. Mol Psychiatry. 2018;23:2363–74.
Article CAS PubMed PubMed Central Google Scholar
Knupp A, Mishra S, Martinez R, Braggin JE, Szabo M, Kinoshita C, Hailey DW, Small SA, Jayadev S, Young JE. Depletion of the AD risk gene SORL1 selectively impairs neuronal endosomal traffic independent of amyloidogenic APP processing. Cell Rep. 2020;31: 107719.
Article CAS PubMed PubMed Central Google Scholar
Lee H, Aylward AJ, Pearse RV 2nd, Lish AM, Hsieh YC, Augur ZM, Benoit CR, Chou V, Knupp A, Pan C, et al. Cell-type-specific regulation of APOE and CLU levels in human neurons by the Alzheimer’s disease risk gene SORL1. Cell Rep. 2023;42: 112994.
Article CAS PubMed PubMed Central Google Scholar
Avey DR, Ng B, Vialle RA, Kearns NA, de Paiva Lopes K, Iatrou A, De Tissera S, Vyas H, Saunders DM, Flood DJ, et al. Uncovering plaque-glia niches in human Alzheimer’s disease brains using spatial transcriptomics. Biorxiv. 2024;22:586.
Google Scholar
Liu T, Zhu B, Liu Y, Zhang X, Yin J, Li X, Jiang L, Hodges AP, Rosenthal SB, Zhou L, et al. Multi-omic comparison of Alzheimer's variants in human ESC-derived microglia reveals convergence at APOE. J Exp Med. 2020; 217.
Takano-Kawabe K, Matoba K, Nakamura Y, Moriyama M. Low density lipoprotein receptor-related protein 2 expression and function in cultured astrocytes and microglia. Neurochem Res. 2024;49:199–211.
Article CAS PubMed Google Scholar
Quesnel MJ, Labonte A, Picard C, Bowie DC, Zetterberg H, Blennow K, Brinkmalm A, Villeneuve S, Poirier J. Alzheimer’s disease neuroimaging I, Group P-AR: osteopontin: a novel marker of pre-symptomatic sporadic Alzheimer’s disease. Alzheimers Dement. 2024;20:6008–31.
Article CAS PubMed PubMed Central Google Scholar
Qiu Y, Shen X, Ravid O, Atrakchi D, Rand D, Wight AE, Kim HJ, Liraz-Zaltsman S, Cooper I, Schnaider Beeri M, Cantor H. Definition of the contribution of an Osteopontin-producing CD11c(+) microglial subset to Alzheimer’s disease. Proc Natl Acad Sci USA. 2023;120: e2218915120.
Article CAS PubMed PubMed Central Google Scholar
Akiyama H, Tooyama I, Kawamata T, Ikeda K, McGeer PL. Morphological diversities of CD44 positive astrocytes in the cerebral cortex of normal subjects and patients with Alzheimer’s disease. Brain Res. 1993;632:249–59.
Article CAS PubMed Google Scholar
Wang Q, Klyubin I, Wright S, Griswold-Prenner I, Rowan MJ, Anwyl R. Alpha v integrins mediate beta-amyloid induced inhibition of long-term potentiation. Neurobiol Aging. 2008;29:1485–93.
Article CAS PubMed Google Scholar
Opanashuk LA, Mark RJ, Porter J, Damm D, Mattson MP, Seroogy KB. Heparin-binding epidermal growth factor-like growth factor in hippocampus: modulation of expression by seizures and anti-excitotoxic action. J Neurosci. 1999;19:133–46.
Article CAS PubMed PubMed Central Google Scholar
Maurya SK, Mishra J, Abbas S, Bandyopadhyay S. Cypermethrin stimulates GSK3beta-dependent Abeta and p-tau proteins and cognitive loss in young rats: reduced HB-EGF signaling and downstream neuroinflammation as critical regulators. Mol Neurobiol. 2016;53:968–82.
Article CAS PubMed Google Scholar
Mansour HM, Fawzy HM, El-Khatib AS, Khattab MM. Repurposed anti-cancer epidermal growth factor receptor inhibitors: mechanisms of neuroprotective effects in Alzheimer’s disease. Neural Regen Res. 2022;17:1913–8.
Article CAS PubMed PubMed Central Google Scholar
Wang L, Chiang HC, Wu W, Liang B, Xie Z, Yao X, Ma W, Du S, Zhong Y. Epidermal growth factor receptor is a preferred target for treating amyloid-beta-induced memory loss. Proc Natl Acad Sci USA. 2012;109:16743–8.
Article CAS PubMed PubMed Central Google Scholar
Jayaswamy PK, Vijaykrishnaraj M, Patil P, Alexander LM, Kellarai A, Shetty P. Implicative role of epidermal growth factor receptor and its associated signaling partners in the pathogenesis of Alzheimer’s disease. Ageing Res Rev. 2023;83: 101791.
Article CAS PubMed Google Scholar
Woo RS, Lee JH, Yu HN, Song DY, Baik TK. Expression of ErbB4 in the neurons of Alzheimer’s disease brain and APP/PS1 mice, a model of Alzheimer’s disease. Anat Cell Biol. 2011;44:116–27.
Article PubMed PubMed Central Google Scholar
Zhang H, Zhang L, Zhou D, Li H, Xu Y. ErbB4 mediates amyloid beta-induced neurotoxicity through JNK/tau pathway activation: implications for Alzheimer’s disease. J Comp Neurol. 2021;529:3497–512.
Article CAS PubMed Google Scholar
Shaftel SS, Griffin WS, O’Banion MK. The role of interleukin-1 in neuroinflammation and Alzheimer disease: an evolving perspective. J Neuroinflammation. 2008;5:7.
