Downregulating Platelet Endothelial Cell Adhesion Molecule‐1 Enhances Learning and Memory and Alleviates Hallmark Pathologies in Alzheimer's Disease
Qiuzhi Zhou, Fei Sun, Yao Zhang, Xiaojian Cao, Mengzhu Li, Haitao Yu, Tao Jiang, Shihong Li, Weixia Wang, Jiazhao Xie, Ting He, Yanchao Liu, Xiuping Liu, Ying Yang, Dan Ke, Xiao‐Chuan Wang, Enjie Liu, Jian‐Zhi Wang
TL;DR
Reducing CD31 levels in mice with Alzheimer's disease improves memory and reduces key disease features like amyloid plaques and neuroinflammation.
Contribution
CD31 is newly identified as a therapeutic target in Alzheimer's disease, with knockdown shown to reverse multiple disease hallmarks.
Findings
CD31 is upregulated in Alzheimer's brains and transgenic mice, correlating with disease pathology.
Knockdown of CD31 improves cognitive function and reduces amyloid and tau pathologies in 5xFAD mice.
CD31 knockdown suppresses neuroinflammation by modulating STAT1/IRF1 signaling and gene expression.
Abstract
Alzheimer's disease (AD) is a neurodegenerative disorder that currently lacks cures; thus, searching for new biomarkers and unraveling its underlying mechanisms are crucial for devising effective therapies. Here, we discovered that both mRNA and protein levels of CD31 (platelet endothelial cell adhesion molecule‐1, PECAM1), a transmembrane glycoprotein in immunoglobulin superfamily, were significantly higher in the brains of AD individuals and different AD transgenic mice, and the elevated CD31 was related to the recognized AD pathologies. Additional studies demonstrated that systemically knockdown of CD31 in 5xFAD mice significantly improved the cognitive functions with decreased AD hallmark pathologies, including β‐amyloid precipitation and tau hyperphosphorylation. Moreover, CD31 knockdown alleviated neuroinflammation, evidenced by the diminished microglial stimulation and suppressed…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
FIGURE 1
FIGURE 2
FIGURE 3
FIGURE 4
FIGURE 5
FIGURE 6
FIGURE 7- —National Natural Science Foundation of China10.13039/501100001809
- —Natural Science Foundation of Henan Province10.13039/501100006407
- —Science and Technology Committee of China
- —Science and Technology Committee of Hubei Province
- —Guangdong Provincial Key S&T Program
- —Zhejiang Provincial Traditional Chinese Medicine Science and Technology Program
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsExtracellular vesicles in disease · Barrier Structure and Function Studies · Cell Adhesion Molecules Research
Introduction
1
The most prevalent cause of senile dementia is Alzheimer's disease (AD), and its occurrence is growing as the population ages [1]. The global patient population of AD underscores the urgency to address this public health challenge [2]. As there is currently a lack of effective methodologies to mitigate or treat AD, it is important to explore novel biomarkers and to reveal the molecular mechanisms underlying the disease progression.
As a part of the immunoglobulin superfamily, CD31 (also named as platelet endothelial cell adhesion molecule‐1, PECAM1) is a type I transmembrane glycoprotein with molecular weight of 130 kDa. This molecule appears in numerous cell types, including vascular endothelial cells, platelets, and monocytes, etc. Comprised of extracellular, transmembrane, and intracellular domains in structure, CD31 acts as a sensor for changes in the local environment and executes functions in adhesion and signaling transduction [3, 4]. As a member of the immunoglobulin superfamily and a recognized adhesion molecule, CD31 is actively involved in inflammatory responses and the permeability of vascular endothelial cells [5, 6].
Previous studies have indicated that CD31 is implicated in AD. Compared to age‐matched individuals, AD patients presented a notable elevation in CD31 protein levels in both plasma and cerebellar tissue [7, 8]. Transcriptomic analyses also suggested a significant elevation of CD31 mRNA in different brain regions of AD patients [9, 10]. Overexpressing tauP301L mutant in the hippocampal CA1 subset of the wild‐type (WT) mice induced vascular damage accompanied by the overexpression of angiogenesis‐related genes in CD31‐positive endothelial cells [11]. Cerebral arterial embolism caused dot‐like and discontinuous aggregation of CD31 in 3xTg‐AD mice [12]. Injection of amyloid‐β_1‐42_ (Aβ_1‐42_) into the hippocampal CA1 of WT mice increased CD31 expression in hippocampal tissue [13]. ApoE ε4 knock‐in mice showed an increased CD31 phosphorylation by binding monomeric C‐reactive protein (mCRP) to endothelial CD31, leading to cerebral vascular impairment and extravasation of T lymphocytes into the mouse brain [14]. These studies strongly suggest the involvement of CD31 in the vascular dysfunction and immune dysregulation which were commonly seen in AD.
However, most existing findings on CD31 remain predominantly correlative and focus on CD31 expression changes in endothelial cells or peripheral compartments, leaving its functional role within the central nervous system (CNS) unresolved. Critically, it is unknown whether CD31 upregulation contributes to core AD pathologies, including Aβ accumulation, tau hyperphosphorylation, and neuronal loss, by directly impairing neuronal functions such as synaptic plasticity, learning, and memory. These mechanistic gaps underscore the imperative for in‐depth investigation. Herein, we first created a CD31 knockout mouse line by using CRISPR/Cas9 technology. By crossing this CD31 knockout mice with 5xFAD mice, we generated a novel AD mouse line with systemic knockdown of CD31 (i.e., 5xFAD‐CD31(±)). Using this mouse model, we observed that systemic reduction of CD31 could remarkably improve the cognitive functions with attenuation of multiple AD‐like pathologies and deregulated gene expression.
Results
2
CD31 level Is Significantly Higher in the Brains of AD Individuals and Transgenic Mice
2.1
Analysis of brain region‐specific omics data from the Alzdata database (http://www.alzdata.org/) illustrated a remarkable elevation in CD31 mRNA expression in various brain areas of AD individuals compared to age‐matched controls (Figure 1A). To validate these findings, we evaluated CD31 mRNA and protein levels in hippocampal extracts from AD individuals and age‐matched controls using q‐PCR and Western blotting. The findings validated a significant elevation of CD31 at the mRNA (Figure 1D) and protein (Figure 1B,C) levels in AD patients compared to controls.
