What this is
- This research investigates the early-stage pathology of cerebral amyloid angiopathy (CAA) in a mouse model of familial Danish dementia (FDD).
- It focuses on the glial responses, particularly the activation of and the role of .
- Findings suggest that early CAA pathology is characterized by severe astrogliosis without significant microglial reactivity, implicating immune and lipid processing networks.
Essence
- Early-stage CAA in the Tg-FDD mouse model shows severe astrogliosis with and decreased levels, indicating a distinct immune response compared to Alzheimer's disease.
Key takeaways
- Severe astrogliosis with an A1 astrocytic phenotype occurs in early-stage CAA without significant microglial activation. This contrasts with typical Alzheimer's disease pathology, where both astrocytes and microglia are activated.
- Dysregulation of immune response and lipid processing genes is evident in the Tg-FDD model, suggesting that these pathways may play a significant role in CAA pathology.
- The study reveals a decrease in protein levels in microglia, which may be influenced by , indicating a complex interplay between glial cells in the context of vascular amyloid deposition.
Caveats
- The findings are based on a mouse model, which may not fully replicate human CAA pathology. Further studies are needed to validate these results in human contexts.
- The study primarily focuses on early-stage pathology; the long-term implications of A1 astrocyte activation and dysregulation require further investigation.
Definitions
- A1 astrocytes: A neurotoxic subtype of astrocytes characterized by increased expression of complement component 3 (C3), linked to neurodegenerative diseases.
- TREM2: A receptor involved in immune response regulation; variants are associated with increased risk for Alzheimer's disease.
Simplified
Introduction
Alzheimer’s disease (AD) is the most common form of dementia and is characterized by the extracellular deposition of parenchymal ß-amyloid (Aß), intracellular accumulation of tau as neurofibrillary tangles (NFTs), synaptic loss, and significant inflammation [1, 2]. Cerebral amyloid angiopathy (CAA) is typified by the cerebrovascular deposition of amyloid and has a close molecular relationship with AD while remaining clinically distinct. Vascular amyloid accumulation is present in approximately 85–95% of individuals with AD [3, 4], positioning CAA as one of the strongest vascular contributors to age-related cognitive decline [5, 6]. The mechanisms responsible for CAA pathogenesis and its downstream effects on the brain are complex and incompletely understood. Despite the strong association of CAA with Aß accumulation [4], other amyloids also associate with the vasculature, suggesting that CAA is a group of biochemically and genetically diverse disorders unified by amyloid deposit accumulation in arterial blood vessel walls, and, in some cases, capillaries of the CNS parenchyma and leptomeninges [7–9].
Increasing evidence suggests a link between neuroinflammation and neuronal dysfunction in AD, orchestrated by the progressive activation of microglia and astrocytes, leading to the overproduction of pro-inflammatory molecules [10]. Microglial activation occurs early in AD pathology [11–13] with known roles in phagocytic activity; however, in the vicinity of amyloid plaques, these glial cells become pro-inflammatory, contributing to neurotoxicity in the late stages of disease [14, 15]. Astrocytes are also suggested to play a key role in AD progression and can be activated by amyloid plaques, resulting in the overexpression of cytokines and excessive oxidative stress [16]. The dysfunction between astrocytes and the neighboring neurons disrupts synaptic connections and initiates a cascade of neuronal injury [17, 18]. It has recently been shown that activation of microglia and astrocytes might not be independent events [19], and that activated microglia can directly polarize a subset of astrocytes (designated A1 astrocytes) toward a neurotoxic phenotype [19]. This astrocytic subtype is characterized by an increased expression of complement 3 (C3) [19]. A1 astrocytes have been identified in human cases and mouse models for AD, Huntington’s disease (HD), ALS, multiple sclerosis (MS), Parkinson’s disease (PD), prion disease, and frontotemporal dementia (FTD) [19–22]; however, the presence of A1 astrocytes has yet to be reported in the context of CAA.
Over 40% of genes identified as late-onset AD (LOAD) risk factors are immune-related genes predominantly expressed in microglia and astrocytes [23–25], enhancing the connection of the immune system to AD pathogenesis. Over the past decades, the molecular and cellular connections between parenchymal amyloid and glial activation and neuroinflammation have been widely studied [13, 26], while the relationship between vascular amyloid pathology and glial reactivity has mainly been reported based on neuropathological observations [27–29]. For instance, TREM2, one of the most studied AD-immune risk factors, has always been analyzed in the context of parenchymal amyloid [30–34], but no study has described its relationship with vascular amyloid pathology.