Article PubMed PubMed Central Google Scholar
Kitazawa M, Cheng D, Tsukamoto MR, Koike MA, Wes PD, Vasilevko V, Cribbs DH, LaFerla FM. Blocking IL-1 signaling rescues cognition, attenuates tau pathology, and restores neuronal beta-catenin pathway function in an Alzheimer’s disease model. J Immunol. 2011;187:6539–49.
Article CAS PubMed Google Scholar
Lopez-Rodriguez AB, Hennessy E, Murray CL, Nazmi A, Delaney HJ, Healy D, Fagan SG, Rooney M, Stewart E, Lewis A, et al. Acute systemic inflammation exacerbates neuroinflammation in Alzheimer’s disease: IL-1beta drives amplified responses in primed astrocytes and neuronal network dysfunction. Alzheimers Dement. 2021;17:1735–55.
Article CAS PubMed Google Scholar
Labandeira-Garcia JL, Costa-Besada MA, Labandeira CM, Villar-Cheda B, Rodriguez-Perez AI. Insulin-like growth factor-1 and neuroinflammation. Front Aging Neurosci. 2017;9:365.
Article PubMed PubMed Central Google Scholar
Sohrabi M, Floden AM, Manocha GD, Klug MG, Combs CK. IGF-1R inhibitor ameliorates neuroinflammation in an Alzheimer’s disease transgenic mouse model. Front Cell Neurosci. 2020;14:200.
Article CAS PubMed PubMed Central Google Scholar
Zhou Y, Chen Y, Xu C, Zhang H, Lin C. TLR4 targeting as a promising therapeutic strategy for Alzheimer disease treatment. Front Neurosci. 2020;14: 602508.
Article PubMed PubMed Central Google Scholar
Zhang X, Sun D, Zhou X, Zhang C, Yin Q, Chen L, Tang Y, Liu Y, Morozova-Roche LA. Proinflammatory S100A9 stimulates TLR4/NF-kappaB signaling pathways causing enhanced phagocytic capacity of microglial cells. Immunol Lett. 2023;255:54–61.
Article CAS PubMed Google Scholar
Frew J, Baradaran-Heravi A, Balgi AD, Wu X, Yan TD, Arns S, Shidmoossavee FS, Tan J, Jaquith JB, Jansen-West KR, et al. Premature termination codon readthrough upregulates progranulin expression and improves lysosomal function in preclinical models of GRN deficiency. Mol Neurodegener. 2020;15:21.
Article CAS PubMed PubMed Central Google Scholar
Lee C, Frew J, Weilinger NL, Wendt S, Cai W, Sorrentino S, Wu X, MacVicar BA, Willerth SM, Nygaard HB. hiPSC-derived GRN-deficient astrocytes delay spiking activity of developing neurons. Neurobiol Dis. 2023;181: 106124.
Article CAS PubMed Google Scholar
Rose SE, Frankowski H, Knupp A, Berry BJ, Martinez R, Dinh SQ, Bruner LT, Willis SL, Crane PK, Larson EB, et al. Leptomeninges-derived induced pluripotent stem cells and directly converted neurons from autopsy cases with varying neuropathologic backgrounds. J Neuropathol Exp Neurol. 2018;77:353–60.
Article CAS PubMed PubMed Central Google Scholar
Washer SJ, Perez-Alcantara M, Chen Y, Steer J, James WS, Trynka G, Bassett AR, Cowley SA. Single-cell transcriptomics defines an improved, validated monoculture protocol for differentiation of human iPSC to microglia. Sci Rep. 2022;12:19454.
Article CAS PubMed PubMed Central Google Scholar
Stine WB, Jungbauer L, Yu C, LaDu MJ. Preparing synthetic Abeta in different aggregation states. Methods Mol Biol. 2011;670:13–32.
Article CAS PubMed PubMed Central Google Scholar
Dana H, Mohar B, Sun Y, Narayan S, Gordus A, Hasseman JP, Tsegaye G, Holt GT, Hu A, Walpita D, et al. Sensitive red protein calcium indicators for imaging neural activity. Elife. 2016;5.
Wang Y, DelRosso NV, Vaidyanathan TV, Cahill MK, Reitman ME, Pittolo S, Mi X, Yu G, Poskanzer KE. Accurate quantification of astrocyte and neurotransmitter fluorescence dynamics for single-cell and population-level physiology. Nat Neurosci. 2019;22:1936–44.
Article CAS PubMed PubMed Central Google Scholar
Wendt S, Johnson S, Weilinger NL, Groten C, Sorrentino S, Frew J, Yang L, Choi HB, Nygaard HB, MacVicar BA. Simultaneous imaging of redox states in dystrophic neurites and microglia at Abeta plaques indicate lysosome accumulation not microglia correlate with increased oxidative stress. Redox Biol. 2022;56: 102448.
Article CAS PubMed PubMed Central Google Scholar
Ku T, Swaney J, Park JY, Albanese A, Murray E, Cho JH, Park YG, Mangena V, Chen J, Chung K. Multiplexed and scalable super-resolution imaging of three-dimensional protein localization in size-adjustable tissues. Nat Biotechnol. 2016;34:973–81.
Article CAS PubMed PubMed Central Google Scholar
Raudvere U, Kolberg L, Kuzmin I, Arak T, Adler P, Peterson H, Vilo J. g:Profiler: a web server for functional enrichment analysis and conversions of gene lists (2019 update). Nucleic Acids Res. 2019;47:W191–8.
Article CAS PubMed PubMed Central Google Scholar