*CD31 is increased in the brains of AD patients and the transgenic mouse models. (A) The increased CD31 mRNA level in the brain of AD patients analyzed by brain region‐specific omics from the Alzdata database. *p < 0.05, **p < 0.01, ***p < 0.001 vs. Ctrl. (B, C) CD31 protein level in the hippocampus of AD patients was increased compared to age‐matched controls measured by Western blotting. n = 7, **p < 0.01 vs. Ctrl. (D, E) CD31 mRNA level in the hippocampal extracts of AD patients(D) and the 2‐, 4‐, and 6‐month‐old 5xFAD mice(E) was increased measured by quantitative PCR. n = 7–8 (AD patients) or n = 4 (5xFAD mice), ***p < 0.001 vs. Ctrl; or ** p < 0.01, *** p < 0.001 vs. WT. (F‐I) CD31 protein level was increased age‐dependently measured by Western blotting, and the increase was more significant in AD transgenic mice. n = 3, **p < 0.01, or **p < 0.001 vs. WT.
We evaluated CD31 mRNA and protein levels in various age groups of AD transgenic mice, specifically 5xFAD and 3xTg‐AD models. Hippocampal tissues were collected from 5xFAD or 3xTg‐AD and the WT control mice at different ages (2, 4, and 6 months for 5xFAD, or 3xTg‐AD mice at 3, 6, 9, and 12 months), and the mRNA and protein levels of CD31 were detected by qPCR and/or Western blotting. Compared with WT controls, the 5xFAD mice showed an age‐dependent increase in CD31 mRNA level in the hippocampus (Figure 1E). An age‐dependent increase of CD31 protein level was also detected in both WT and AD transgenic mice (Figure 1F–I), and the increase was more prominent in 5xFAD (Figure 1F, G) and 3xTg‐AD (Figure 1H, I) cases compared to age‐matched WT mice.
These data together demonstrate that CD31 is significantly increased in the brains of AD patients and the transgenic mice.
Systemically Downregulating CD31 Improves Learning and Memory in 5xFAD Mice
2.2
To elucidate the function of CD31 in AD pathology and its potential impact on disease progression, we employed CRISPR/Cas9 technology to generate transgenic mice with a global CD31 knockout (CD31‐KO). By further crossing these CD31 knockout (CD31‐KO) mice with 5xFAD mice, we obtained an AD model with systemic reduction of CD31 (termed 5xFAD‐CD31(±), Figure 2A). Hippocampal and cortical tissues from 6‐month‐old 5xFAD, 5xFAD‐CD31(±), and age‐matched WT mice were analyzed for CD31 expression via Western blotting. CD31 levels in the hippocampus and cortical regions of 5xFAD‐CD31(±) mice were significantly reduced to 54% of the control level compared to 5xFAD mice, confirming effective CD31 knockdown in these brain areas (Figure 2B, C).
*Downregulating CD31 attenuates AD‐like learning and memory deficits in 5xFAD mice. (A) Schematics showing CD31 knockdown in 5xFAD mice (5xFAD‐CD31(±)) by crossing 5xFAD mice with CD31‐KO mice. (B, C) The efficient knockdown of CD31 in the hippocampus and cortex of 5xFAD mice was confirmed by Western blotting. n = 4, **p < 0.01, ***p < 0.001 vs. WT; ## p < 0.01, ### p < 0.001 vs. 5xFAD. (D) Behavioral paradigms and their sequential order used in this study. (E, F) CD31 knockdown did not significantly change the traveled distance (E) and time spent in the center area (F) in open field (OF) test. n = 8–10. (G, H) CD31 knockdown restored the discrimination index in novel object recognition (NOR) test, though the restoration in recognition index was not significant. n = 8–10, *p < 0.05, **p < 0.01 vs. WT; # p < 0.05 vs. 5xFAD. (I) CD31 knockdown significantly decreased the latency to find the hidden platform during training phase in Morris water maze (MWM) test. n = 8–10, *p < 0.05, **p < 0.01 vs. WT; # p < 0.05 vs. 5xFAD. (J‐M) During MWM probe trial, mice with CD31 knockdown showed significantly decreased latency to find the target platform area (J), increased the number of target platform crossings (K), and the time spent in the target quadrant (L), with no influence on swimming speed (M). n = 8–10, *p < 0.05, **p < 0.01, ***p < 0.001 vs. WT; # p < 0.05 vs. 5xFAD. (N) CD31 knockdown did not significantly restore freezing time in contextual fear conditioning (FC) test. n = 8–10, *p < 0.05, *p < 0.01 vs. WT.
Cognitive decline is a prominent clinical feature in individuals with AD, and 5xFAD mice exhibit notable cognitive dysfunction as early as 5 months of age [15]. Therefore, we first assessed the impact of CD31 reduction on cognitive functions in WT mice. Cognitive assessments at 6 months revealed no significant changes following CD31 knockdown, as determined by novel object recognition (NOR, Figure S1A,B), Morris water maze (MWM, Figure S1C–G), and fear conditioning (FC, Figure S1H) paradigms. These findings revealed that systemic CD31 knockdown in WT mice did not affect the cognitive abilities of 6‐month‐old mice.
Subsequently, we assessed the influence of CD31 on cognitive functions in 5xFAD mice (Figure 2D). In the open field test, systemic CD31 knockdown did not impact the motor skills or anxiety‐like behavior of 5xFAD mice, as evidenced by similar motor abilities (Figure 2E) and time spent in the center area across all groups (Figure 2F). In the NOR trial, 5xFAD mice exhibited significantly lower recognition (Figure 2G) and discrimination index (Figure 2H) than WT mice. The discrimination index was notably higher in 5xFAD‐CD31(±) mice than in 5xFAD mice (Figure 2H), suggesting that systemic CD31 reduction markedly enhanced the 5xFAD mice's capacity to differentiate new objects.
In the MWM test, 5xFAD mice exhibited a prolonged latency in locating the hidden platform compared to WT mice. However, 5xFAD‐CD31(±) mice demonstrated enhanced learning ability, as evidenced by their reduced time to discover the platform over the 6‐day training period (Figure 2I). During the probe phase, 5xFAD‐CD31(±) mice demonstrated enhanced spatial memory, as demonstrated by low latency to locate the target platform area (Figure 2J), more prevalent platform crossings (Figure 2K), and extended time spent in the target quadrant (Figure 2L), suggesting that systemic CD31 knockdown improved their spatial memory. Swimming speeds were similar across WT, 5xFAD, and 5xFAD‐CD31(±) mice (Figure 2M), indicating that systemic CD31 reduction did not affect motor abilities.
