AAV mediated carboxyl terminus of Hsp70 interacting protein overexpression mitigates the cognitive and pathological phenotypes of APP/PS1 mice

Abstract

The E3 ubiquitin ligase, carboxyl terminus of heat shock protein 70 (Hsp70) interacting protein (CHIP), also functions as a co-chaperone and plays a crucial role in the protein quality control system. In this study, we aimed to investigate the neuroprotective effect of overexpressed CHIP on Alzheimer’s disease. We used an adeno-associated virus vector that can cross the blood-brain barrier to mediate CHIP overexpression in APP/PS1 mouse brain. CHIP overexpression significantly ameliorated the performance of APP/PS1 mice in the Morris water maze and nest building tests, reduced amyloid-β plaques, and decreased the expression of both amyloid-β and phosphorylated tau. CHIP also alleviated the concentration of microglia and astrocytes around plaques. In APP/PS1 mice of a younger age, CHIP overexpression promoted an increase in ADAM10 expression and inhibited β-site APP cleaving enzyme 1, insulin degrading enzyme, and neprilysin expression. Levels of HSP70 and HSP40, which have functional relevance to CHIP, were also increased. Single nuclei transcriptome sequencing in the hippocampus of CHIP overexpressed mice showed that the lysosomal pathway and oligodendrocyte-related biological processes were up-regulated, which may also reflect a potential mechanism for the neuroprotective effect of CHIP. Our research shows that CHIP effectively reduces the behavior and pathological manifestations of APP/PS1 mice. Indeed, overexpression of CHIP could be a beneficial approach for the treatment of Alzheimer’s disease.

Introduction

Alzheimer’s disease (AD) is a predominant form of neurodegenerative disease and dementia. It is characterized by a gradual deterioration of cognitive function and behavioral abilities, with the main pathological changes including senile plaques and neurofibrillary tangles (Scheltens et al., 2021; Gonçalves et al., 2023). The pathogenesis of AD involves the misfolding and accumulation of abnormal proteins. Senile plaques are primarily composed of amyloid-β (Aβ), which arises from abnormal cleavage of the amyloid precursor protein (APP; Greenberg et al., 2020; Kelley et al., 2023). Hyperphosphorylated tau forms paired helical filaments that induce neurofibrillary tangles, damaging the cytoskeleton and resulting in neuronal degeneration (Ballatore et al., 2007; Musi et al., 2018). Because of the imbalance between protein synthesis and degradation in AD, the elimination of aberrant proteins may provide protection (Cozachenco et al., 2023). Organisms accomplish protein degradation through two main pathways: the ubiquitin-proteasome system and autophagy-lysosome pathway (Sun-Wang et al., 2020). Abnormal protein degradation via the ubiquitin-proteasome system involves E3 ubiquitin ligase. The carboxyl terminus of heat shock protein (Hsp) 70 interacting protein (CHIP) is encoded by the STIP1 homology and U-Box containing protein 1 (STUB1) gene, and consists of a tetratricopeptide repeat domain, a U-box domain, and a coiled coil domain. The tetratricopeptide repeat domain interacts with molecular chaperones such as HSP70 or HSP90, while the U-box domain exhibits ubiquitin ligase activity. Consequently, CHIP has co-chaperone and E3 ubiquitin ligase functions and plays a key role in protein quality control (Ballinger et al., 1999).

CHIP is also associated with dementia. In our previous study, rats with a CHIP T246M mutation displayed declined cognitive function and increased AD-associated proteins, such as tau and phosphodiesterase 9A (Shi et al., 2018). Both total and phosphorylated tau were significantly increased in CHIP knockout mice (Dickey et al., 2006). Notably, CHIP is decreased in the brain of patients with AD (Ravalin et al., 2019). CHIP interacts directly with APP, and co-localization signals can be detected in the endoplasmic reticulum and Golgi apparatus (Kumar et al., 2007). CHIP also interacts with tau and facilitates its degradation through the ubiquitin-proteasome system (Petrucelli et al., 2004). Intragastric administration of sulforaphane in triple transgenic (3×Tg) AD mice considerably enhanced CHIP expression and improved performance in novel object recognition and scene fear experiments (Lee et al., 2018). Upregulating CHIP via exercise also decreased AD-related neuroinflammation (Xu et al., 2022). Therefore, modulating CHIP could be a promising therapeutic strategy for AD.

Adeno-associated virus (AAV)-mediated gene therapy is a promising therapeutic strategy. Recently, AAV-PHP. B vector has been shown to penetrate the blood-brain barrier and mediate widespread and long-term expression of genes in the nervous system (Deverman et al., 2016). In the present study, we overexpressed CHIP in APPswe/PS1dE9 (APP/PS1) mice (double transgenic mice expressing mouse/human amyloid precursor protein [Mo/HuAPP695swe] and mutant human presenilin 1 [PS1-dE9]), using an AAV vector that crosses the blood-brain barrier, to explore the neuroprotective effects of CHIP in APP/PS1 mice.

Methods

Animals and AAV vectors

APP/PS1 double transgenic mice (Stock no. 14001A, 3 and 6 months old, 26–30 g) and age- and background-matched C57BL/6 (wild-type [WT]) mice (Stock No. 11001A, 26–30 g) were purchased from HFK Bioscience Co., Ltd., Beijing, China (license No. SCXK (Jing) 2019-0008). Male APP/PS1 mice typically have higher cholesterol and triglyceride levels, while female mice tend to exhibit a heavier Aβ burden (Li et al., 2016; Wu et al., 2016). However, gender-related differences were not considered in this study. To minimize the number of experimental animals used, only male mice were used. All the animals were housed in a specific pathogen-free environment with the room temperature controlled at 22 ± 1°C and humidity at (60 ± 5)%. Mice were kept on a 12-hour light/dark cycle with access to water and food ad libitum. The study was approved by the Life Science Ethics Review Committee of Zhengzhou University on 10, August, 2018. All experiments were designed and reported according to Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines (Percie du Sert et al., 2020).

The AAV vector sequence was from Hanbio Biotechnology (Shanghai, China), and was constructed according to Deverman et al. (2016); it was named AAV/BBB. The vector used a CMV promoter. The use of T2A peptide enabled CHIP and green fluorescent protein (GFP) to be separately expressed under the same promoter. AAV at a titer 1.2 × 10¹² vg/mL was injected into the tail vein. In our previous studies, AAV/BBB induced protein expression across the blood-brain barrier in mouse brain without causing any significant adverse effects (Zhang et al., 2020; Hu et al., 2021). APP/PS1 mice were injected with 100 μL AAV/BBB-CHIP-GFP, AAV/BBB-GFP, or phosphate buffer solution (PBS) (#P1020, Solarbio, Beijing, China). A group of WT mice were injected with the same volume of PBS as a control. Subsequent experiments were performed after three months. Six animals were used in each group. For simplicity, the four groups were named as follows: APP/PS1-CHIP, APP/PS1-GFP, APP/PS1-PBS, and WT-PBS. The experimental design is shown in Figure 1.