Download references

Acknowledgements

We would like to thank Ken McArthur, the Aunie Foundation and the Copland family for their generous donations to fund this project. We also thank the Alzheimer’s Association, Canadians For Leading Edge Alzheimer Research (CLEAR) and the Krembil foundation for their support. For identifying the APOE genotype of the used iPSC line we thank Dr. Mari DeMarco for assistance.

Funding

We would like to thank Ken McArthur, the Aunie Foundation and the Copland family for their generous donations to fund this project. We also thank the Alzheimer’s Association, Canadians For Leading Edge Alzheimer Research (CLEAR) and the Krembil foundation for their support.

Author information

Authors and Affiliations

Department of Psychiatry, Djavad Mowafaghian Centre for Brain Health, University of British Columbia, Vancouver, V6 T 1Z3, Canada
Stefan Wendt, Sarah N. Ebert, Yanyang Bai, Christopher J. Groten, Hyun B. Choi, Mathias Delhaye & Brian A. MacVicar
Division of Neurology, Djavad Mowafaghian Centre for Brain Health, University of British Columbia, Vancouver, V6 T 1Z3, Canada
Ada J. Lin, Declan J. Brennan, Wenji Cai, Da Young Kong, Stefano Sorrentino, Christopher Lee, Jonathan Frew & Haakon B. Nygaard
Department of Medical Genetics, University of British Columbia, Vancouver, V6 T 1Z3, Canada
David R. Kaplan & Freda D. Miller
Michael Smith Laboratories, University of British Columbia, Vancouver, BC, V6 T 1Z4, Canada
Sarah N. Ebert & Freda D. Miller
Opalia Co, Montreal, QC, H2X 3Y7, Canada
Jonathan Frew
Program in Neurosciences and Mental Health, Hospital for Sick Children, Toronto, ON, M5G 0 A4, Canada
Konstantina Karamboulas & David R. Kaplan
Department of Pathology, Vancouver General Hospital, University of British Columbia, Vancouver, BC, V5Z 1M9, Canada
Ian R. Mackenzie
Department of Molecular Genetics, University of Toronto, Toronto, ON, M5S 1 A8, Canada
David R. Kaplan