In the FC test, while both 5xFAD and 5xFAD‐CD31(±) mice showed markedly reduced freezing compared to WT mice, the variation between these two groups was not statistically significant. (Figure 2N), suggesting that systemic CD31 reduction did not affect fear memory in the 5xFAD mice.
These results together demonstrate that systemic CD31 reduction does not influence normal cognitive functions but can significantly improve learning and memory abilities in 5xFAD mice.
Systemic Downregulation of CD31 Attenuates Hallmark Brain Pathologies in 5xFAD Mice
2.3
The two hallmark pathologies in the brains of AD individuals are extracellular senile plaques comprised of Aβ and intraneural neurofibrillary tangles formed primarily by hyperphosphorylated tau accumulation [16, 17]. It is currently not known whether the elevated brain CD31 level affects Aβ and tau. To address this, we utilized ELISA to measure Aβ_1‐40_ and Aβ_1‐42_ levels in brain tissue and plasma of 5xFAD and 5xFAD‐CD31(±) mice. Systemic CD31 knockdown in 5xFAD mice led to a marked decrease of Aβ_1‐40_ and Aβ_1‐42_ in the hippocampus and cortex, while plasma levels showed a decrease in Aβ_1‐40_ and no alteration in Aβ_1‐42_ (Figure 3A–D). Thioflavin S staining indicated a reduction in senile plaques in the brains of 5xFAD mice following CD31 knockdown (Figure 3E,F). We also observed that systemic knockdown of CD31 significantly reduced phosphorylated tau at pS396 and AT8 (phosphorylated at Ser202/Thr205) (Figure 3G,H). Immunohistochemistry staining also revealed that, compared to 5xFAD mice, the levels of pS396 and pT205 in the hippocampus of 5xFAD‐CD31(±) mice were significantly decreased (Figure 3I). These data indicate that systemic CD31 knockdown effectively reduces AD‐like Aβ and pTau pathologies in the brains of 5xFAD mice, potentially enhancing cognitive functions.
*Downregulating CD31 reduces Aβ and tau pathologies in 5xFAD mice. (A–D) CD31 knockdown significantly reduced Aβ1‐40 levels in both the hippocampus and cortex (A) as well as in plasma (B), and markedly decreased Aβ1‐42 levels in the hippocampus and cortex (C), while plasma Aβ1‐42 levels remained unchanged (D). n = 4, *p < 0.05, **p < 0.01, ***p < 0.001 vs. 5xFAD. (E, F) CD31 knockdown decreased Aβ plaques in mouse hippocampus measured by Th‐S staining. n = 6–7, **p < 0.01 vs. 5xFAD, bar = 50 µm. (G‐I) CD31 knockdown significantly reduced phosphorylation level of tau measured by Western blot (G, H) and immunohistochemical staining (I). n = 4, *p < 0.05, **p < 0.01, **p < 0.001 vs. WT; # p < 0.05, ## p < 0.01, ### p < 0.001 vs. 5xFAD, bar = 50 µm.
Systemic CD31 Knockdown Restores Synaptic Plasticity in 5xFAD Mice
2.4
To explore the mechanisms underlying the improved cognitive function by systemic CD31 reduction in 5xFAD mice, we evaluated the expression of synapse‐correlated proteins in the hippocampal and cortical tissues of the mice (Figure 4A,B). 5xFAD mice displayed a significant lessening in levels of synapse‐associated proteins compared to WT mice, indicating severe synaptic damage. However, CD31 knockdown mice significantly increased expression of synaptotagmin (SYT), synapsin1 (SYN1), synaptophysin (SYN), and postsynaptic density protein 93 (PSD93) and postsynaptic density protein 95 (PSD95) in the hippocampus and cortex, suggesting that systemic CD31 reduction significantly alleviated synaptic damage in 5xFAD mice. In contrast, CD31 knockdown only slightly altered the expression of glutamate receptor subunits GluA1, GluN2A, and GluN2B compared to the 5xFAD mice. These findings indicate that reducing CD31 may primarily enhance synaptic integrity by preserving structural proteins, rather than directly restoring glutamate receptor levels.
*Downregulating CD31 ameliorates synaptic deficits and improves synaptic plasticity in 5xFAD mice. (A, B) CD31 knockdown significantly restored the levels of multiple synapse‐ or cognition‐associated proteins in the hippocampus and cortex measured by Western blotting. n = 4, *p < 0.05, **p < 0.01, ***p < 0.001 vs. WT; # p < 0.05, ## p < 0.01, ### p < 0.001 vs. 5xFAD. (C, D) Representative transmission electron microscopy images of the CA1 region (C) showing synaptic structures (red arrows), and quantification of synapse number per 100 µm2 (D). CD31 knockdown significantly increased synaptic density compared to 5xFAD mice. n = 5, **p < 0.01, ***p < 0.001 vs. WT; # p < 0.05 vs. 5xFAD, bar = 1 µm. (E‐G) Electrophysiological assessment of synaptic function and plasticity in hippocampal slices. (E) Input‐output curves showing fEPSP amplitude in response to increasing stimulus intensities, indicating impaired synaptic responses in 5xFAD mice, which were partially rescued by CD31 knockdown. n = 5 hippocampal slices from 3–4 mice, ***p < 0.001 vs. WT; ### p < 0.001 vs. 5xFAD. (F) Representative traces and time course of normalized fEPSP slope following high‐frequency stimulation (HFS), demonstrating that CD31 knockdown significantly restored LTP deficits in 5xFAD mice. (G) Quantification of fEPSP slope at 60 min post‐HFS further confirmed enhanced synaptic plasticity in CD31‐deficient 5xFAD mice. n = 8 hippocampal slices from 3–4 mice, ***p < 0.01, **p < 0.001 vs. WT; # p < 0.05 vs. 5xFAD.
Transmission electron microscopy (TEM) demonstrated a notable decrease in synapse density in the CA1 region of 5xFAD mice, which was substantially restored by CD31 knockdown (Figure 4C,D). To evaluate synaptic function, field excitatory postsynaptic potential (fEPSP) recordings were performed. Input‐output curves demonstrated an impaired synaptic responsiveness in 5xFAD mice, which was partially restored by CD31 knockdown (Figure 4E). Moreover, long‐term potentiation (LTP), a main indicator of synaptic plasticity, was notably impaired in 5xFAD mice. However, CD31 knockdown significantly improved LTP, as demonstrated by the increased fEPSP slope post high‐frequency stimulation (HFS) (Figure 4F,G).