Therefore, we decided to dissect the glial responses associated with early-stage CAA in a mouse model of familial Danish dementia (FDD). FDD is characterized by the presence of CAA, consisting of the ~ 4-kDa ADan amyloid, in leptomeninges and vessels of the gray and white matter. Genetic analysis in patients with FDD revealed the presence of a 10-nucleotide duplication insertion in the 3′-end of the coding region of the BRI2 gene. This frameshift mutation generates an ADan precursor protein of 277 amino acids, of which the ~ 4-kDa ADan amyloid subunit comprises the last 34 amino acids [35]. Cotton wool-like plaques in the vicinity of blood vessels with amyloid and tau NFTs are also observed in FDD patients [36]. The mouse model for FDD (Tg-FDD) used in this study consistently exhibits CAA, primarily in leptomeningeal cerebellar vessels [37] and in the large and medium-sized parenchymal and penetrating vessels of the brain. At a late stage of CAA pathology, perivascular tau immunoreactive deposits and qualitative glial activation have also been observed in this model [38]. These observations make the Tg-FDD mice an ideal model for the study of glial reactivity and neuroimmune responses in early-stage CAA pathology.
In the present study, we show that early-stage vascular amyloid pathology in the Tg-FDD model is associated with severe astrogliosis with an A1 astrocytic phenotype without significant reactive microgliosis. We also observed that a genetic network of immune and lipid processing genes is highly dysregulated, directly impairing cholesterol, ApoE, and TREM2. Our results suggest that the lipid-ApoE-TREM2 pathway could play a preponderant role in CAA, as previously described in AD, and that in the context of vascular amyloid, astrogliosis is an early event in which activated astrocytes may directly influence microglial homeostasis.
Materials and methods
Transgenic mouse model
Tg-FDD and wild-type C57/BL6J (WT) (JAX stock #000664) male and female mice were used for our experiments, including cellular, biochemical, and immunohistochemistry (IHC) analyses. The Tg-FDD mouse model expresses an FDD-associated human mutant BRI2 transgene that leads to the vascular accumulation of ADan amyloid [37]. Mice were housed at the Indiana University School of Medicine (IUSM) animal care facility and were maintained according to USDA standards (12-hr light/dark cycle, food and water ad libitum), per the Guide for the Care and Use of Laboratory Animals (National Institutes of Health, Bethesda, MD). Animals were anesthetized and euthanized according to the IUSM Institutional Animal Care and Use Committee-approved procedures. Mice were deeply anesthetized prior to decapitation. After sacrifice, brains were removed and stored at − 80 °C or formalin-fixed as previously described [37]. For all described experiments, 3- or 9-month-old animals were utilized.
Mouse brain samples preparation and immunoblot analysis
WT and Tg-FDD brains were homogenized at a 1:10 (w/vol) ratio of brain and T-PER tissue protein extraction reagent with complete protease inhibitor cocktail (Roche), then sonicated for 30 s. Samples were then centrifuged at 13,200 rpm for 15 min at 4 °C. Next, samples were run on a NuPAGE 4–12% Bis-Tris protein gel (Invitrogen) and transferred to a nitrocellulose membrane. Primary antibodies used were anti-ApoE (1:1000, AB1907, Abcam), anti-TREM2 (1:10000, AF1729, R&D), and anti-Vinculin (1:10000, V9131, Sigma). The secondary antibodies used were goat anti-mouse HRP IgG (1:1500, A16066, Invitrogen) and goat anti-rabbit HRP IgG (1:1500, PI31460, Invitrogen). Western blot (WB) quantification results are expressed as the ratio of ApoE or TREM2 normalized by the loading control Vinculin. In all cases, we considered the control group to be 100%.
Brain sections immunofluorescence
Paraffin sections were deparaffinized in xylene and rehydrated in ethanol (EtOH) and washed with deionized water. Then, the sections were heated with a microwave oven in low pH antigen retrieval solution (eBioscience) twice for 4 min each. After washing in PBS twice for 5 min each, the sections were blocked with PBS 5% goat serum, 5% horse serum, 2% fish gel, and 0.01% Triton X-100 for 1 h at room temperature (RT). Sections were then incubated overnight at 4 °C with the following antibodies: anti-GFAP (G3893, Sigma-Aldrich), anti-IBA1 (MABN92, Millipore), anti-CD11B (1:100, AB8878, Abcam), anti-C3 (PA5-21349, ThermoFisher), anti-ADan 1699 (gift from Dr. Ruben Vidal), anti-Desmin (smooth muscle actin) (PA5-21349, Millipore), anti-ApoE (AB1907, Abcam), and anti-TREM2 (AF1729, R&D) diluted 1:100 in blocking solution. The next day, sections were quickly washed 3 times in PBS and incubated with 1:500 biotinylated horse anti-mouse antibody (BA-200, Vector) and/or biotinylated goat anti-rabbit antibody (BA-1000, Vector) for 1 hr at RT. Thirty minutes in advance, the Vectastain Elite ABC peroxidase kit (PK-6100, Vector) was prepared according to the manufacturer’s instructions. After the secondary antibody incubation, sections were incubated with the A + B solution for 30 min at RT. After 3 quick washes with PBS, tyramide dyes were prepared 1:500 in PBS, and slides were incubated with them for 10 min at RT. Slides were incubated with 3% H2O2 for 10 min at RT to stop peroxidase activity. Vascular amyloid was stained with 1% Thioflavin-S for 8 min at RT, followed by two washes in EtOH 50% and 30% for 3 min each, and a final wash in deionized water for 5 min. For cholesterol detection, sections were stained with filipin at a concentration of 250 μg/mL and incubated overnight at 4 °C, then rinsed in PBS 3 × 5 min in the dark to prevent fading. Finally, sections were washed in PBS and mounted with Vectashield mounting medium with or without DAPI (Vector Laboratories).