Authors

Stefan Wendt
View author publications
You can also search for this author inPubMed Google Scholar
Ada J. Lin
View author publications
You can also search for this author inPubMed Google Scholar
Sarah N. Ebert
View author publications
You can also search for this author inPubMed Google Scholar
Declan J. Brennan
View author publications
You can also search for this author inPubMed Google Scholar
Wenji Cai
View author publications
You can also search for this author inPubMed Google Scholar
Yanyang Bai
View author publications
You can also search for this author inPubMed Google Scholar
Da Young Kong
View author publications
You can also search for this author inPubMed Google Scholar
Stefano Sorrentino
View author publications
You can also search for this author inPubMed Google Scholar
Christopher J. Groten
View author publications
You can also search for this author inPubMed Google Scholar
Christopher Lee
View author publications
You can also search for this author inPubMed Google Scholar
Jonathan Frew
View author publications
You can also search for this author inPubMed Google Scholar
Hyun B. Choi
View author publications
You can also search for this author inPubMed Google Scholar
Konstantina Karamboulas
View author publications
You can also search for this author inPubMed Google Scholar
Mathias Delhaye
View author publications
You can also search for this author inPubMed Google Scholar
Ian R. Mackenzie
View author publications
You can also search for this author inPubMed Google Scholar
David R. Kaplan
View author publications
You can also search for this author inPubMed Google Scholar
Freda D. Miller
View author publications
You can also search for this author inPubMed Google Scholar
Brian A. MacVicar
View author publications
You can also search for this author inPubMed Google Scholar
Haakon B. Nygaard
View author publications
You can also search for this author inPubMed Google Scholar

Contributions

Conceptualization, H.B.N., S.W.; Methodology, S.W., S.N.E., F.M., K.K., J.F.; Experimental and Analysis, S.W., A.J.L., S.N.E., D.J.B., W.C., Y.B., D.Y.K., H.B.C., M.D.; S.S., C.J.G., C.L.; Resources, I.R.M.; Writing, S.W., H.B.N., A.J.L., B.A.M., F.M.; Visualization, S.W., A.J.L, K.K.; Supervision, H.B.N., B.A.M., F.M., D.R.K.; Funding Acquisition, H.B.N., B.A.M., F.M., S.W.

Corresponding authors

Correspondence to Stefan Wendt or Haakon B. Nygaard.

Ethics declarations

Ethics approval and consent to participate

Our work on hiPSCs was approved by the University of British Columbia Clinical Research Ethics Board (H21 - 02261).

Competing interests

The authors declare no competing interests.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

12974_2025_3433_MOESM1_ESM.pdf

Supplementary Material 1: Figure S1: Alternative marker expression patterns confirming the presence of neurons, astrocytes and microglia-like cells in hiNS. A: Cell population annotations from our snRNA-seq data set. B: Neuronal nuclei labeled by immunostaining for NeuN and its gene expression pattern shown on the right. C: Astrocyte/NPC expression of SLC7 A11 overlaps with GFAP labelled by immunostaining and their gene expression patterns shown on the right. D: Microglia-like cells were labelled with lectin, overlapping with CD45 immunostaining and its gene expression pattern shown on the right

12974_2025_3433_MOESM2_ESM.pdf

Supplementary Material 2: Figure S2: Maintaining neural wave activity requires glutamate transporter activity in hiNS. A: Glutamate transporter EAAT2and EAAT1expression in hiNS. While SLC1 A2 expression is widespread in neurons, progenitors and astrocytes, SLC1 A3 expression appears more specific for progenitor and astrocyte cell populations. B: iGluSnFr fluorescence, indicating extracellular glutamate waves. C: Extracellular glutamate levels gradually increase by blocking EAAT1/2 with 100 nM TFB-TBOAwhich prevents wave formation within minutes in hiNS indicating a functional role of glutamate transporters in maintaining calcium/glutamate wave activity

12974_2025_3433_MOESM3_ESM.pdf

Supplementary Material 3: Figure S3: hiMG differentiation timeline shows a gradual increase in IBA1 expression. A: hiMG were plated on coverslips and fixed at different time points during differentiation. Immunofluorescence for IBA1is detectable at all time points. Co-staining with tomato lectin- 488 confirms that lectin can be used to label hiMG in vitro. B: Western blot for IBA1 confirms stable expression in hiMG up to at least day 21. C: Stimulation of hiMG with LPS for 24 h results in significant TNFα and IL1β release confirming their capacity to react to proinflammatory stimuli. D: Full length blots for GAPDH, IBA1, β-actin, Il1β and TNFα. E: All individual blots for Il1β and TNFα used for the quantification in panel C