These data demonstrate that CD31 is involved in regulating the expression of synaptic proteins, and the improvement of synaptic plasticity by CD31 knockdown at least contributes to the amelioration of cognitive damage in 5xFAD mice.
CD31 knockdown Suppresses Microglial Activation and Neuroinflammation
2.5
Then, we further explored the potential mechanisms underlying the improved synaptic plasticity. CD31 is associated with inflammation, and glial cells are pivotal in AD by facilitating inflammatory responses [18, 19]. Therefore, we investigated the impacts of CD31 on glia cells in the brains of 5xFAD and 5xFAD‐CD31(±) mice. The outcomes illustrated that CD31 knockdown significantly decreased IBA1 (microglial marker) with unchanged GFAP (astrocyte marker) staining (Figure 5A,B), suggesting that CD31 may specifically activate microglia but not astrocytes. Given the known role of activated microglia in secreting cytokines and chemokines to stimulate neuroinflammation [19], we further explored the CD31 effect on this process. In 5xFAD mice, high levels of pro‐inflammatory factors (IL‐1β/6, TNF‐α, and NF‐kB) were observed in the hippocampus and cortex, but systemic CD31 knockdown reduced these elevations (Figure 5C,D). These data reveal the role of CD31 in modulating microglia‐associated neuroinflammation, suggesting that attenuation of inflammation by CD31 knockdown may underlie the improved expression of synaptic proteins.
*CD31 knockdown reduces microgliosis with reduced pro‐inflammatory cytokine level in 5xFAD mice. (A, B) CD31 knockdown reduced microglia (A) with unchanged astrocytes (B) gliosis in the hippocampus measured by co‐immunofluorescence staining. n = 6, *p < 0.05 vs. 5xFAD, bar = 200 µm. (C, D) CD31 knockdown significantly reduced the expression of several pro‐inflammatory cytokines in the hippocampus (C) and cortex (D) of the mice. n = 4, *p < 0.05, **p < 0.01, **p < 0.001 vs. WT, ## p < 0.01, ### p < 0.001 vs. 5xFAD.
CD31 knockdown Alleviates Hippocampal Gene Dysregulation in 5xFAD Mice
2.6
To further evaluate the specific molecular mechanisms by which CD31 participates in modulating cognitive functions and pathological damages in 5xFAD mice, RNA‐seq analysis was conducted on hippocampal tissue from 6‐month‐old male 5xFAD, 5xFAD‐CD31(±), and age‐matched WT mice. A total of 18,316 genes and 1314 differentially expressed genes (DEGs) were observed. In the comparison between the 5xFAD‐WT and Ctrl groups, 1152 genes revealed altered expression, with 882 overexpressed and 270 suppressed (Figure S2A). In the comparison between the 5xFAD‐CD31(±) and WT groups, 288 DEGs were determined, of which 117 were overexpressed and 171 suppressed (Figure S2B). In the comparison between the 5xFAD‐CD31(±) and 5xFAD groups, 687 genes displayed differential expression, with 59 overexpressed and 628 suppressed (Figure S2C).
Further analysis revealed that CD31 knockdown restored the expression of 524 DEGs identified by comparing 5xFAD with the WT groups, which we refer to as “reverted genes”. Among these reverted genes, 33 were upregulated, while 491 were downregulated (Figure 6A). These genes showed clear expression normalization in 5xFAD‐CD31(+/–) mice compared to 5xFAD, as visualized in the heatmap (Figure 6B). A strong negative correlation between the two comparisons further confirmed the reversal effect of CD31 knockdown on gene expression changes (Figure 6C). The overexpressed reverted genes were primarily enriched in synapse‐correlated pathways, such as GABAergic synapse pathway, Glutamatergic synapse pathway, and Serotonergic synapse pathway, as well as lipid metabolism‐related pathways, including the ovarian steroidogenesis pathway and cortisol synthesis and secretion pathway, along with other relevant pathways (Figure 6D). The suppressed reverted genes were primarily enriched in inflammation‐correlated pathways, such as cytokine‐cytokine receptor interaction, chemokine and Toll‐like receptor pathways, as well as immune‐correlated pathways, such as the NOD‐like receptor and B cell receptor pathways, and phagocytic‐related pathways, including phagosome formation, FC gamma R‐mediated phagocytosis, and other relevant pathways (Figure 6E).
*CD31 knockdown modulates the deregulated gene expression with amelioration of cognition‐ and inflammation‐associated clusters in hippocampus of 5xFAD mice. (A) Venn diagram illustrating overlapping and unique DEGs in 5xFAD vs. WT (Comparison 1) and 5xFAD‐CD31(±) vs. 5xFAD (Comparison 2). CD31 knockdown reversed 524 DEGs in 5xFAD mice, with 33 genes upregulated and 491 downregulated. (B) A heatmap displaying the normalized expression z‐scores of the reversed 524 genes (adjusted p value < 0.05, upregulated: red; downregulated: blue), which were significantly dysregulated in the hippocampus of 5xFAD mice and notably improved by systemic CD31 knockdown. (C) Correlation plot of Log2 fold changes (FC) in DEGs between 5xFAD vs. WT and 5xFAD‐CD31(±) vs. 5xFAD. A strong negative correlation (r = ‐0.941, p < 0.0001) suggests a reversal effect of CD31 knockdown on the gene expression changes observed in 5xFAD mice. (D, E) KEGG pathway enrichment analysis of Up‐regulated reverted genes (D) and Down‐regulated reverted genes (E). (F) Analysis of transcription factors regulating DEGs in the 5xFAD vs. WT and 5xFAD‐CD31(±) vs. 5xFAD groups using the TRRUST transcription factor regulatory database. The results identify key transcription factors, including Stat1 and Irf1, as major regulators of the reversed DEGs upon CD31 knockdown. (G) Transcription levels of STAT1 and Irf1 in the hippocampal tissues of mice from 6‐month‐old male WT, 5xFAD, and 5xFAD‐CD31(±) groups. n = 4, p < 0.05 vs. WT; # p < 0.05 vs. 5xFAD.