Astrocytes and microglia primary cell culture
Primary glial cell cultures were prepared from newborn (postnatal day 0–3) brain cortex of C57/BL6J mice. Briefly, animals were euthanized, and their brains extracted. Brains were cut into small pieces, collected in HBSS (H9269, Sigma), and treated with 2.5% trypsin (15090-046, Gibco) and 1% DNase (EN0521, Thermo) at 37 °C for 15 min. Then, the tissue was disaggregated by pipetting, passed through a 70-μm pore cell strainer (352350, Corning), and collected in FBS. Cells were centrifuged for 5 min at 1000×g, and the obtained pellet was resuspended in 10 mL glial medium (Advanced DMEM/F12, 10% FBS, 100 g/mL streptomycin, 100 UI/mL penicillin, 200 μg/mL glutamine). Cells were then counted using a Luna Dual Fluorescence Cell Counter (Logos Biosystem). Cells were plated into 75-cm2 cell culture flask at a density of 1 × 106 and incubated until 90% confluent, changing the media every 2 days. For primary microglia cell culture, confluent flasks were shaken for 1 hr at 200 rpm at 37 °C to detach microglial cells. Then, the supernatants containing microglia were collected and centrifuged for 5 min at 300×g. The cells were resuspended into 1 mL of glial media and counted using an automatic cell counter (Logos Biosystem) and seeded in 6-well culture plates for further experiments. The flask was treated with 0.5% trypsin (15400054, Gibco) for 10 min at 37 °C to detach the astrocytes. Then, the supernatant was collected, centrifuged at 300×g for 5 min, and the pellet was resuspended in glial medium. Astrocytes were quantified using an automated cell counter (Logos Biosystem) and seeded into 12-well (2 × 105 cells) plates for further experiments. Primary astrocytes were treated with A1 inducing cocktail IL-1α (3 ng), TNFα (30 ng), and C1q (400 ng) for 24 h to promote A1 induction. The next day, media was replaced with fresh serum-free media and was allowed to condition for another 24 h, which we define as astrocytic conditioned media (ACM). Control-ACM or A1-ACM was then collected and lyophilized with a Speed Vac Plus SC110A and resuspended in 500 μL of PBS. The total protein concentration was determined using the Pierce BCA protein assay kit (23246, Thermo Scientific), and 50 μg of total protein was added to mouse primary microglia.
Cell culture immunofluorescence
Astrocytes or microglia were seeded at 2 × 105 cells/well in 18 mm diameter coverslips. Once the culture reached 90% confluence, the cells were fixed in PBS containing 4% paraformaldehyde (PFA) for 15 min. Cells were permeabilized for 5 min at RT in 0.25% Triton X-100 in PBS, washed twice with PBS, and incubated for 1 h at 37 °C in PBS containing 5% goat, 5% horse serum, and 2% fish gel (blocking solution). Cells were then incubated overnight at 4 °C in primary antibody diluted 1:100 in blocking solution with anti-IBA1 (MABN92, Millipore) and anti-TREM2 (AF1729, R&D). After incubation, cells were washed with PBS, then incubated with Alexa 488-conjugated goat anti-mouse antibody (1:100, Invitrogen) and Alexa 568-conjugated goat anti-rabbit antibody (1:100, Invitrogen) for 1 h at RT, then washed with PBS and mounted with Vectashield mounting medium with DAPI (Vector Laboratories). Samples were examined using a Nikon A1-R laser scanning confocal microscope coupled with Nikon AR software for orthogonal images of reconstructed three-dimensional views. At least 5–10 cells were analyzed from each image. Eight to ten images were used for each experiment, and three independent culture experiments were performed.
Microscopy and image analysis
We used the ImageJ software (NIH) to create one index that represented changes in both astrocyte and microglia. The number of C3 (+), GFAP (+), or IBA1 (+) pixels was divided by the total number of pixels in the image and expressed as a GFAP (+) or IBA1 (+) area % [39] or C3 (+) GFAP (+) cells % [19]. For colocalization analysis, the ImageJ Coloc2 plugin was used for quantification and expressed as a percentage (% Colocalization Area). Microglia or astrocyte cell morphology was analyzed using IBA1 or GFAP and nuclear staining for Dapi. Cells were analyzed for the number of glial branches and junction using cross-sectional measurements of 10–30 randomly selected cells from 5 different micrographs per animal obtained from 20 μm image z-stacks as described [40]. The z-stack images for IBA1 or GFAP were obtained with a × 63 objective with a × 0.7 zoom and a 0.2-μm z-step. To fully analyze the cellular processes, only IBA1 (+) or GFAP (+) labeled cells in which the cell body was located toward the middle portion of the z-plane were selected for imaging. For every single staining, as a negative control, primary antibodies were omitted to determine background and autofluorescence (not shown). WT and Tg-FDD (3–4 animals per genotype) cerebral cortex and cerebellum were examined using a Leica DMi 8 epifluorescence microscope coupled with the LAS X program (Leica) or Nikon A1-R laser scanning confocal microscope coupled with Nikon AR software (Nikon).