12974_2025_3433_MOESM4_ESM.pdf

Supplementary Material 4: Figure S4: Aβ aggregates form within 7 days of oligomeric Abtreatment in hiNS. A: hiNS cell culture medium was supplemented with three different doses of oAβ. Only the highestconcentration resulted in robust formation of aggregated Aβpositive area in hiNS. One-way ANOVA followed by Holm-Sidak’s post-hoc test C: Three-dimensional reconstruction of Aβ aggregates in hiNS. While oAβ treatment results in plaque-like aggregates inside the tissue, supplementing cell culture media with pre-aggregated fibrillar Aβresults in large Aβ sheets covering the hiNS tissue which do not mimic plaque-like structures

12974_2025_3433_MOESM5_ESM.pdf

Supplementary Material 5: Figure S5: Over-night live confocal imaging of hSyn -roGFP1 expressing hiNSduring chronic Aβ treatment. Recordings of 18 - 22 h duration were conducted between DIV 87 - 104 with chronic Aβ treatment ranging from 7 to 23 days. A: Example section of an Aβ treated hiNS displaying multiple neurons oxidizing during the recording time period. B: The total number of oxidizing cells per hour was quantified for both Aβ treated as well as control hiNS. In Aβ treated spheres 0.29 ± 0.02 cells oxidized per hour compared to 0.07 ± 0.01 under control conditions

12974_2025_3433_MOESM6_ESM.pdf

Supplementary Material 6: Figure S6: Chronic amyloidosis effects on hiNS derived from an alternative iPSC line). A: We employed our 5w Aβ treatment strategy and roGFP1 and GCaMP6f imaging to monitor Aβ induced neurodegeneration. B: Significant manifestation of oxidative stress, evident by an increase in roGFP ratios, after 27 days of Aβ exposure. Continuous Aβ exposure results in progressively more oxidative stress after 35 days of treatment. C: KOLF2.1 J hiNSdisplay calcium wave activity which frequencies get significantly reduced by chronic Aβ treatment. Statistical testing was performed using an unpaired student’s t-tests

12974_2025_3433_MOESM7_ESM.pdf

Supplementary Material 7: Figure S7: Chronic amyloidosis does not result in Aβ induced Tau hyperphosphorylation. A: Immunofluorescence staining for phosphorylated Tau T181and total Tau in hiNS of all 6 experimental conditions. B: Quantification of pTau181/total Tau ratio in hiNS indicates pTau modulation by hiMG independent of Aβ exposure. C: Soluble pTau181 was quantified using ELISA on supernatants from hiNSconfirms the lack of Aβ induced Tau hyperphosphorylation

12974_2025_3433_MOESM8_ESM.pdf

Supplementary Material 8: Figure S8: Example transmitted light images of control and Aβ treated hiNS with and without hiMG. hiMG attach to the surface surrounding hiNS Note that 5w Aβ hiNSdisplay a darker contrast and appear to detach from the surface

12974_2025_3433_MOESM9_ESM.pdf

Supplementary Material 9: Figure S9: Electrophysiological properties of hiMG in 3w and 5w Aβ hiNS. A: Current-voltage relationship of hiMG ranging from - 150 to + 50 mV after chronic Aβ treatment appears similar to hiMG from ctrl hiNSwith no significant increase in outward or inward currents. B: Reversal potentials of hiMG do not differ between treatment groups indicating membrane potentials of ~- 40 mV. c: Membrane capacitance of hiMG was significantly increased in hiMG of 5w Aβ hiNSindicating potentially larger cell sizes with prolonged Aβ treatment: N= 9 cells each)