Figure 1.

Experimental timeline.

AAV: Adeno-associated virus; BBB: blood–brain barrier; CHIP: carboxyl terminus of Hsp70 interacting protein; m: months; GFP: green fluorescent protein; PBS: phosphate buffer solution.

Behavioral tests

Morris water maze

The Morris water maze (MWM) was used to test the learning and memory of mice (D’Hooge and De Deyn, 2001). Three months after injection of AAV or PBS, MWM testing was performed on mice. The test was carried out in a 120 cm diameter pool with graphic marks on the inner wall. Two virtual principal axes were used to divide the maze into four equal quadrants. The water temperature was maintained at 22 ± 1°C, and non-toxic white dye was added to the water for a contrasting effect with the mice. In the place navigation test, a platform (10 cm in diameter) was placed into the middle of one quadrant and hidden underwater (0.5 cm). Each mouse was trained four times daily to find the hidden platform, starting each session in one of the four quadrants. Mice were placed on the platform for 10 seconds before the start of the session. Each training session ended when the mouse found the platform and stayed on it for 2 seconds. The mouse was guided onto the platform if it was not found within 60 seconds. At the end of each training session, the mouse was allowed to stay on the platform for 10 seconds, then dried and put back into their cage. On the fifth day, the latency of the mouse to find the platform was recorded. The spatial probe test was conducted 24 hours after the place navigation test. The hidden platform was removed from the water and the mouse was placed into the apparatus, with the number of crossings over the platform’s original location within 60 seconds counted. Experimental data were recorded and analyzed using an automatic tracking system (Shuo Lin Yuan Technology, Beijing, China).

Nest building test

Three months after tail vein injection of AAV or PBS, the nest building test was performed in each group of mice. On the first day of the experiment, a piece of thick absorbent cotton (5 cm × 5 cm) was placed into a cage for nesting before the start of the night rhythm. The nest was rated the next day. The 5-point scale was as follows: 1 = not noticeably touched, 2 = partially torn, 3 = mostly shredded but no identifiable nest site, 4 = identifiable but flat nest, and 5 = nearly perfect or perfect nest (Deacon, 2006; Yuan et al., 2019).

Immunohistochemistry and immunofluorescence

After behavioral testing, mice were anesthetized by inhalation of isoflurane (3% for induction and 2% to maintenance; Cat# R510-22, RWD Life Science Co., Ltd., Shenzhen, China) and sequentially perfused with PBS and 4% paraformaldehyde (Cat# G1101, Servicebio, Wuhan, China) through the heart. Brains were rapidly removed, soaked in paraformaldehyde overnight for fixation, and then dehydrated in sucrose solution. Coronal frozen sections (thickness of 20 μm) were cut using a cryostat (Lecia Biosystems, Wetzlar, Germany). For immunohistochemistry, a kit was used (Cat# PV-9000, Zsbio, Beijing, China). After sections were restored to room temperature, microwave antigen retrieval was performed using sodium citrate solution (pH 6.0) (Cat# C1010, Solarbio). Hydrogen peroxide was used to block endogenous catalase, then goat serum block was used, and the sections were incubated with the primary antibody at 4°C overnight. 3,3-Diaminobenzidine (DAB) staining was used to visualize the target protein (Cat# ZLI-9017, Zsbio). For immunofluorescence, antigen retrieval was performed in the same way. Sections were permeabilized with 0.3% Triton X-100 (Cat# T8200, Solarbio), blocked with 10% bovine serum albumin, and then incubated with primary antibodies overnight at 4°C. Secondary antibodies were incubated at room temperature for 2 hours in the dark. 4′,6′-Diamidino-2-phenylindole (DAPI) (#S2110, Solarbio) was used to stain nuclei. Details of the primary and secondary antibodies are listed in Additional Table 1. A fluorescence microscope (Lecia Biosystems) was used for image acquisition. Immunohistochemistry or immunofluorescence was performed on brain sections, with the cortex or hippocampus as regions of interest. Six brain sections per mouse were used to calculate averages, which were then statistically analyzed, with three mice per group. ImageJ software (version 1.52o, NIH, Bethesda, MD, USA; Schneider et al., 2012) was used for analysis. Images were converted to 8-bit grayscale and subjected to a binary threshold. Ionized calcium binding adaptor molecule-1 (IBA1) or glial fibrillary acidic protein (GFAP) immunoreactivity around 6E10-labeled Aβ plaques was quantified, with the occupied field of view area calculated and used for statistical analysis. The intensity of myelin basic protein (MBP) in the field of view was also calculated for statistical analysis.

Additional Table 1.