qPCR
Total RNA was isolated from mouse brains with the RNeasy Plus Universal Mini Kit (Qiagen). cDNA was prepared from 1 μg total RNA with High-Capacity cDNA reverse transcription kit (Life Technologies). All qPCRs were performed on QuantStudio 6 Flex Real-Time PCR system (Life Technologies). The mouse TREM2 relative gene expression was evaluated with the delta Ct method using Taqman probe sets (TREM2: Mm00451744_m1, GAPDH 4351309, Life Technologies) and TaqMan Universal PCR Master Mix.
RNA sequencing
Library preparation and sequencing
Three mice per group were used for RNA sequencing experiments. The cerebellum was homogenized in Trizol, and total RNA was isolated by chloroform extraction as previously described [41]. The concentration and quality of total RNA samples were first assessed using an Agilent 2100 Bioanalyzer. An RNA Integrity Number (RIN) of five or higher was required to pass the quality control. Then, 500 ng of RNA per sample was used to prepare a dual-indexed strand-specific cDNA library using TruSeq Stranded mRNA Library Prep Kit (Illumina). The resulting libraries were assessed for quantity and size distribution using a Qubit and an Agilent 2100 Bioanalyzer, respectively. Two hundred picomolar pooled libraries were utilized per flowcell for clustering amplification on cBot using HiSeq 3000/4000 PE Cluster Kit and sequenced with 2 × 75 bp paired-end configuration on HiSeq4000 (Illumina) using HiSeq 3000/4000 PE SBS Kit. A Phred quality score (Q score) was used to measure the quality of sequencing. More than 97% of the sequencing reads reached Q30 (99.9% base call accuracy).
Sequence alignment and gene counts
The sequencing data were first assessed using FastQC (Babraham Bioinformatics, Cambridge, UK) for quality control. Then, all sequenced libraries were mapped to the mm10 mouse genome using STAR RNA-Seq aligner [42] with the following parameter: “--outSAMmapqUnique 60.” The reads distribution across the genome was assessed using bamutils (from ngsutils) [43]. Uniquely, mapped sequencing reads were assigned to mm10 refSeq genes using featureCounts (from subread) [44] with the following parameters: “-s 2 -p –Q 10.” Quality control of sequencing and mapping results was summarized using MultiQC [45]. Genes with read count per million (CPM) > 0.5 in more than 3 of the samples were kept. The data were normalized using TMM (trimmed mean of M values) method. Differential expression analysis was performed using edgeR [46, 47]. False discovery rate (FDR) was computed from p values using the Benjamini-Hochberg procedure.
Genomic pathway analysis
Ingenuity Pathway Analysis (IPA) core analysis was used to identify the perturbed gene networks in 9-month-old Tg-FDD mice compared to WT mice. The differential expression (DE) cutoff we used in the analysis is fold change greater than 1.2 and adjusted p value lower than 0.05. In the network figure generated by IPA, the genes marked by red color are upregulated while the genes marked by green are downregulated, and the main biological function of the network was provided. Then, for a specific perturbed gene network of interest, a heatmap showing differential expression of genes was generated to visualize the expression perturbation of genes contained in the network. Genes of interest were cross-referenced with the Agora open access portal. Agora hosts evidence for whether or not genes are associated with Alzheimer’s disease (AD). Agora also contains a list of over 500 nascent drug targets for AD that were nominated by AD researchers. The list of nominated targets was contributed by researchers from the National Institute on Aging’s Accelerating Medicines Partnership in Alzheimer’s Disease (AMP-AD) consortium as well as other research teams. Other evidence presented in Agora was either generated by AMP-AD research teams or is aggregated from publicly available data sources [48, 49].
Statistical analyses
Experimental analysis and data collection were performed blind unless otherwise stated. P values were determined using the appropriate statistical method via GraphPad Prism as described. Statistical comparisons were made using a two-tailed unpaired Student’s t test. Data are presented as mean ± SEM unless otherwise stated. *, **, ***, and **** denote p < 0.05, p < 0.01, p < 0.001, and p < 0.0001, respectively.
Data availability
The accession number for the RNA-Seq data reported in this paper is Gene Expression Omnibus (GEO): GSE150394↗.