Analysis using the TRRUST transcription factor regulatory database revealed that the DEGs in both the 5xFAD versu WT and 5xFAD‐CD31(±) versus 5xFAD comparison groups were most significantly regulated by the transcription factors signal transducer and activator of transcription 1 (STAT1) and Irf1 (Figure 6F). Simultaneously, a significant elevation of STAT1 and interferon regulatory factor 1 (Irf1) was detected in 5xFAD mice, whereas systemic CD31 knockdown attenuated the elevation (Figure 6G). These findings suggest that CD31 knockdown may ameliorate gene expression dysregulation in 5xFAD mice by modulating multiple pathways through STAT1 and Irf1.
CD31 knockdown Suppresses STAT1/IRF1 Expression by Reducing Histone Lactylation in 5xFAD Mice
2.7
Transcriptomic analysis showed that CD31 knockdown significantly reduced hippocampal mRNA levels of STAT1 and Irf1 in 5xFAD mice (Figure 6G). The protein levels of STAT1 and IRF1 were markedly elevated in 5xFAD mice, while CD31 knockdown significantly suppressed the elevation measured by Western blotting (Figure 7A,B).
*Downregulating CD31 suppresses STAT1/IRF1 expression by decreasing histone lactylation in 5xFAD mice. (A, B) STAT1 and IRF1 expression was significantly elevated in the hippocampus of 5xFAD mice and it was reversed by CD31 knockdown, as measured by Western blotting. n = 3, *p < 0.05, **p < 0.01 vs. WT; # p < 0.05, ## p < 0.01 vs. 5xFAD. (C, D) Global lactylation level (Pan‐lactylation) was elevated in the hippocampus of 5xFAD mice and it was attenuated by CD31 knockdown. n = 3, *p < 0.05, ***p < 0.001 vs. WT; # p < 0.05 vs. 5xFAD. (E, F) Lactylation levels at H3K14 and H4K12 were significantly increased in the hippocampus of 5xFAD mice and they were markedly reversed by CD31 knockdown, as measured by Western blotting. n = 3, *p < 0.05, ***p < 0.001 vs. WT; # p < 0.05, ### p < 0.001 vs. 5xFAD. (G, H) CD31 knockdown significantly decreased the binding of H3K14la and H4K12la to the promoter regions of STAT1 and IRF1, as measured by the CHIP assay. n = 4, *p < 0.05, *p < 0.01 vs. 5xFAD.
Given that histone lactylation promotes gene transcription [20] and it is elevated in 5xFAD mice [21, 22], we hypothesized that CD31 may regulate STAT1/Irf1 via histone lactylation. We observed that global pan‐lactylation was increased in 5xFAD mice compared to the WT controls, while CD31 knockdown attenuated the global lactylation (Figure 7C,D). Further analyses demonstrated that H3K14 and H4K12 lactylation (H3K14la, H4K12la) were enhanced in 5xFAD mice, whereas CD31 knockdown restored the lactylation of H3K14la, H4K12la (Figure 7E,F and Figure S3). ChIP assays also revealed an enriched H3K14la and H4K12la at STAT1 and Irf1 promoters in 5xFAD mice, whereas CD31 knockdown diminished the enrichment (Figure 7G,H).
These findings suggest that CD31 knockdown suppresses STAT1/IRF1 transcription by reducing histone lactylation (H3K14la, H4K12la), by which it modulates the disease‐related transcriptional activity in 5xFAD mice.
Discussion
3
CD31, encoded by chromosome 17, is part of the immunoglobulin superfamily of cell adhesion molecules. Composed of extracellular, transmembrane, and intracellular segments, CD31 functions both as an adhesion molecule and a signaling molecule in the body [4, 23]. CD31 is abundantly distributed at endothelial cell junctions, thus it has been commonly used as a marker in the research of vascular diseases [24]. Bioinformatic analyses indicate an increased CD31 expression in the brain tissues of both AD patients and the animal models [7, 8, 9, 10, 11, 12]. This study biochemically verified that CD31 mRNA and protein levels were significantly increased in the hippocampal tissues of AD patients. In AD transgenic mice, such as 5xFAD and 3xTg‐AD, the hippocampal level of CD31 was also significantly higher than that of the controls, and the increase exhibited an age‐dependent manner. These findings indicate that increased CD31 levels can significantly influence AD onset and progression. Therefore, downregulating CD31 may halt AD‐like pathological damage and cognitive decline.
Growing evidence illustrates that neuroinflammation has a crucial function in driving AD. In the AD brain, overactivation of astrocytes and microglia leads to oxidative stress and high production of pro‐inflammatory mediators [25]. Moreover, neuroinflammation induced by immune cells in the AD brain is positively related to the progression and severity of the disease [26, 27, 28]. Excessive stimulation of microglia results in increased generation of IL‐1β/6 and TNF, which are related to alterations in mood, memory, cognition, and neuronal death [27, 29, 30]. Our study found that systemic knockdown of CD31 significantly reduced microglial activation with decreased pro‐inflammatory cytokines in 5xFAD mice. Transcriptomic analysis further revealed that CD31 knockdown downregulated pro‐inflammation‐related genes. These observations are in agreement with the pro‐inflammatory role of CD31 in peripheral diseases, such as atherosclerosis and acute lung injury [3, 6]. It is suggested that CD31 may regulate microglial activation and the related inflammatory responses in AD, making it a promising therapeutic target at least for delaying the progression of neurodegeneration.
Cognitive impairment is a major clinical manifestation of AD [31]. Synaptic loss is central to AD‐related cognitive dysfunction, as synaptic integrity is crucial for maintaining cognitive function [31]. Inhibiting Aβ accumulation can improve synaptic transmission and restore cognitive function in AD models [32, 33]. Transcriptomic research indicates that alterations in synaptic transmission are linked to the dysregulation of genes associated with AD, with synaptic dysfunction playing a crucial role in AD pathology progression [34, 35, 36]. We found in the present study that CD31 knockdown significantly improved cognitive performance in 5xFAD mice. Simultaneously, CD31 knockdown remarkably restored the expression of synapse‐associated proteins, such as PSD95, a crucial scaffolding protein essential for maintaining synaptic plasticity. In addition, electron microscopy revealed a restoration of synapse density in the hippocampal CA1 region, indicating an improvement in structural plasticity. Meanwhile, the results of LTP further demonstrated the recovery of functional plasticity, as evidenced by the alleviation of synaptic impairments. These outcomes also illustrate that CD31 has a function in cognitive functions by modulating the expression of synapse‐related proteins and synaptic plasticity.