Antibodies and detailed information

Antibody	Host organism	Dilution	Cat#	RRID	Supplier	Application
ADAM10	Rabbit	1:1000	14194	AB_2798420	Cell Signaling Technology, Danvers, MA, USA	WB
APP	Rabbit	1:1000	25524-1-AP	AB_2880118	Proteintech, Wuhan, China	WB
Aβ (clone 6E10)	Mouse	1:1000	803014	AB_2728527	Biolegend, San Diego, CA, USA	IF
BACE1	Rabbit	1:1000	5606	AB_1903900	Cell Signaling Technology	WB
Beclinl	Rabbit	1:50	11306-1-AP	AB_2259061	Proteintech	CE
CTSB	Rabbit	1:50	31718	AB_2687580	Cell Signaling Technology	CE
CHIP	Rabbit	1:200	55430-1-AP	AB_10949225	Proteintech	IF
CHIP	Rabbit	1:1000	2080	AB_2198052	Cell Signaling Technology	WB
GAPDH	Rabbit	1:2000	10494-1-AP	AB_2263076	Proteintech	WB
GAPDH	Rabbit	1:100	5174	AB_10622025	Cell Signaling Technology	CE
GFAP	Rabbit	1:200	12389	AB_2631098	Cell Signaling Technology	IF
GFP	Mouse	1:200	66002-1-Ig	AB_11182611	Proteintech
Goat anti-mouse IgG (H+L), HRP conjugate	Goat	1:10000	SA00001-1	AB_2722565	Proteintech
Goat anti-mouse IgG H&L (Alexa Fluor 488)	Goat	1:500	ab150113	AB_2576208	Abcam, Cambridge, MA, USA
Goat anti-mouse IgG H&L (Alexa Fluor 594)	Goat	1:500	ab150116	AB_2650601	Abcam
Goat anti-rabbit IgG (H+L), HRP conjugate	Goat	1:10000	SA00001-2	AB_2722564	Proteintech
Goat anti-rabbit IgG H&L	Goat	1:500	ab150077	AB_2630356	Abcam
(Alexa Fluor 488) Goat anti-rabbit IgG H&L (Alexa Fluor 594)	Goat	1:500	ab150080	AB_2650602	Abcam
HSP40/DNAJB1	Rabbit	1:1000	13174-1-AP	AB_2230742	Proteintech	WB
HSP70	Rabbit	1:5000	10995-1-AP	AB_2264230	Proteintech	WB
HSP90	Rabbit	1:2000	13171-1-AP	AB_2120924	Proteintech	WB
IBA1	Rabbit	1:200	17198	AB_2820254	Cell Signaling Technology	IF
IDE	Rabbit	1:1000	501630	NA	Zenbio, Chengdu, China	WB
MBP	Rabbit	1:100	78896	AB_2799920	Cell Signaling Technology	IF
NEP	Rabbit	1:1000	382263	NA	Zenbio	WB
Phosphorylated-tau	Rabbit	1:1000	ab32057	AB_778254	Abcam	WB
PLP1	Rabbit	1:50	28702	NA	Cell Signaling Technology	CE
PS1	Rabbit	1:1000	5643	AB_10706356	Cell Signaling Technology	WB
PSAP	Rabbit	1:50	10801-1-AP	AB_2172462	Proteintech	CE
RPS19	Rabbit	1:50	15085-1-AP	AB_2180202	Proteintech	CE
SQSTM1/p62	Rabbit	1:100	18420-1-AP	AB_10694431	Proteintech	CE
tau	Rabbit	1:2000	ab76128	AB_1524475	Abcam	WB
β-tubulin	Rabbit	1:2000	10094-1-AP	AB_2210695	Proteintech	WB

ADAM10: A disintegrin and metalloprotease 10; APP: amyloid precursor protein; Aβ: amyloid-β; BACE1: β-site APP cleaving enzyme 1; CE: Capillary electrophoresis; CHIP: carboxyl terminus of the Hsp70 interacting protein; CTSB: cathepsin B; GAPDH: glyceraldehyde 3-phosphate dehydrogenase; GFAP: glial fibrillary acidic protein; GFP: green fluorescent protein; HRP: horseradish peroxidase; HSP: heat shock protein; IBA1: Ionized calcium binding adaptor molecule-1; IDE: insulin degrading enzyme; IF: immunofluorescence; MBP: myelin basic protein; NA: not available; NEP: neprilysin; PLP1: proteolipid protein 1; PSAP: prosaposin; RPS19: ribosomal protein S19; SQSTM1: sequestosome 1; WB: western blotting.

Western blotting

Immediately after the mice were anesthetized and sacrificed, the brain was removed and the cortex and hippocampus separated on ice. Total protein was extracted using radio immunoprecipitation assay (RIPA) lysis buffer (Cat# R0010, Solarbio) with protease inhibitor (Cat# HY-K0010, MedChemExpress, Monmouth Junction, NJ, USA) and phosphatase inhibitor (Cat# HY-K0021, MedChemExpress). The mixture was fully lysed and then centrifuged at 12,000 × g for 10 minutes at 4°C. Protein quantification was performed using a bicinchoninic acid (BCA) kit (Cat# PC0020, Solarbio), then 5× protein loading buffer (Cat# P0015, Beyotime, Shanghai, China) was added to the supernatant and boiled for protein denaturation. Samples were separated on 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis gels (Cat# P0456, Beyotime), transferred to 0.22 μm polyvinylidene fluoride membranes (Cat# ISEQ00010, Merck Millipore, Billerica, MA, USA), blocked with 5% skimmed milk, and incubated with primary antibody overnight at 4°C. The membranes were probed with primary antibody at 4°C overnight. Detailed information on the antibodies is provided in Additional Table 1. An enhanced chemiluminescence kit was used for signal development. Images were then collected with a GE Amersham Imager 600 System (GE Healthcare, Little Chalfont, UK) and quantitatively analyzed using ImageJ. The expression of CHIP was normalized to β-tubulin. The other proteins were normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH).

Enzyme linked immunosorbent assay

Soluble and insoluble proteins were extracted from the brain or hippocampus using a protein extraction kit (Cat# C500071, Sangon Biotech, Shanghai, China). Detection of Aβ₄₀ and Aβ₄₂ in soluble and insoluble proteins was performed using relevant enzyme linked immunosorbent assay (ELISA) kits (Cat# E-EL-H0542c for Aβ₄₀ and E-EL-H0543c for Aβ₄₂, Elabscience, Wuhan, China) according to the manufacturer’s instructions. The unilateral cortex or hippocampus of three mice per group was used for experiments.

Single nuclei RNA sequencing

The hippocampus of mice injected with 100 μL of AAV/BBB-CHIP-GFP or AAV/BBB-GFP (Hanbio Biotechnology) was used for single nuclei (sn)RNA sequencing; and named as AD_CHIP or AD_GFP, respectively. One side of the hippocampus was removed from three mice in each group and mixed to prepare single cell suspensions for subsequent experiments. The GEXSCOPE Single Cell Nuclear Transcription Library Kit (Cat# 1370062, Singleron Biotechnologies, Nanjing, China) was used for single cell nuclear suspension preparation and library construction. The resulting snRNA-seq libraries were sequenced on an Illumina Novaseq6000 instrument (San Diego, CA, USA) with 150 bp paired end reads. Upstream analysis was performed using CeleScope software (Singleron Biotechnologies) for sequence alignment and format conversion. Analysis of sequencing data was performed by R software (version 4.2.1, https://cran.r-project.org/src/base/R-4/). Output data were processed using the Seurat package (version 4.3.0, https://satijalab.org/seurat). To eliminate low quality cells, only cells with > 200 genes, and < 5000 genes and < 5% mitochondrial RNA were retained. The Harmony package (version 0.1.1, https://portals.broadinstitute.org/harmony/) was used to integrate data. The obtained matrix was standardized, and principal components (PCs) were extracted. The top 20 PCs were chosen for cell clustering. The threshold for clustering was 0.2, which was determined using the clustree package (version 0.5.0, https://lazappi.github.io/clustree/). The resulting clusters were visualized using the UMAP function, and differentially expressed genes (DEGs) were identified by the FindMarkers function, with logfc.threshold = 0.1 set as the differential expression threshold. Enrichment analyses including Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and Gene Set Enrichment Analysis (GSEA) were conducted using the clusterProfiler package (version 4.4.4, https://guangchuangyu.github.io/software/clusterProfiler/). Protein–protein interaction (PPI) networks were generated using the STRINGdb package (version 2.8.4, https://git.bioconductor.org/packages/STRINGdb; and online at https://cn.string-db.org/).