Results
Reactive astrogliosis in early-stage vascular amyloid pathology in Tg-FDD mice
Glial immunoreactive changes in Tg-FDD mice.Double-stained immunofluorescent images of amyloid (Thio-S, green) and astrocytes (GFAP, red). Quantification of GFAParea (%) of WT or Tg-FDD mice.Double-stained immunofluorescent images of amyloid (Thio-S, green) and microglia (IBA1, red). Quantification of IBA1area (%) of WT or Tg-FDD mice. All are representative images of 9-month-old mice. Results are shown as the mean ± SEM of= 3–4, where *< 0.05 and **< 0.01 as determined by unpaired Student’stest. Scale bar 100 μm a b + + n p p t
Glial morphometric changes in Tg-FDD mice.Individual astrocytes (GFAP, green) were selected, isolated, thresholded, and skeletonized for analysis from brain sections of WT or Tg-FDD mice.Quantification of individual astrocyte number of branches and junctions from the hippocampus and cortex.Individual microglia (GFAP, green) were selected, isolated, thresholded, and skeletonized for analysis within cortex and hippocampus in brain sections of WT or Tg-FDD miceQuantification of individual microglia number of branches and junctions from the hippocampus and cortex.andare representative images of the hippocampus of 9-month-old mice. Results are shown as the mean ± SEM of= 4 per genotype and 10–30 astrocytes or microglia per animal, where *< 0.05 and ****< 0.0001 as determined by unpaired Student’stest. Scale bar 100 μm a b c d a c n p p t
A1 astrocytes are highly abundant in Tg-FDD mice
A1 Astrocytes are highly abundant in Tg-FDD.Double-stained immunofluorescent images of complement component 3 (C3, red) and astrocytes (GFAP, green) in the hippocampus of WT and Tg-FDD mice. C3 and GFAP immunoreactivity overlay (Merge).Quantification of C3GFAPcell % from 3 regions per animal of the hippocampus of 4 WT and 3 Tg-FDD mice.Quantification of % colocalization area from the hippocampus of 4 WT and 3 Tg-FDD mice.Double-stained immunofluorescent images of complement component 3 (C3, red) and astrocytes (GFAP, green) in the cerebellum of WT and Tg-FDD mice. C3 and GFAP immunoreactivity overlay (Merge).Quantification of C3GFAPcells % from 3 regions per animal of the cerebellum of 4 WT and 3 Tg-FDD mice.Quantification of % colocalization area from the cerebellum of 4 WT and 3 Tg-FDD mice.andare representative images of the brain regions of 9-month-old mice. Colocalization analysis (CC) was performed to determine pixel intensity correlation between C3 and GFAP signals. White pixels indicate colocalization between C3 and GFAP signal. Plot profiles of representative C3 and GFAP intensities show higher overlapping of C3 with GFAP in Tg-FDD mice in comparison with WT mice, indicating an increase of C3 positive astrocytes. Results are shown as the mean ± SEM of= 3–4. Asterisks indicate significant differences, where ****< 0.0001 by unpaired Student’s t test. Scale bar 100 μm a b c d e f a d + + + + n p
Dysregulation of immune response and lipid processing networks associated with CAA
Dysregulation of immune response and lipid processing network in Tg-FDD mice.Heatmap of gene expression (red = upregulation, blue = downregulation) andnetwork analysis of significantly affected genes determined by RNA-Seq in WT versus Tg-FDD mice (= 3; 9-month-old).Bolded genes from lipid processing and immune response network indicate genetic association with LOAD or if the gene has been nominated as a potential target for AD. logFC, log2 (fold change); FDR, false discovery rate a b c n
Cholesterol accumulation is associated to vascular amyloid in Tg-FDD mice.Double immunofluorescence of smooth muscle actin (SMA, red) and cholesterol fFilipin, green) in Tg-FDD mice show filipinstaining without association to SMA (Merge). Colocalization analysis (CC) was performed to determine pixel intensity correlation between filipin and SMA. White pixels indicate colocalization between filipin and SMA signal. Plot profiles of representative intensities showing almost no overlapping intensities of cholesterol (filipin, green) and SMA (red).Double immunofluorescence of Danish amyloid (ADan, red) and cholesterol (filipin, green) in Tg-FDD mice show strong filipincolocalization and overlapping intensities (Merge). Colocalization analysis confirmed major association between filipin and vascular amyloid. White pixels indicate colocalization between filipin and ADan (red) signal. Plot profiles of representative intensities showing major overlapping intensities of cholesterol (filipin, green) and ADan (red).Double immunofluorescence of astrocytes (GFAP, red) and cholesterol (filipin, green) in Tg-FDD mice show filipinshow a high degree of cellular association (Merge). White pixels in CC indicate colocalization between filipin and GFAP (red) signal. Plot profiles of representative intensities showing minor overlapping intensities of cholesterol (filipin, green) and GFAP (red). Scale bar 25 μm a b c + + +
ApoE is associated with vascular amyloid in Tg-FDD mice.,WB analysis and quantification of ApoE from the cerebral hemisphere and cerebellum of WT and Tg-FDD mice. Vinculin was used as a loading control. For quantifications, error bars represented ± SEM= 3, where *< 0.05 of ***< 0.001, unpaired Student’stest.Double immunofluorescence images of apolipoprotein E (ApoE, green) and Danish amyloid (ADan, red) in WT and Tg-FDD brain sections show a high degree of association of ApoE with vascular amyloid in the Tg-FDD (merge). All are representative images of 9-month-old mice cortex. Scale bar 100 μm a b c n p p t
