Early transcriptional and cellular abnormalities in choroid plexus of a mouse model of Alzheimer's disease

Abstract

Background: Alzheimer's disease (AD) is a progressive neurodegenerative disorder characterized by the accumulation of amyloid-β plaques, tau hyperphosphorylation, and neuroinflammation. The choroid plexus (ChP), serving as the blood-cerebrospinal fluid-brain barrier, plays essential roles in immune response to stress and brain homeostasis. However, the cellular and molecular contributions of the ChP to AD progression remain inadequately understood.

Methods: To elucidate the molecular abnormalities during the early stages of AD, we acquired single-cell transcription profiling of ChP from APP/PS1 mice with early-stage of Aβ pathology and litter-mate controls. The transcriptional alterations that occurred in each cell type were identified by differentially expressed genes, cell-cell communications and pseudotemporal trajectory analysis. The findings were subsequently validated by a series of in situ and in vitro assays.

Results: We constructed a comprehensive atlas of ChP at single-cell resolution and identified six major cell types and immune subclusters in male mice. The majority of dysregulated genes were found in the epithelial cells of APP/PS1 mice in comparison to wild-type (WT) mice, and most of these genes belonged to down-regulated module involved in mitochondrial respirasome assembly, cilium organization, and barrier integrity. The disruption of the epithelial barrier resulted in the downregulation of macrophage migration inhibitory factor (MIF) secretion in APP/PS1 mice, leading to macrophage activation and increased phagocytosis of Aβ. Concurrently, ligands (e.g., APOE) secreted by macrophages and other ChP cells facilitated the entry of lipids into ependymal cells, leading to lipid accumulation and the activation of microglia in the brain parenchyma in APP/PS1 mice compared to WT controls.

Conclusions: Taken together, these data profiled early transcriptional and cellular abnormalities of ChP within an AD mouse model, providing novel insights of cerebral vasculature into the pathobiology of AD.

Keywords: Alzheimer’s disease; Choroid plexus; Lipid accumulation; Neuroinflammation; Single-cell transcriptome.

Fig. 1

Single-cell transcriptional atlas of the ChP in APP/PS1 mice exhibiting early amyloid pathology. A Co-staining of Aβ with 4G8 (green) and astrocytes with GFAP (red) in the hippocampus and cortex of 4-month-old (n = 5 per group) and 8-month-old (n = 5 per group) APP/PS1 and WT mice, with nuclei stained by DAPI (blue). B Workflow for scRNA-seq of the ChP. Two pooled samples of ChP from the APP/PS1 (n = 8) and WT (n = 8) mice groups were sequenced. C UMAP of cell types identified in ChP of male APP/PS1 and WT mice. D Expression of conventional cell markers in broad cell types. The color label of each column corresponds to each cell type. E Cell markers, clustering, and GO biological processes, with underlined genes representing conventional or reported cell markers. The colors at the top of the clustering and the marker gene sets on the far right correspond to the cell type colors on the left, while the GO enrichment colors in the middle are purely for visual aesthetics and carry no specific meaning. LV, lateral ventricle; 4V, fourth ventricle. F Validation of typical cell type markers by in situ multiplex immunohistochemistry based on TSA method, with nuclei stained by DAPI (blue). Histological validations in (F) were performed on independent age-matched male cohorts (n = 8 per group) processed separately from those used for scRNA-seq, Scale bar, 20 μm. Origin in (A and F), original

Fig. 2

Cellular and transcriptional alterations in ChP of APP/PS1 mice. A Changes in cell ratio of each cell type in male APP/PS1 mice compared to wild-type (WT) mice. B Automatically and mechanically recognized epithelial cells (red box, OTX2-positive cells) and other cells (white box, OTX2-negative cells) based on staining signal of OTX2 (green fluorescence, specifically labeling epithelial cells) and DAPI (blue color, labeling all cell nuclei). Scale bar, 10 μm. C-D OTX2-positive cells (epithelial cells) were measured in the ChP of the lateral (LV, C) and the fourth ventricles (4V, D). Histological validations in (B-D) were performed on independent age-matched male cohorts (n = 8 per group) processed separately from those used for scRNA-seq. E Number and proportion of DEGs in each cell type. F Volcano plot showing distribution of epithelial DEGs. G Pathways enriched in up-regulated and down-regulated genes in epithelial cells. H Volcano plot showing distribution of DEGs in immune cells. Statistical analyses in (A) were performed using Chi-square test; Statistical analyses in (C-D) were performed using student’s t-test, and data were presented as mean ± SD; **, p < 0.01***; p < 0.001; ****, p < 0.0001