Capillary electrophoresis analysis

Capillary electrophoresis (Jess Simple Western Bio-Technology, San Jose, CA, USA) was used to confirm the results obtained by snRNA-sequencing. The 12–230 kDa Jess Separation Module for chemiluminescence was applied to sample separation and detection. The hippocampus of mice was lysed using RIPA buffer with protease and phosphatase inhibitors, and quantified using a BCA kit (#PC0020, Solarbio). The antibodies used for capillary electrophoresis are provided in Additional Table 1. Capillary electrophoresis was performed according to the default process provided in the manual. Compass for SW software was used (version 6.1.0, Simple Western Bio-Technology) to analyze the results. Three mice per group were used for analysis.

Statistical analysis

No statistical methods were used to predetermine sample sizes; however, our sample sizes are similar to those reported in previous publications (Du et al., 2017; Lopez-Sanchez et al., 2021). GraphPad Prism 8 software (GraphPad Software, San Diego, CA, USA, www.graphpad.com) was used for data analysis, and the mean ± standard error of the mean (SEM) was used to represent the statistical results. One-way analysis of variance followed by Tukey’s post hoc test was used to determine significant differences. A P value < 0.05 was considered to be significant.

Results

Overexpression of CHIP rescues the behavioral impairments of APP/PS1 mice

The MWM and nest building test were used to evaluate the cognitive and social abilities of mice. In the MWM test, learning ability was measured by the time it took mice to find a hidden platform, while memory was measured by the number of times they crossed the location after removing the platform. Animals have a natural tendency to use materials to build nests; the nest building test reflects the social interaction ability of mice. Three months after AAV injection, compared with that in the APP/PS1-GFP or APP/PS1-PBS groups, latency in the place navigation test was significantly shorter in APP/PS1 mice with CHIP overexpression (P < 0.05; Figure 2A and C). Meanwhile, the number of platform crossings in the spatial probe test increased (Figure 2B and D), and the nesting rate also increased (P < 0.05; Figure 2E and Additional Figure 1 ^{(1.2MB, tif)} ). No significant differences in MWM or nest building tests were observed between APP/PS1-GFP and APP/PS1-PBS groups.

Figure 2.

Overexpression of CHIP in APP/PS1 mice rescues behavioral impairments in the Morris water maze and nest building tests.

Behavioral tests were performed three months after AAV or PBS injection. Overexpression of CHIP reduced the latency time of APP/PS1 mice to find the underwater platform (A, C) and increased the number of platform crossings in the spatial probe test (B, D). Mice in the APP/PS1-CHIP group also scored better in the nest building test compared with mice in the other two APP/PS1 groups (E). n = 6 per group. Data are shown as mean ± SEM. *P < 0.05 (one-way analysis of variance followed by Tukey’s post hoc test). APP: Amyloid precursor protein; CHIP: carboxyl terminus of the Hsp70 interacting protein; GFP: green fluorescent protein; PBS: phosphate buffer saline; PS1: presenilin 1; WT: wild-type.

CHIP reduces the burden of Aβ plaques and decreases Aβ and tau expression

We detected the expression of proteins closely related to AD pathology. Aβ plaques are one of the pathological hallmarks of AD, and are observed in brain tissue of 6-month-old APP/PS1 mice. Protein expression in the cortex and hippocampus of each group was detected by western blotting and immunofluorescence, which confirmed that AAV/BBB-CHIP-GFP caused overexpression of CHIP (P < 0.05; Figure 3A and B and Additional Figure 2 ^{(2.5MB, tif)} ). The deposition of Aβ plaques in mouse brain was detected using Aβ antibody (6E10) for immunohistochemistry. No plaques were observed in the brain of WT mice, while the plaque area was reduced significantly in the cerebral cortex and hippocampus of APP/PS1-CHIP mice compared with the findings in APP/PS1-GFP or APP/PS1-PBS mice (P < 0.05; Figure 3C and D).

Figure 3.

Adeno-associated virus–mediated CHIP overexpression in the brain of APP/PS1 mice improves Aβ plaque deposition and reduces Aβ and tau in the cortex and hippocampus.

Three months after injection, CHIP expression was significantly overexpressed in the cortex and hippocampus (A, B). Immunohistochemistry showed that CHIP overexpression reduced Aβ plaque area in the brain of APP/PS1 mice. Human Aβ was not detected in WT-PBS mice. Arrows indicate typical Aβ plaques (C, D). Scale bars: 1 mm. ELISA of soluble and insoluble proteins showed that CHIP overexpression significantly decreased both Aβ₄₀ and Aβ₄₂ in the cortex and hippocampus of APP/PS1 mice (E, F). CHIP overexpression also decreased the expression of APP, tau, and phosphorylated tau (G–I). n = 3 per group for western blotting and ELISA; and n = 6 per group for immunohistochemistry. Data are shown as mean ± SEM. *P < 0.05 (one-way analysis of variance followed by Tukey’s post hoc test). APP: Amyloid precursor protein; Aβ: amyloid-β; CHIP: carboxyl terminus of the Hsp70 interacting protein; GAPDH: glyceraldehyde-3-phosphate dehydrogenase; GFP: green fluorescent protein; PBS: phosphate buffer saline; PS1: presenilin 1; p-tau: phosphorylated tau; WT: wild-type.

Aβ₄₀ and Aβ₄₂ are two common subtypes of amyloid proteins. Aβ₄₂ is more toxic and easier to aggregate, and is believed to be the initiating factor of amyloid plaque formation. Soluble and insoluble proteins were extracted from the mouse cortex and hippocampus. ELISA showed decreased expression of Aβ₄₀ and Aβ₄₂ in soluble and insoluble proteins of CHIP-overexpressed APP/PS1 mice compared with the findings in those that received AAV/BBB-GFP or PBS (P < 0.05; Figure 3E and F). Furthermore, APP was also reduced after CHIP overexpression in APP/PS1 mice. As another marker protein for AD, tau phosphorylation was increased in the brain of APP/PS1 mice, while both tau and phosphorylated tau decreased after CHIP overexpression. All results were compared with the findings in APP/PS1-GFP and APP/PS1-PBS mice (P < 0.05; Figure 3G–I).