Decreased TREM2 protein levels in Tg-FDD mice microglia.WB analysis and quantification of TREM2 protein levels in the cerebral hemisphere of WT and Tg-FDD mice.Double immunofluorescence images of microglia (IBA1, red) and TREM2 (green) in the hippocampus of WT and Tg-FDD mice.WB analysis and quantification of TREM2 protein levels in the cerebellum of WT and Tg-FDD mice.Double immunofluorescence images of microglia (IBA1, red) and TREM2 (green) in the cerebellum of WT and Tg-FDD mice. For WB analysis (and), TREM2 levels were normalized by vinculin. For WB quantifications, error bars represented ± SEM= 4, where *< 0.05 of ***< 0.001, unpaired Student’stest. For double immunofluorescence (and), a decrease of TREM2 (green) positive microglia (IBA1, red) was observed in WT versus Tg-FDD mice in both brain regions (Merge). Colocalization analysis (CC) was performed to determine pixel intensity correlation between TREM2 and IBA1. White pixels indicate colocalization between TREM2 and IBA1 signal. All are representative images of 9-month-old WT or Tg-FDD mice. Scale bar 100 μm a b c d a c b d n p p t
A1 astrocyte conditioned media decreases TREM2 protein in mouse primary microglia culture.Schematic diagram showing A1 astrocytes induction by cytokine cocktail treatment (TNFα, IL-1α, and C1q) followed by microglia incubation with astrocyte conditioned media (ACM) from A1 astrocytes or control astrocytes solely treated with PBS.Double immunofluorescence images of microglia (IBA1, red) and TREM2 (green) of primary WT mouse microglia culture treated with control or A1 ACM. Right panel shows orthogonal images of reconstructed three-dimensional views. c Intensity quantification of TREM2 per cell shows a decrease in TREM2 proteins in microglia incubated with A1 ACM.Quantification of IBA1cell surface area shows no change in cell surface area between control and A1 ACM-treated groups. For quantifications, error bars represented ± SEM where eight to ten images were used for each experiment with= 3 independent cultures and 20 cells per culture, where ****< 0.0001 by unpaired Student’stest. Scale bar 50 μm a b d + n p t
Discussion
This study demonstrates that early pathology in a mouse model of CAA is associated with severe astrogliosis with an A1 astrocytic phenotype without significant microgliosis. Additionally, the immune response and lipid processing network are dysregulated in the Tg-FDD model. Further, validation demonstrated an impairment of cholesterol, ApoE, and TREM2 in this CAA model, as has been broadly demonstrated in AD mouse models, characterized by the accumulation of parenchymal amyloid plaques. Finally, A1 astrocytes induce a decrease in TREM2 protein levels in microglial culture.
The role of glial cells in the pathogenesis of several neurodegenerative diseases, including AD, has been a major research focus in recent years [10–16]. Since astrocytes and microglia activation has been described in CAA patients [29, 76], we studied the early glial response associated with early CAA pathology in Tg-FDD mice. Here, we report that vascular amyloid deposition in 9-month-old Tg-FDD animals contributes to robust immunoreactive and morphometric changes in astrocytes with increased glial branches and junctions in early CAA. Interestingly, astrocytic alterations associated with a premature vascular amyloid accumulation have also been reported in transgenic artic β-amyloid (acrAβ) mice [77], suggesting that severe astrocytic reactivity in early CAA is independent of amyloid species. In response to early vascular amyloid accumulation, microglial reactivity is unchanged, with no CD11B or IBA1 immunoreactive or morphometric changes in glial branches or junctions. Traditionally, studies of AD mouse models with parenchymal amyloid accumulation report the presence of both activated astrocytes and microglia [78–81]; however, our findings demonstrated that in a mouse model of CAA robust astrogliosis independent of microglial immune activation is a primary response to vascular amyloid deposition. The early astrocytic response in CAA could be due to the central role of astrocytes in the neurovascular unit in maintaining BBB integrity and cerebral blood flow regulation in physiological and pathological conditions [82–84]. For instance, changes in vascular components due to CNS injury have a direct influence on astrocyte signaling [85]. Astrocytes respond to and influence immune and inflammatory responses following insults such as excitotoxicity, ischemia, apoptosis, necrosis, and inflammatory cues by undergoing a pronounced transformative state called “reactive astrogliosis”, regulating inflammatory responses that can be either neuroprotective or neurotoxic and participating in migratory, phagocytic, and proteolytic activity [72]. The notion that the astrocyte and microglia phenotype observed in 9-month-old Tg-FDD results from a response to vascular amyloid deposits as opposed to overexpression of the mutant form of human BRI2 is supported by the fact that the transgene expression is under the control of the mouse prion promoter, which is exclusively expressed in neurons [37]. Therefore, no BRI2 is over-expressed in glial cells. The role of CAA in glial response in this model is also supported by the observation that no astrogliosis is detected in 3-month-old Tg-FDD mice, an age where no vascular amyloid deposits are observed [37].