Plaques generated from extracellular Aβ deposits and neurofibrillary tangles created by intracellular hyperphosphorylated tau protein accumulation are hallmark pathologies in the AD brains [37]. Our study revealed an important relationship between CD31 and these AD pathological features. In 5xFAD mice, systemic CD31 knockdown not only significantly reduced soluble and deposited Aβ but also significantly inhibited tau hyperphosphorylation. These findings indicate that CD31 is crucial in Aβ and Tau pathologies. However, given that tau pathology is relatively limited in the 5xFAD model and likely arises as a secondary consequence of Aβ‐induced neurotoxicity, it remains unclear whether the observed reduction in tau hyperphosphorylation stems from a direct effect of CD31 knockdown or an indirect outcome of the decreased Aβ burden. Based on current evidence, we propose that the attenuation of tau pathology may be primarily driven by the reduced Aβ accumulation. Further studies using tau‐specific models will be essential to determine the direct function of CD31 in tau pathogenesis.
Notably, the dynamic interaction between Aβ/tau pathologies and neuroinflammation may contribute to a self‐perpetuating vicious cycle. The stimulation of microglia by Aβ deposition results in the secretion of pro‐inflammatory cytokines, which subsequently induce tau hyperphosphorylation [19]. Neuroinflammation exacerbates Aβ production and deposition, while also enhancing tau phosphorylation and aggregation via inflammatory responses, ultimately causing neuronal dysfunction and death [38, 39]. This interplay between Aβ, tau, and neuroinflammation reinforces each other, accelerating the progression of AD pathology and cognitive decline [19]. Our results showed that CD31 knockdown effectively disrupted this vicious cycle by significantly reducing Aβ production, decreasing pathological tau phosphorylation, alleviating inflammation, and improving synaptic damage. These outcomes illustrate that CD31 knockdown has a critical function in controlling neuroinflammation and breaking the reciprocal promotion among Aβ, tau, and inflammation, offering a new therapeutic target for slowing the progression of AD pathology.
Our further transcriptomic analysis illustrated that CD31 knockdown reversed many DEGs primarily regulated by STAT1 and IRF1. In 5xFAD mice, STAT1 and IRF1 transcription levels were significantly elevated, but CD31 knockdown suppressed this increase, suggesting a potential mechanism through which CD31 regulates neuroinflammation and improves AD pathology. This aligns with earlier research on STAT1 and IRF1, crucial transcription factors in immune responses and inflammation, closely linked to AD pathology. In AD, heightened activation of STAT1 in microglia triggers the secretion of pro‐inflammatory cytokines, intensifying neuroinflammation [40]. The activated STAT1 also upregulates BACE1 expression, promoting Aβ production [41]. Our previous research showed that the activation of STAT1 can directly downregulate key synapse‐related proteins and directly impair synaptic function in mice [42]. In AD brain tissues, IRF1 levels are elevated and may serve as a potential diagnostic biomarker for AD [43, 44]. IRF1 silencing significantly reduces microglial phagocytosis of Aβ and the production of pro‐inflammatory cytokines post‐LPS stimulation [45]. Given that the roles of the STAT1 and IRF1 pathways in AD have been well studied, we instead explored how changes in CD31 regulate STAT1 and IRF1 as upstream factors.
To further clarify how CD31 regulates STAT1/IRF1 expression, we investigated the potential role of histone lactylation—an emerging epigenetic modification recently linked to the transcriptional regulation of immune and inflammatory genes [20]. Elevated histone lactylation has been observed in AD models, where it is thought to drive AD‐related transcriptional dysregulation and neuroinflammation [21, 22]. Consistent with previous reports, our findings demonstrated that CD31 knockdown markedly decreased global pan‐lactylation and site‐specific histone lactylation at H3K14 and H4K12, both of which are inked to active gene transcription. Notably, CD31 deficiency also decreased the enrichment of H3K14la and H4K12la at the STAT1 and IRF1 promoters, pointing to a direct transcriptional regulatory mechanism. While histone lactylation has been associated with microglial activation and neuroinflammation, our study is the first to connect CD31 to this epigenetic pathway in AD. This finding provides new mechanistic insights into how CD31 acts as an upstream regulator of STAT1/IRF1 signaling by modulating histone lactylation. Future studies should explore how CD31 signaling interacts with lactate metabolism and histone‐modifying enzymes, and whether this axis can be targeted therapeutically.
Although CD31 upregulation was observed in both 5xFAD and 3xTg‐AD mice, we selected the 5xFAD model for mechanistic studies due to its fast‐progressing, Aβ‐dominant pathology and early onset—features that facilitate experimental control. In contrast, 3xTg‐AD mice display more complex, slower‐developing pathological profile involving both Aβ and tau, which may mask target‐specific effects. Nevertheless, given the marked neuroinflammation in 3xTg‐AD mice, future studies are warranted to evaluate whether CD31 knockdown elicits similar benefits in this model, particularly with respect to tau pathology.
In summary, we first verified that CD31 was significantly higher in the brains of AD individuals, and the elevated CD31 played a detrimental role in AD. Systemic CD31 knockdown attenuated the AD‐related pathologies, including accumulation of Aβ and pTau, activation of microglia, elevation of pro‐inflammatory cytokines, and reduction of synaptic proteins, with a remarkably improved cognitive function in 5xFAD mice. Furthermore, CD31 knockdown ameliorated dysregulated gene expression by inhibiting STAT1 and IRF1 signaling, a process mediated through the modulation of histone lactylation at their promoter regions. Therefore, CD31 may act as a promising therapeutic target for novel drug development in AD.
Materials and Methods
4
Reagents and Human Brain Tissue
4.1
Table S1 summarizes the antibodies used. Postmortem human brain tissues were sourced from the National Human Brain Bank for Development and Function (Chinese Academy of Medical Sciences, Peking Union Medical College, Beijing, China). Donor information is detailed in Table S2
Animals
4.2
C57BL/6J and 5×FAD mice were obtained from the Experimental Animal Central (Beijing Vital River Laboratory Animal Technology Co., Ltd.), and CD31 knockout mice were generated by Cyagen Biosciences. Animals were housed under standard conditions (12‐h light/dark cycle, 24 ± 2°C) with ad libitum food and water. All procedures were approved by the Animal Care and Use Committee of Huazhong University of Science and Technology.