CHIP alleviates the microglia and astrocytes concentrated around amyloid plaques

AD is typically accompanied by microglia proliferation. In early AD, microglia become activated and attack Aβ plaques. However, they may play a harmful role as the disease progresses. To detect the distribution of microglia around plaques, we used double-labelled immunofluorescence. Immunofluorescence was performed using IBA1 and Aβ antibodies, and the area percentage of microglial cells around plaques in the visual field was calculated. In APP/PS1 mice injected with AAV/BBB-GFP or PBS, there were more microglia around plaques, but the area occupied by microglia decreased after CHIP overexpression (P < 0.05; Figure 4A and C).

Figure 4.

CHIP reduces microglia and astrocytes around Aβ plaques.

Double-labelled immunofluorescence using Aβ and IBA1 antibodies showed that CHIP overexpression reduced microglia around Aβ plaques in the brains of APP/PS1 mice (A, C). Using the same method for co-labeling with Aβ and GFAP antibodies, astrocytes were decreased around Aβ plaques (B, D). n = 6 per group. For non-magnified images, the scale bar is 50 μm and magnification is 200×; for magnified images, the scale bar is 20 μm and magnification is 630×. Data are shown as mean ± SEM. *P < 0.05 (one-way analysis of variance followed by Tukey’s post hoc test). APP: Amyloid precursor protein; CHIP: carboxyl terminus of the Hsp70 interacting protein; GFAP: glial fibrillary acidic protein; GFP: green fluorescent protein; IBA1: ionized calcium binding adaptor molecule 1; PBS: phosphate buffer saline; PS1: presenilin 1; WT: wild-type.

Reactive astrocytes also participate in neuroinflammation. Therefore, we also used GFAP antibody to detect astrocytes surrounding Aβ plaques. APP/PS1 mice injected with CHIP overexpressing AAV showed reduced astrocytes around plaques in the brain. There was no significant difference between mice injected with the control virus and PBS (P < 0.05; Figure 4B and D).

CHIP affects enzymes that participate in Aβ metabolism

Next, we examined enzymes involved in Aβ production and metabolism. Of proteins related to APP metabolism, β-site APP cleaving enzyme 1 (BACE1) was decreased in the cortex and hippocampus of APP/PS1-CHIP mice compared with the findings in APP/PS1-GFP and APP/PS1-PBS mice. In contrast, there was no significant change in a disintegrin and metalloprotease 10 (ADAM10), PS1, Aβ degradation protein insulin degrading enzyme (IDE), and neprilysin ([NEP] P < 0.05; Figure 5A–C).

Figure 5.

The effect of CHIP overexpression on APP-related metabolic enzymes.

Six-month-old APP/PS1 mice were injected with an AAV vector for three months, then total protein was extracted for western blotting; data are expressed as “9 mon Cortex” and “9 mon Hippocampus”. In both groups, only a decrease in BACE1 was detected after CHIP overexpression (A–C). CHIP was also overexpressed in 3-month-old APP/PS1 mice and the same detection approach performed after three months; data are expressed as “6 mon Cortex” and “6 mon Hippocampus.” In these two groups, ADAM10 expression increased and IDE expression decreased in APP/PS1-CHIP mice, in addition to the decrease of BACE1 after CHIP overexpression. Decreased NEP was detected in the hippocampus (D–F). n = 3 per group. Data are shown as mean ± SEM. *P < 0.05 (one-way analysis of variance followed by Tukey’s post hoc test). APP: Amyloid precursor protein; BACE1: β-site APP cleaving enzyme 1; CHIP: carboxyl terminus of the Hsp70 interacting protein; GAPDH: glyceraldehyde-3-phosphate dehydrogenase; GFP: green fluorescent protein; IDE: insulin-degrading enzyme; NEP: neprilysin; PBS: phosphate buffer saline; PS1: presenilin 1; WT: wild-type.

We then used 3-month-old WT and APP/PS1 mice (which are known to exhibit early changes in AD), and administered either AAV or PBS via injection. After 3 months, total protein from the cortex and hippocampus was extracted for western blotting. Compared with the other groups, APP/PS1 mice receiving AAV/BBB-CHIP-GFP showed a significant increase in CHIP expression (P < 0.05; Additional Figure 3 ^{(1.2MB, tif)} ). Overexpression of CHIP in the cortex and hippocampus resulted in increased ADAM10 and decreased BACE1 and IDE expression, while NEP expression was decreased in the hippocampus only (P < 0.05; Figure 5D–F).

CHIP regulates HSP expression in APP/PS1 mice

HSPs are believed to prevent the aggregation of harmful proteins like Aβ during the disease process. Because CHIP functions as a co-chaperone, we investigated HSPs that are functionally related to CHIP. Compared with the findings in APP/PS1-GFP or APP/PS1-PBS mice, overexpression of CHIP for three months in 6-month-old APP/PS1 mice significantly increased HSP70 and HSP40 in the cortex and hippocampus. In contrast, HSP70 and HSP40 expression was lower in APP/PS1-GFP and APP/PS1-PBS mice compared with that in WT-PBS mice (P < 0.05; Figure 6A–C).

Figure 6.

Changes in heat shock proteins associated with co-chaperone function of CHIP.

Adeno-associated virus was injected into 6- and 3-month-old APP/PS1 mice, and detected after three months. Hsp70 and Hsp40 increased following CHIP overexpression in “9 mon Cortex” and “9 mon Hippocampus” (A–C), and in “6 mon Cortex” and “6 mon Hippocampus” (D–F). No change in HSP90 was observed. n = 3 per group. Data are shown as mean ± SEM. *P < 0.05 (one-way analysis of variance followed by Tukey’s post hoc test). APP: Amyloid precursor protein; CHIP: carboxyl terminus of the Hsp70 interacting protein; GAPDH: glyceraldehyde-3-phosphate dehydrogenase; GFP: green fluorescent protein; HSP: heat shock protein; PBS: phosphate buffer saline; PS1: presenilin 1; WT: wild-type.

In addition, we also examined HSPs in 3-month-old APP/PS1 and WT mice injected with AAV or PBS. Three months after injection, HSP70 and HSP40 increased significantly in the cortex and hippocampus of APP/PS1 mice with overexpression of CHIP compared with the findings in mice injected with the control virus and PBS (P < 0.05; Figure 6D–F). No changes were observed in HSP90.