12974_2025_3433_MOESM10_ESM.pdf

Supplementary Material 10: Figure S10: Neuronal gene expression changes during chronic amyloidosis in the presence of hiMG include APOE. A: Isolated neuronal cell populations from snRNA-seq on hiNSin ctrl, 3w and 5w Aβ conditions. Merged UMAPs of all 6 experimental groups displaying excitatory neuronsand inhibitory neurons. Note that the majority of 5w Aβ hiNSneurons cluster together in the individual neuronal populations. B: Quantification of total neuronal nuclei count in all 6 experimental groups. Note that only few neurons remain after chronic amyloidosis in the absence of hiMG. C: UMAP overlay of the three most significant DEGs APOE, SPP1 and FTL. Cell populations enriched with 5w Aβ hiNSneurons are encircled in dashed black lines. D: Heatmap depicting the top 15 most significant DEGs in 5w Aβ hiNSneurons, with APOE being the most significant upregulated gene. E: Neuronal gene ontology analysis in 3w/5w Aβ hiNSneurons. F: Direct comparison of neuronal DEGs depending on Ab or hiMG for ctrl and 3w Aβ hiNS shown in volcano plots. Top 10 DEGs in respect to fold change and/or adjusted pvalue are labeled. Note that AD-associated genes, such as CCL2, STAT3 and VIM, are only upregulated if hiMG were present during Aβ treatment

12974_2025_3433_MOESM11_ESM.pdf

Supplementary Material 11: Figure S11: Top 10 most significant DEGs per subcluster of hiMG from 5w Aβ hiNSdisplayed in a heatmap

12974_2025_3433_MOESM12_ESM.pdf

Supplementary Material 12: Figure S12: Additional western blots from hiNS supernatant for APOE. A: Full length blot for APOE. B: All individual blots used for the soluble APOE quantification as shown in Fig. 8C.

Supplementary Material 13: Supplementary Tables.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Wendt, S., Lin, A.J., Ebert, S.N. et al. A 3D human iPSC-derived multi-cell type neurosphere system to model cellular responses to chronic amyloidosis. J Neuroinflammation 22, 119 (2025). https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s12974-025-03433-3

Download citation

Received: 18 January 2025
Accepted: 31 March 2025
Published: 24 April 2025
DOI: https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s12974-025-03433-3

A 3D human iPSC-derived multi-cell type neurosphere system to model cellular responses to chronic amyloidosis

Abstract

Background

Methods

Results

Conclusions

Highlights

Introduction

Results

Formation of hiPSC-derived neurospheres as a 3D tissue culture system includes neurons, astrocytes and microglia

hiNS are highly functional 3D human tissue cultures

Chronic treatment of hiNS with oligomeric Aβ results in plaque-like aggregates and a series of neurotoxic effects

hiMG dependent phagocytosis and removal of plaque-like Aβ aggregates

Time-dependent neuroprotective properties of hiMG

hiMG alter astrocyte gene expression profiles in hiNS in response to Aβ

hiMG drive unique expression profiles in neurons, including APOE

Heterogenous gene expression profiles in hiMG

hiMG cytokine and ligand expression indicating signaling to neurons and astrocytes

APOE expression in hiNS is hiMG-dependent

Discussion

Functional characteristics of hiNS demonstrate healthy brain physiology

Chronic Aβ exposure models plaque-like aggregation and neurotoxic effects

Impact of hiMG on Aβ clearance and neuroprotection

AD-related astrogliosis and astrocyte activation states depend on hiMG

Gene expression changes in surviving neuronal populations after Aβ treatment

hiMG drive astrocytic and neuronal APOE expression during chronic amyloidosis

hiMG are required for APOE-Aβ co-localization

Amyloidosis triggered AD-like pathology independent of Tau hyperphosphorylation

Pathway analysis of hiMG ligands matching astrocytic and neuronal receptors

Limitations

Conclusions

Methods

Generation of human iPSCs and induction of NPCs

Neurosphere generation

AAV transduction of hiNS

Differentiation of human iPSC-derived microglia

hiMG transfer into neurospheres

ELISA

Oligomeric Aβ preparation

Fibrillar Aβ preparation

Chronic Aβ treatment paradigm

Confocal microscopy of neurospheres

Human formalin-fixed paraffin-embedded (FFPE) magnified analysis of proteome (MAP) tissue clearing and expansion

Deparaffinization and rehydration

Post-fixation

AA-integration

Inactivation

Monomer incubation

Mounting and gel embedding

Denaturation, clearing, and expansion

Immunofluorescence

Western Blot

Electrophysiology and 2-photon imaging

Single nuclei RNA sequencing

DEG statistical analysis

Gene ontology analysis

Ligand-receptor interaction analysis

Pearson and single-cell correlation analyses

Statistical analysis

Data availability

References

Acknowledgements

Funding

Author information

Authors and Affiliations

Contributions

Corresponding authors

Ethics declarations

Ethics approval and consent to participate

Competing interests

Additional information

Publisher's Note

Supplementary Information

Rights and permissions

About this article

Cite this article

Share this article

Journal of Neuroinflammation

Contact us