Behavioral Tests
4.3
In the NOR test, mice were acclimated to an empty chamber for 5 min, 24 h before testing. On the 1st day, participants explored two similar objects, labeled A and B, for a duration of 10 min. On Day 2, a novel object (C) replaced a familiar object (B), and mice explored the chamber for 10 min. The time spent exploring object A (TA) and object C (TC) was recorded. The chamber was sanitized with 75% ethanol between trials to eliminate olfactory cues. The recognition index (RI) is defined as the ratio of TC to the sum of TA and TC, expressed as RI = TC/(TA + TC). Similarly, the discrimination index (DI) is measured by subtracting TA from TC and dividing by the sum of TA and TC, expressed as DI = (TC − TA)/(TA + TC).
In the MWM test, the pool was sectioned into four equivalent quadrants, with a concealed platform located in 1 quadrant. In the training phase, mice were placed in the pool to observe the platform within 60 s. Upon locating the platform, the escape latency was documented, and the mouse stayed on the platform for 30 s. If unsuccessful, the mouse was gently guided to the platform, the latency was set to 60 s, and it stayed there for the same duration. Each mouse completed three trials per day over 6 consecutive days. Following a 1‐day hiatus, a probe trial was performed without the platform, permitting mice to swim freely for 60 s. An automated tracking system (TaiMeng, Chengdu) recorded parameters (escape latency, swim path, swimming speed, and platform crossings).
During the FC test, mice were subjected to three 0.9 mA electric foot shocks, each lasting 3 s, within a 3‐min timeframe on Day 1 in the test chamber. On the 2nd day, mice were positioned back in the same chamber for 3 min without receiving a shock, and their freezing time was measured.
Western Blotting
4.4
The collection of brain tissues was conducted on ice, and the homogenization was performed in a lysis buffer composed of 50 mM Tris‐HCl, 100 mM NaCl, 1% Triton X‐100, and 5 mM EDTA, with added PMSF and a protease inhibitor cocktail. Homogenates were spun at 12,000 rpm for 30 min, and supernatants were combined with SDS loading buffer, and boiling was conducted for 10 min. Protein separation was conducted by SDS‐PAGE, and then transferred to nitrocellulose membranes, blocked with 5% nonfat milk in TBS for 1 h at room temperature (RT), and an overnight incubation was conducted with primary antibodies at 4°C. After rinsing with TBST, a 1‐h incubation of membranes was conducted with HRP‐conjugated secondary antibodies at RT. An ECL detection system (QinXiang, Shanghai) was utilized to visualize the protein bands.
Immunostaining
4.5
A 4% paraformaldehyde in phosphate buffer was utilized to fix the brain tissues at 4°C for 48 h, followed by cryoprotection in 20% and 30% sucrose solutions. Coronal brain sections (40 µm thick) were obtained using a cryostat microtome (Leica, USA).
Immunofluorescence staining involved permeabilizing and blocking free‐floating sections in PBS with 5% bovine serum albumin and 0.5% Triton X‐100 at RT for 1 h. Sections were incubated overnight at 4°C with primary antibodies, then with Alexa Fluor‐conjugated secondary antibodies for 1 h at 37°C. Sections were mounted in 50% glycerol in PBS following PBS washing. Photographs were captured with a Zeiss LSM800 two‐photon confocal laser‐scanning microscope.
For immunohistochemistry, sections were initially subjected to 0.3% hydrogen peroxide in PBS for 30 min, then blocking and permeabilization were conducted. The primary antibody was incubated overnight at 4°C, then a 1‐h incubation was conducted with an HRP‐conjugated secondary antibody at 37°C. Visualization utilized a DAB substrate kit (ZSGB‐Bio, ZLI‐9018), followed by dehydrating in graded ethanol, xylene clearing, and mounting with neutral balsam. Imaging was conducted via a virtual slide scanner (SV200, Olympus).
Thioflavin S Staining
4.6
Brain sections were treated with PBS and 0.5% Triton X‐100 for 30 min at RT to enhance permeability, then a 5‐min incubation was conducted in 0.1% Thioflavin S (dissolved in 50% ethanol) in the dark at RT. After rinsing with 50% ethanol and PBS, counterstaining of the sections was conducted with DAPI, and mounting was conducted in PBS with 50% glycerol.
Detection of Aβ
4.7
Hippocampal and cortical tissues were dissected on ice, homogenized in chilled PBS with a protease inhibitor cocktail, and spun at 5000 rpm for 30 min at 4°C. The resulting supernatant was collected for analysis. For blood processing, whole blood samples were kept at 4°C for 5 h, then a 1‐min centrifugation was conducted at 2000 rpm at 4°C, after which plasma was harvested. Aβ_1‐40_ and Aβ_1‐42_ levels were assessed with commercial ELISA kits (Elabscience, E‐EL‐H0542 and E‐EL‐H0543) as per the manufacturer's guidelines.
Transcriptomics
4.8
Hippocampal tissues were obtained from 6‐month‐old male 5xFAD, 5xFAD‐CD31(±), and WT C57 mice (n = 3 per group). Total RNA isolation was conducted via the Trizol method, followed by mRNA enrichment with Oligo(dT)‐coated magnetic beads targeting polyA tails. Then, the enriched mRNA fragmentation was conducted, and first‐strand cDNA synthesis was conducted via random hexamer primers. Then, the second strand was synthesized to produce double‐stranded cDNA.
The cDNA was subjected to end repair, 5′phosphorylation, and 3′A‐tailing. cDNA fragments were ligated to T‐overhang adaptors and amplified by PCR with specific primers. The denaturation of products was conducted into single strands, and then circularized using a bridging primer to generate single‐stranded circular DNA libraries. Libraries passing quality assessment were sequenced on high‐throughput platforms.
Sequencing data were normalized with the Perseus software. DEGs were determined via t‐tests, with significance considered as p < 0.05 relative to controls. Heatmaps and hierarchical clustering were created via the heatmap.2 package in R, and volcano plots were created with GraphPad Prism 8.0. Gene ontology (GO), pathway enrichment, and protein‐protein interaction (PPI) network analyses were conducted using Metascape (http://metascape.org).