It has recently been shown that activated microglia may directly polarize A1 astrocytes toward a neurotoxic phenotype [19]. Induced by a cytokine cocktail composed by TNFα, IL-1α, and C1q, A1 astrocytes can be identified by their upregulation of C3 and are shown to lose many normal homeostatic functions, such as the promotion of neuronal survival, neurite outgrowth, and synapse formation. This suggests that A1s are either unable to maintain synapses or actively disassemble them by releasing multiple complement components that help drive synaptic degeneration. A1 astrocytes also exert a toxic gain of function by secreting soluble neurotoxin(s) that induce neuronal and oligodendrocyte death, supporting the notion that A1 astrocytes are involved in the development of neurodegenerative diseases [19, 86]. In this study, we report for the first time that A1 astrocytes are highly abundant in a mouse model of CAA. No activated microglia were observed at early stages of CAA pathology, suggesting that inflammatory signaling derived from another source may be responsible for, and sufficient, to induce A1 astrocytes in CAA. While it is unknown if damaged endothelial or vascular cells polarize astrocytes to an A1 type, each can produce pro-inflammatory cytokines under toxic stimulation, supporting the notion that these cells are capable of A1 astrocytic induction [87, 88]. Blocking A1 astrocyte conversion is neuroprotective in Parkinson’s disease models [21]; however, their abolishment accelerated prion disease course in a mouse model [22]. This suggests that the role of these astrocytes is disease-specific, and further analysis is required to determine their contribution to CAA pathogenesis.
Recently, genome-wide association studies (GWAS), whole-genome sequencing (WGS), and gene-expression network analyses have revealed gene networks and common and rare genetic variants that are associated with LOAD. The majority of these genes are involved in lipid processing, endocytosis, and innate immunity [55, 56, 89, 90]. Further validations of many of these risk factors have focused on their effect on parenchymal amyloid pathology, with their influence in vascular amyloid pathology largely understudied. Gene expression analyses revealed dysregulation of immune response networks associated with lipid processing in the Tg-FDD model. Our RNA-Seq results showed an upregulation of the rare AD-associated genetic factors ABCA7 and TREM2 [57, 58], an upregulation of lipoprotein lipase (LPL), known to be involved in AD susceptibility and pathogenesis [59, 60], and downregulation of CX3CL1, which has known roles in neuroprotection and neurotoxicity in the context of AD [61–64]. This suggests that many of the immune and lipid-related genes, identified as risk factors for AD, may have a preponderant involvement in CAA pathogenesis. In support of this, a recent population-based study tested the association of distinct neuropathological features of AD with risk loci identified in GWAS studies. Remarkably, the authors observed an association between the ABCA7 and TREM2 loci with CAA and capillary-Aβ, respectively [91], further demonstrating a role for immune-related AD risk factors in vascular amyloid deposition.
Multiple threads of evidence connect cholesterol with AD [92]. Epidemiological studies suggest a positive correlation between hypercholesterolemia and increased AD risk, though the exact impact and mechanisms involved remain largely unknown. Previous reports have shown cholesterol accumulation in senile plaques of AD patients and mouse models alike [66], suggesting that cholesterol could play a role in the formation of amyloid plaques, and that amyloid impairs cholesterol trafficking and homeostasis. Here, our findings show prominent cholesterol accumulation associated with vascular amyloid in the Tg-FDD. This indicates that the dysregulation of cholesterol redistribution in CAA could play an important role in disease progression.
ApoE, predominantly produced by astrocytes, has been widely confirmed to regulate the redistribution and homeostasis of cholesterol within the brain and to affect the accumulation and clearance of Aβ-amyloid [70, 93, 94]. Here, we observed an association between ApoE and vascular amyloid deposits in Tg-FDD mice, possibly decreasing its solubility. These observations are supported by previous studies implicating ApoE as the principal component of parenchymal amyloid plaques and CAA [95, 96] and the major risk factor for both AD [97] and CAA [98]. The ApoE4 isoform is associated with total and vascular Aβ-amyloid levels [99], and the ApoE2 isoform is a risk factor for hemorrhagic CAA [97, 98]. Conversely, the ApoE2 isoform is protective in AD [100], suggesting that the mechanistic basis of ApoE-amyloid association differs between AD and CAA. The different influence of ApoE on the development of parenchymal plaques versus CAA has also been observed in mouse models. For instance, the over-expression of human ApoE4 in the APPswe mouse model, characterized by the accumulation of amyloid plaques, results in a shift in the amyloid deposition from parenchyma to the vasculature [101]. Also, it was recently shown how in the 5XFAD model expressing murine ApoE and human ApoE4 parenchymal plaques colocalized with much more murine ApoE while vascular amyloid deposits contained more human ApoE4, suggesting that the type of ApoE dictates whether ApoE will lead to greater parenchymal plaques versus CAA [102]. Additional studies will be required to determine if ApoE impairment in the context of vascular amyloid originates from astrocytes or microglia. Pericytes may also be involved as a recent study demonstrated that ApoE plays a critical role in these vascular cells in the context of CAA [103].