Bioinformatics analysis suggests putative pathways involved in CHIP overexpression

Three-month-old APP/PS1 mice were injected with AAV/BBB-CHIP-GFP or AAV/BBB-GFP. After 3 months, the hippocampus was isolated for snRNA sequencing; the results are expressed as AD_CHIP and AD_GFP, respectively. After quality control filtration, 48,677 cells were used for subsequent analysis, including 26,713 in the AD_CHIP group and 21,964 in the AD_GFP group. Based upon the expression of different genes, cells were divided into 12 different types annotated as: dentate, CA1, oligodendrocyte, CA2/CA3, subiculum, interneuron, entorhinal, microglia, astrocyte, oligodendrocyte precursor cell, mural, and Cajal-Retzius (Additional Figure 4 ^{(2.7MB, tif)} A). There was no significant difference in the overall distribution of cells between the two groups (Additional Figure 4 ^{(2.7MB, tif)} B). The calculated cell ratios are shown (Additional Figure 4 ^{(2.7MB, tif)} C). The proportion of oligodendrocytes in the AD_CHIP group significantly increased to 10.72%, while the proportion was 7.74% in the AD_GFP group. Marker genes in each cluster were displayed using a heatmap and bubble map (Additional Figure 4 ^{(2.7MB, tif)} D and E). The up-regulated and down-regulated genes in the AD_CHIP group relative to the AD_GFP group were displayed using a volcano map (Additional Figure 4 ^{(2.7MB, tif)} F).

To identify pathways that may change after CHIP overexpression, we performed several enrichment analyses. GO enrichment analysis showed that the top 10 up-regulated biological processes were mainly related to the ensheathment of neurons and axons, the myelin sheath, and oligodendrocytes, while the top 10 down-regulated biological processes were mainly related to ion transport (Figure 7A). For KEGG enrichment analysis, the up-regulation of genes in the lysosomal pathway was the most prominent (Figure 7B). GSEA analysis was performed using the KEGG pathway as the functional gene set. The top 10 activated and suppressed pathways are represented using a bubble plot. The lysosomal pathway was activated, while cholinergic synapses and endogenous cannabinoid receptor pathways were suppressed (Figure 7C–E). The network diagram shows the relationships among pathways for GSEA analysis. Bile secretion and the mitogen-activated protein kinase pathway were more closely related to one another than to other pathways (Figure 7F and G). DEGs between the AD_CHIP and AD_GFP groups were selected for PPI network analysis (Additional Figure 5 ^{(4MB, tif)} A). The network can be divided into three clusters using kmeans clustering. Cluster 1 included 14 nodes and 6 edges, and was mainly about “Neuronal cell body” in functional enrichment analysis. Cluster 2 included 22 nodes and 23 edges, mainly related to the regulation of Aβ and tau. Cluster 3 included 16 nodes and 10 edges, and was associated with axon ensheathment (Additional Figure 5 ^{(4MB, tif)} B–D).

Figure 7.

Enrichment analysis of altered pathways in AD_CHIP mice.

GO analysis shows the top 10 up- and down-regulated biological processes (A). KEGG analysis shows the top 10 up-regulated and only three down-regulated pathways (B). Activated and suppressed pathways in GSEA are shown (C–E). Networks show a relationship among altered pathways in GSEA (F, G). Each group contained a mixture of unilateral hippocampi from three mice. AD: Alzheimer’s disease; APP: amyloid precursor protein; CHIP: carboxyl terminus of the Hsp70 interacting protein; GFP: green fluorescent protein; GO: Gene Ontology; GSEA: Gene Set Enrichment Analysis; KEGG: Kyoto Encyclopedia of Genes and Genomes; PBS: phosphate buffer saline; PS1: presenilin 1; WT: wild-type.

To confirm the bioinformatics analysis results, we used capillary electrophoresis and immunofluorescence to detect protein expression. Sequestosome 1 (SQSTM1/p62) and Beclin 1 are related to the lysosomal pathway, with up-regulated prosaposin (PSAP), cathepsin B (CTSB), proteolipid protein 1 (PLP1), and down-regulated ribosomal protein S19 (RPS19) detected among DEGs. We found increased SQSTM1/p62 in the hippocampus of three groups of APP/PS1 mice, which was decreased after CHIP overexpression compared with the findings in mice injected with AAV-GFP and PBS. Up-regulation of CHIP also increased Beclin 1 expression. The increase of PLP1 and PSAP and decrease of RPS19 found in snRNA-seq were confirmed by protein expression. However in contrast to the sequencing results, protein levels of CTSB were decreased (Figure 8A and C). In each group of mice, we stained oligodendrocytes in the hippocampus with MBP antibody. The results showed that CHIP overexpression increased oligodendrocytes in the hippocampus of APP/PS1 mice, which was consistent with the results of pathway enrichment analysis (Figure 8B and D).

Figure 8.

Verification of bioinformatics analysis.

Protein expression in mouse hippocampus detected by capillary electrophoresis (A, C). In APP/PS1 mice with CHIP overexpression, SQSTM1/p62 and Beclin 1 are consistent with activation of the lysosomal pathway. SnRNA-seq changes in PSAP, PLP1, and RPS19 were also verified, but CTSB was found to be reduced at the protein level. Immunofluorescence showed increased MBP following CHIP overexpression, indicating an increase of oligodendrocytes (B, D). n = 3 per group for capillary electrophoresis; and n = 6 per group for immunofluorescence. Scale bar = 50 μm. Data are shown as mean ± SEM. *P < 0.05 (one-way analysis of variance followed by Tukey’s post hoc test). APP: Amyloid precursor protein; CHIP: carboxyl terminus of the Hsp70 interacting protein; CTSB: cathepsin B; GAPDH: glyceraldehyde-3-phosphate dehydrogenase; GFP: green fluorescent protein; MBP: myelin basic protein; PBS: phosphate buffer saline; PLP1: proteolipid protein 1; PS1: presenilin 1; PSAP: prosaposin; RPS19: ribosomal protein S19; SnRNA-seq: single nuclei ribonucleic acid Sequencing; SQSTM1: sequestosome 1; WT: wild-type.

Discussion

In this study, we used an AAV vector capable of crossing the blood-brain barrier to overexpress CHIP in APP/PS1 mice. CHIP overexpression rescued behavioral deficits, alleviated amyloid plaque formation in the brain, and reduced microglia and astrocytes around plaques. Aβ and phosphorylated tau are closely related to the pathological manifestations of AD, and expression of both were significantly reduced by CHIP overexpression. Moreover, CHIP altered the expression of proteins related to Aβ metabolism (namely BACE1, ADAM10, IDE, and NEP), and increased HSP70 and HSP40 expression. SnRNA transcriptome sequencing showed that CHIP may also activate the lysosomal pathway and play a role in oligodendrocytes.