Fig. 3

Structural and functional impairments of ChP epithelial cells in APP/PS1 mice. A Co-staining of ARL13B (green, cilium body) and γ-tubulin (red, cilium base) in APP/PS1 and wild-type (WT) mice. B-C Measurement of ARL13B niche numbers in ChP of the lateral (LV, B) and the fourth ventricles (4V, C) in (A). D UQCRB staining (green, a component of the mitochondrial respiratory complex III) of ChP in APP/PS1 and WT mice. E–F Measurement of fluorescence intensity in ChP of the lateral (LV, E) and the fourth ventricles (4V, F) in (D). G-I EZR staining (green, G) and measurement of fluorescence intensity distance from apical membrane (white dotted line) of epithelial cells in ChP of the lateral (LV, H) and the fourth ventricles (4V, I). J Workflow of permeability assessment experiment. K Comparison of dextran level in plasma after intracerebroventricular (ICV) injection in APP/PS1 and WT mice. Cell nuclei were stained by DAPI (blue). Scale bar, 10 μm. Histological validations (A-I; n = 8 per group) and dextran detection assays (J-K; n = 5 per group) were performed on independent age-matched male animals processed separately from those used for scRNA-seq. Statistical analyses in (F-G), (I-J) and (K) were performed using Student’s t-test and data were presented as mean ± SD; statistical analyses in (H-I) were performed using two-way ANOVA test and data were presented as mean ± SEM; *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001

Fig. 4

Down-regulation of epithelial MIF resulted in macrophages activation in APP/PS1 mice. A UMAP of immune subclusters in APP/PS1 and wild-type (WT) mice. B Subcluster markers and categorization of immune cells defined in (A). C Validation of immune cells and myeloid monocytes with co-staining of CORO1A (green, immune cells) and IBA1 (red, monocyte) in WT mice (n = 3). D Comparative analysis of cellular proportions across distinct cell types between APP/PS1 and WT mice. E MIF mediated ligand-receptor subcluster communications. F Comparison of epithelial Mif expression levels between APP/PS1 and WT mice. G MIF (green) in situ expression profiles in APP/PS1 and WT mice. Quantitative analysis of the lateral (LV, H) and the fourth ventricles (4V, I) in (G). J Comparison of immune subcluster interaction strength. K Pseudotime trajectory originating from M2 macrophages to M1 macrophages in UMAP. L Dynamic gene expression in the trajectory in (K). M Comparison of trajectory-associated expression levels of M2 markers in (L) along trajectory in APP/PS1 and WT mice. N Activated macrophages (M1) were validated by CD68 (green, classical marker for activated macrophages) and IBA1 (red) in ChP from APP/PS1 and WT mice. O-P Measurement on ChP in the lateral (LV, O) and the fourth ventricles (4V, P) in (N). Cell nuclei were stained by DAPI (blue). Scale bar, 20 μm. Histological validations (G-I, N-P) were performed on independent age-matched male animals (n = 8 per group) processed separately from those used for scRNA-seq. Statistical analyses in (D) were performed using Chi-square test. Statistical analyses in (H-I) and (O-P) were performed using Student’s t-test, and data were presented as mean ± SD. *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001

Fig. 5

Increased APOE of ChP altered brain lipid homeostasis via interactions with receptors in ependymal cells. A Detection of monocytes (positive for IBA1, red) in CSF-ChP interface in APP/PS1 mice (n = 3). B Monocytes stained positive for IBA1 (red) but not TMEM119 (green) were detected in ependymal cells (white dotted line) in ChP from the APP/PS1 mice (n = 3). C Western blotting of the retained Aβ in culture medium of BMDMs from 4-month-old male WT mice (n = 3). D Quantification of TNF-α in supernatants from BMDMs culture in (C) using ELISA. E Up-regulated (red) and down-regulated ligands (blue) from ChP interact with their receptors in ependymal cells (green). F Co-staining of APOE (green) and LRP1 (red) in brain slice of APP/PS1 and WT mice. G-H Quantification of APOE (G) and LRP1 (H) fluorescence intensity in ChP of APP/PS1 (n = 7) and WT (n = 7) mice in (F). I Measurement of lipoprotein particles in insoluble fraction of cortex and hippocampus tissues from APP/PS1 (n = 5) and WT (n = 5) mice by using ELISA. HDL, high-density lipoprotein; LDL, low-density lipoprotein; VLDL, very low-density lipoprotein. J Working model showing pathological abnormalities in ChP of mice at the early stage of AD pathology. The image was created by Biorender. Cell nuclei in (A, B and F) were stained by DAPI (blue). Histological validations (A-B, F-H) and lipid particles measurement assay (I) were performed on independent age-matched male animals processed separately from those used for scRNA-seq. Scale bar, 20 μm. Data in (D), (G-H), and (I) were presented as mean ± SD, and statistical analyses were performed using Student’s t-test; ns, not significant; *, p < 0.05; **, p < 0.01; ***, p < 0.001