The role of TREM2 in AD has become increasingly relevant as human variants of TREM2 have been associated with a 2- to -3-fold increased risk for developing LOAD [104]. Several cellular functions have been ascribed to TREM2, including the regulation of phagocytosis, inhibition of inflammatory signaling, promotion of cell survival, and cholesterol efflux [105, 106]. New studies have demonstrated that several apolipoproteins, including ApoE, are TREM2 ligands, and that the APOE-TREM2 pathway plays a fundamental role in the microglial homeostatic signature [15, 107]. However, the majority of TREM2 studies in AD relate to parenchymal amyloid deposition, and the role of TREM2 in vascular amyloid deposition and CAA is largely unknown [31, 34, 72, 73, 105]. Here, we demonstrated an increase in TREM2 mRNA levels in the Tg-FDD model in comparison with WT controls. Recently, a novel study demonstrated an increase in TREM2 mRNA levels in a rat model for CAA [108], supporting our findings. Interestingly, our biochemical analysis revealed a significant decrease in TREM2 protein levels. It is well known that transcriptome changes do not always correlate with protein abundance [109]; nevertheless, this discrepancy between mRNA and protein levels suggests the presence of either a post-translational mechanism promoting TREM2 degradation, or a post-translational modification directly in the TREM2 protein that decreases its stability. This decrease in protein levels could trigger an increase in TREM2 mRNA expression as a compensatory mechanism. Noteworthily, despite the significant decrease in TREM2 protein levels, no changes in microglial immunoreactivity were observed, suggesting that the decrease of TREM2 could be an early response of microglial reactivity to vascular amyloid deposits. Similar observations were made in an APP/PS1 mouse model characterized by the accumulation of parenchymal amyloid plaques, where the authors reported an increase of TREM2 mRNA and a decrease of TREM2 protein [110]. It was suggested that the lack of TREM2 protein may reduce microglia activity, ultimately leading to neuroinflammation [110]. Collectively, these data indicate that TREM2-related pathways could have an influence on vascular amyloidosis. Further, studies will be necessary to dissect in detail the effect of TREM2 in CAA [104] and the effect that ApoE impairment has on TREM2 via its ligand function [107] or the recently identified TREM2-APOE regulatory pathway responsible for microglial phenotypic change [15].
The decrease in TREM2 protein levels without significant changes in microglial reactivity in early vascular amyloid deposits in the Tg-FDD model is an unexpected and unique finding. However, activation of microglia can result in phenotypic and functional diversity—detrimental or beneficial—depending on the activation conditions. This duality in microglia has been reported in detail, but complex mechanisms that regulate these diverse states of activation remain unclear [111, 112]. It has been suggested that different stages of microglial activation may also depend on signaling from another CNS cell type [113]. Therefore, based on our analysis of the Tg-FDD model, where a preponderant A1 astrocyte reactivity was observed at the early stages of vascular amyloid pathology, we theorized that A1 astrocytes may contribute to microglia activation and decreased TREM2 levels. Interestingly, we observed that treating primary microglia with conditioned A1-ACM media decreases TREM2 intensity without significantly affecting microglia morphology, demonstrating the ability of A1 astrocytes to influence microglial steady state. The concept of A1 astrocytes influencing microglial activation is particularly interesting as studies have mainly focused on the ability of reactive microglia to influence A1 astrocytic phenotypes in disease and injury [19]; however, this concept has not been thoroughly explored in the context of astrocytes influencing microglia reactivity. Only a handful of studies have suggested a direct effect of astrocytes on microglia [114]. For instance, a high level of S100ß production in astrocytes contributes to iNOS and NO production in microglia [115]. In another study, astrocytes were able to downregulate the antigen-presenting function of invading monocytes [116]. Furthermore, a new study demonstrated that exposure to conditioned media from oxygen-glucose-deprived astrocytes activated microglia-promoted neuronal dendritogenesis [113]. Overall, these studies support the notion that in the Tg-FDD model, A1 astrocytes could secrete factor(s) that affect TREM2 protein stability and subsequently microglial activity.
Identifying the signals or factors used by A1 astrocytes to alter TREM2 protein levels in microglia and the role that ApoE could play provides a means to potentially regulate the neuroinflammatory response associated with CAA pathogenesis and in other cerebrovascular diseases where astrocytic reactivity could precede microglial response.
Conclusions
Our study demonstrates that initial glial response associated with early-stage CAA is characterized by the upregulation of A1 astrocytes without significant microglial reactivity. This glial response is distinct from AD, where reactive microglia influence the A1 astrocyte phenotype. Noteworthy, gene expression analysis revealed that several AD risk factors involved in immune response and lipid processing could also have a preponderant role in CAA. This study contributes to the increasing evidence that brain cholesterol metabolism, ApoE, and TREM2 signaling are major players in the pathogenesis of AD-related dementias, including CAA. Understanding the basis for possible differential effects of the glial response, ApoE, and TREM2 signaling on parenchymal plaques versus vascular amyloid deposits provides important insight into neuroimmune mechanisms further understanding of neurodegenerative processes associated with AD and CAA, respectively, and promote the development of novel therapeutics and prevention strategies.