APP/PS1 mice show cognitive decline and disordered daily behavior. The Morris water maze is commonly used to assess cognitive ability and learning (Cui et al., 2023). Overexpression of CHIP in APP/PS1 mice improved their performance in place navigation and spatial probe tests. Nest building is a common activity among rodents that also serves as a means of social communication. In our experiments, CHIP-overexpressing APP/PS1 mice outperformed both control groups in nest building using cotton pieces. Combined with previous studies, these results suggest that CHIP can rescue the behavioral deficits of AD mice.

According to the amyloid cascade hypothesis, as a downstream molecule of Aβ, tau is involved in its toxic effects, and there is a synergistic effect between Aβ and tau (Viola and Klein, 2015). Aβ₄₀ and Aβ₄₂ are common subtypes of amyloid protein, with Aβ₄₂ being more toxic and susceptible to aggregation (Thorwald et al., 2022). We found that CHIP caused a significant reduction in Aβ, especially insoluble Aβ. This may be due to an improved scavenging effect of CHIP on misfolded proteins. The ability of hyperphosphorylated tau to bind to microtubules decreases, which affects the structural stability of neurons and induces neuronal degeneration. The total amount of tau and phosphorylated tau increases in AD brain. CHIP promotes tau ubiquitination, with overexpression of CHIP effectively promoting tau degradation and reducing its aggregation (Saidi et al., 2015). We found that CHIP reduced tau and phosphorylated tau and alleviated pathological manifestations in APP/PS1 mice.

Microglia are intraparenchymal macrophages. In AD, they migrate to the area around plaques and phagocytize the plaques. Activation of microglia attenuates the pathological changes of early AD. Yet excessive microglial activation leads to the production of toxic substances and neuroinflammation (Hansen et al., 2018). Microglia may migrate to unaffected brain regions as Aβ carriers, promoting the spread of Aβ (d’Errico et al., 2022). Astrocytes maintain the structural integrity of the central nervous system, and also play a regulatory role in the immune system, helping to protect the brain from infection and inflammation (Colombo and Farina, 2016). In AD, astrocytes contribute to the accumulation of Aβ plaques and neurofibrillary tangles (Frost and Li, 2017). Postmortem observations of AD brain reveal the presence of reactive astrocytes in areas with accumulation of Aβ or tau (Kumar et al., 2023). In our study, both microglia and astrocytes around plaques were reduced after CHIP overexpression, which was accompanied by improved Aβ pathology.

When intracellular APP is metabolized via the non-amyloid pathway, α-secretase and γ-secretase act in sequence without producing Aβ. Activation of α-secretase or inhibition of β-secretase reduces Aβ production. Additionally, γ-secretase, which is a direct catalyst for Aβ production, is also a potential target for the treatment of AD (O’Brien and Wong, 2011). The ADAM family is the most extensively studied group of α-secretases. In particular, ADAM10 has been found to be beneficial in reducing Aβ production and regulating tau levels by promoting degradation (Yuan et al., 2017; Musardo et al., 2022). Inhibition of β-secretase is a crucial research direction for the treatment of AD. BACE2 is less widely distributed in the brain, leading researchers to speculate that BACE1 plays a minor role in AD pathogenesis (Bennett et al., 2000). As a result, BACE1 inhibitors have become a focus in the development of drugs to treat AD. CHIP interacts with BACE1 to promote its ubiquitination and degradation. Furthermore, CHIP negatively regulates the BACE1 promoter via p53, thereby reducing BACE1 expression at the transcriptional level (Singh and Pati, 2015; Chanana and Pati, 2018). Because γ-secretase cleaves several transmembrane proteins, inhibitors with more robust selectivity could potentially have a positive effect (Yang et al., 2019). We found that BACE1 was reduced after three months in 6-month-old APP/PS1 mice that received CHIP-overexpressing AAV, while in 3-month-old mice there was also an increase in ADAM10. At both time points, no changes were detected in PS1. This shows that CHIP may play a role by activating the non-amyloid pathway and inhibiting the amyloid pathway.

Aβ-degrading enzymes facilitate the metabolic degradation of Aβ, with IDE and NEP considered to be crucial enzymes in this process. IDE degrades Aβ in soluble protein, preventing its accumulation in deposits and reducing its toxicity (Sahoo et al., 2021). NEP degrades Aβ₄₂, and inhibition of NEP protein activity causes apparent injury and cognitive decline (Grimm et al., 2013). We did not observe any changes in IDE and NEP after AAV injection in 6-month-old APP/PS1 mice. However, in the 3-month-old group, APP/PS1 mice showed increased IDE and NEP expression, which was decreased by overexpression of CHIP. Because IDE and NEP play a role in Aβ elimination, this may explain their increased expression in APP/PS1 mice. Overexpression of CHIP reduced Aβ burden, resulting in lower levels of IDE and NEP in APP/PS1-CHIP mice. Because 3-month-old APP/PS1 mice are in the early stage of AD, early interventions may have a better effect.

HSPs help to fold and assemble other proteins and assist in the repair of damaged proteins. Elevated HSP70 is associated with protection in the brains of AD transgenic mice (Bobkova et al., 2014). HSP70 forms a complex with HSP40 and HSP90 to play a combined role. HSP40 binds to tau and counteracts the formation and proliferation of toxic tau aggregates (Irwin et al., 2021). As an E3 ubiquitin ligase with co-chaperone functions, CHIP simultaneously binds HSP70 and substrate protein. This interaction is crucial for maintaining protein homeostasis. Studies have confirmed that CHIP and HSP70 not only affect the metabolism of Aβ by APP but also regulate the degradation of tau (Kumar et al., 2007). Our findings show that CHIP overexpression leads to increased HSP70 and HSP40. During protein quality control, HSP90 and HSP70 have opposing roles. For AD, abnormal tau has a high affinity for HSP90, and HSP90 inhibitors are thought to reduce tau pathology. However, we did not observe any effect of CHIP on HSP90 in this study.

We used snRNA-seq to investigate the impact of CHIP on transcriptional signatures and cell types in the hippocampus of APP/PS1 mice. Bioinformatics analysis indicated that overexpression of CHIP maintained the health of neurons and activated the lysosomal pathway. A PPI network showed that interacting proteins could be classified into three clusters associated with the neuronal cell body, regulation of Aβ, and tau and axon ensheathment. The lysosome pathway is responsible for the breakdown and recycling of various biomolecules. In individuals with AD, lysosomes may not function efficiently, resulting in the accumulation of Aβ plaques and tau tangles (Lee et al., 2022). SQSTM1/p62 is involved in the formation of autophagosomes and is selectively targeted for encapsulation. Following the fusion of autophagosomes and lysosomes, SQSTM1/p62 is degraded along with the substrate. Moreover, elevated levels of Beclin 1 are indicative of enhanced autophagy (Uddin et al., 2019). PSAP is involved in the lysosomal pathway and has a neuroprotective effect when elevated (Mendsaikhan et al., 2019). Our findings show some contradictory results. CTSB is a lysosomal protease and Ctsb-knockout mice have shown protective effects in various neurological diseases (Hook et al., 2022). However, we found that CTSB was up-regulated by snRNA sequencing, but decreased by protein expression. Another puzzle is that we observed down-regulation of the ribosome pathway and a decrease in RPS19. Previous studies have suggested that RPS19 reduction may promote oxidative stress, but its relationship with CHIP remains unclear (Kang et al., 2021). Therefore, further investigations are needed to investigate the interaction between CHIP and ribosomes.

We found an increased proportion of oligodendrocytes in the AD_CHIP group. Oligodendrocyte-related pathway activation and myelin were also up-regulated on enrichment analysis. Detection of MBP and PLP1 are consistent with these results. Oligodendrocytes produce and maintain the myelin sheath in the central nervous system, which is essential for proper nerve function (Simons and Nave, 2015). Studies have shown a reduced number of oligodendrocytes in AD brains, which is associated with a decrease in myelin content (Nasrabady et al., 2018). Our study suggests that CHIP also plays a neuroprotective role in AD by regulating oligodendrocytes.

Our study has limitations. The phenotype of APP/PS1 mice is caused by overexpression of disease-related proteins, therefore intervention with CHIP may provide a certain degree of compensation. However, as a widely used model, examination of APP/PS1 mice can provide new insights into AD. To summarize, our study provides further evidence for the therapeutic effects of CHIP on AD, with improvements in cognitive impairment and pathological manifestations through various mechanisms. Regulation of CHIP will assist in developing therapeutic strategies for this progressive disease.

Additional files:

Additional Figure 1 ^{(1.2MB, tif)} : Representative photos for nest building experiments of mice in each group.

Additional Figure 1

Representative photos for nest building experiments of mice in each group.

After 3 months of CHIP up-regulation, the nests of APP/PS1 mice were better than those from mice injected with AAV-GFP or PBS. AAV: adeno-associated virus; APP: amyloid precursor protein; CHIP: carboxyl terminus of the Hsp70 interacting protein; GFP: green fluorescent protein; PBS: phosphate buffer saline; PS1: presenilin 1; WT: wild-type.

Additional Figure 2 ^{(2.5MB, tif)} : Immunofluorescence of CHIP and GFP in the brain of APP/PS1 mice.

Additional Figure 2

Immunofluorescence of CHIP and GFP in the brain of APP/PS1 mice.

GFP expression increased significantly in the cerebral cortex and hippocampus of APP/PS1 mice injected with AAV/BBB-CHIP-GFP or AAV/BBB-GFP. CHIP expression increased in mice injected with AAV/BBB-CHIP-GFP. n = 3 per group. Scale bars: 100 µm. Data are shown as mean ± SEM. ^*P < 0.05 (one-way analysis of variance followed by Tukey’s post hoc test). AAV: adeno-associated virus; APP: amyloid precursor protein; BBB: blood-brain barrier; CHIP: carboxyl terminus of the Hsp70 interacting protein; GFP: green fluorescent protein; PBS: phosphate buffer saline; PS1: presenilin 1; WT: wild-type.

Additional Figure 3 ^{(1.2MB, tif)} : Three-month-old APP/PS1 mice were injected with adeno-associated virus or PBS.

Additional Figure 3

Three-month-old APP/PS1 mice were injected with adeno-associated virus or PBS.

CHIP overexpression was detected in the cortex and hippocampus three months later. n = 3 per group. Data are shown as mean ± SEM. ^*P < 0.05 (one-way analysis of variance followed by Tukey’s post hoc test). APP: Amyloid precursor protein; CHIP: carboxyl terminus of the Hsp70 interacting protein; GFP: green fluorescent protein; PBS: phosphate buffer saline; PS1: presenilin 1; WT: wild-type.

Additional Figure 4 ^{(2.7MB, tif)} : SnRNA-seq analysis of the hippocampus in APP/PS1-CHIP and APP/PS1-GFP mice.

Additional Figure 4

SnRNA-seq analysis of the hippocampus in APP/PS1-CHIP and APP/PS1-GFP mice.

AD_CHIP and AD_GFP represent the two groups, respectively. Based upon gene expression, 48,677 cells were divided into 12 clusters, including 26,713 cells in the AD_CHIP group and 21,964 cells in the AD_GFP group (A, B). The stacked bar chart shows the proportion of cells in each group (C). The top 5 DEGs in each cluster are shown in the heatmap. DEGs are labeled in yellow color (D). Marker genes of cell types were selected according to http://dropviz.org/ (E). Up- and down-regulated genes are colored in red in the volcano plot of DEGs (F). AD: Alzheimer’s disease; APP: amyloid precursor protein; CHIP: carboxyl terminus of the Hsp70 interacting protein; DEG: differentially expressed genes; GFP: green fluorescent protein; PBS: phosphate buffer saline; PS1: presenilin 1.

Additional Figure 5 ^{(4MB, tif)} : PPI network among DEGs.

Additional Figure 5

PPI network among DEGs.

A total of 52 proteins were included in the PPI network. Up-regulated genes are labeled red, and down-regulated genes are labeled green. (A). The network was divided into three clusters using kmeans clustering (B–D). Functional enrichment results suggests that cluster 1 is associated with the neuronal cell body, cluster 2 is mainly involved in regulating Aβ and tau, and cluster 3 may be involved in axonal ensheathment. DEG: Differentially expressed genes; PPI: protein-protein interaction.

Additional Table 1: Antibodies and detailed information.

Funding Statement

Funding: This work was supported by the National Natural Science Foundation of China, Nos. 91849115 and U1904207 (to YX), 81974211 and 82171247 (to CS); and Non-profit Central Research Institute Fund of Chinese Academy of Medical Sciences, No. 2020-PT310-01 (to YX).