Electrophysiological Recordings
4.9
Electrophysiological measurements were conducted on acute hippocampal slices prepared from deeply anesthetized mice. Once full anesthesia was confirmed, brains were quickly removed and immersed in ice‐cold oxygenated cutting solution composed of 225 mM sucrose, 3 mM KCl, 1.25 mM NaH_2_PO_4_, 24 mM NaHCO_3_, 6 mM MgSO_4_, 0.5 mM CaCl_2_, and 10 mM D‐glucose. Coronal sections, 300 µm thick, were prepared with a Leica VT1000S vibratome at 4°C and incubated in oxygenated cutting solution at 30°C for 1 h prior to recording. For LTP recordings, slices were placed in a recording chamber with a planar 8 × 8 microelectrode array (50 × 50 µm electrodes, 150 µm spacing) and a continuous perfusion was conducted with artificial cerebrospinal fluid (aCSF) at 1–2 mL/min. Slices were secured with a nylon mesh and platinum anchor. fEPSPs were triggered by Schaffer collateral stimulation and recorded with the MED64 multichannel system (Alpha MED Sciences, Panasonic). LTP was induced using three HFS trains, each at 100 Hz for 1 s. The degree of LTP was quantified by averaging normalized fEPSP amplitudes 60 min after induction.
Transmission Electron Microscopy Analysis
4.10
Fresh brain tissues were quickly dissected and immediately placed into fixative to minimize mechanical artifacts. The tissues were trimmed into ∼1 mm^3^ blocks, fixed at 4°C, and washed with 0.1 M phosphate buffer (pH 7.4), then post‐fixation was performed in 1% osmium tetroxide at RT for 2 h. Following thorough buffer rinses, samples underwent dehydration via a graded ethanol and acetone series before being infiltrated and embedded in 812 resin. The polymerization process was conducted at 60°C over a period of 48 h. Ultrathin sections (70 nm) were prepared via a Leica UC7 ultramicrotome and followed by mounting onto 200‐mesh formvar‐coated copper grids. Uranyl acetate and lead citrate were utilized for sequential staining of the sections before being tested using a HITACHI HT7800 transmission electron microscope.
CHIP Analysis
4.11
CHIP analysis utilized the BeyoChIP ChIP Assay Kit (P2080, Beyotime) as per the manufacturer's guidelines. Hippocampal tissues were crosslinked using 1% formaldehyde, quenched with glycine, and lysed in SDS lysis buffer with protease inhibitors. Sonication was utilized to shear chromatin, producing DNA fragments ranging from 200 to 1000 base pairs. An overnight incubation of the fragmented chromatin was conducted at 4°C with Protein A/G magnetic beads and antibodies anti‐H3K14la (PTM‐1414RM, PTM BIO) and anti‐H4K12la (PTM‐1411RM, PTM BIO). The immunoprecipitated complexes underwent sequential washing, followed by elution, reverse‐crosslinking, and purification of the bound DNA for subsequent PCR analysis.
Statistical Analysis
4.12
Data are presented as mean ± SEM. ImageJ and GraphPad Prism were used for analysis and visualization. Student's t‐test was applied for two‐group comparisons, and one‐ or two‐way ANOVA for multiple groups. Significance was defined as p < 0.05.
Author Contributions
Qiuzhi Zhou, Jian‐Zhi Wang, and Enjie Liu: research design. Qiuzhi Zhou, Fei Sun, Yao Zhang, Xiaojian Cao, Mengzhu Li, Haitao Yu, Tao Jiang, Shihong Li, Weixia Wang, Jiazhao Xie, Ting He, Yanchao Liu, Dan Ke, Xiuping Liu, and Ying Yang: performing experiments. Enjie Liu, Qiuzhi Zhou, Ying Yang, Xiuping Liu, Xiao‐Chuan Wang, and Jian‐Zhi Wang: data analysis. Jian‐Zhi Wang and Enjie Liu: data verification. Jian‐Zhi Wang, Qiuzhi Zhou, and Enjie Liu: writing the manuscript. All authors reviewed and gave their approval to the final manuscript.
Funding
This investigation was funded from National Natural Science Foundation of China (82230041, 91949205, 82301620, 82001134, 31730035, 81721005), the Natural Science Foundation of Henan Province (242300421081), Science and Technology Committee of China (2016YFC1305800), Science and Technology Committee of Hubei Province (2018ACA142), Guangdong Provincial Key S&T Program (018B030336001), Zhejiang Provincial Traditional Chinese Medicine Science and Technology Program (2023ZL109).
Ethical Approval
Animal experiments were approved by the Institutional Animal Care and Use Committee of Huazhong University of Science and Technology ([2022] IACUC 2751). Human postmortem brain tissues were obtained from the National Human Brain Bank with approval from the Ethics Committee of Huazhong University of Science and Technology ([2022] IEC(A228)); informed consent was obtained from all donors or their legal representatives.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Figure S1: (A) Recognition index during the novel object recognition test. (B) Discrimination index during the novel object recognition test. (C) Latency to find the hidden platform during the training phase of the Morris water maze. (D) Latency to locate the target platform area during the probe trial of the Morris water maze. (E) Number of times mice crossed the target platform area during the probe trial of the Morris water maze. (F) Time spent in the target quadrant where the platform was located during the probe trial of the Morris water maze. (G) Average swimming speed of mice in the water during the probe trial of the Morris water maze. (H) Percentage of freezing time during the contextual fear conditioning test. n = 8‐10, Data were presented as mean ± SEM. Figure S2: Differential gene expression among WT, 5xFAD and 5xFAD‐CD31(+/‐) mice.The volcano plots illustrate significant transcriptional alterations and the distribution of differentially expressed genes across experimental groups. (A) Volcano plot of 5xFAD mice vs WT group. A total of 882 genes were upregulated (red) and 270 genes were downregulated (blue). (B) Volcano plot of 5xFAD‐CD31(+/‐) mice vs WT group. A total of 117 genes were upregulated and 171 genes were downregulated. (C) Volcano plot of 5xFAD‐CD31(+/‐) mice vs 5xFAD mice. A total of 59 genes were upregulated and 628 genes were downregulated. The x‐axis represents Log_2_ fold change (FC), and the y‐axis represents ‐Log₁₀p value. Dashed lines indicate the significance threshold (p < 0.05), and vertical lines represent Log_2_FC thresholds. Red dots represent significantly upregulated genes, blue dots represent significantly downregulated genes, and gray dots represent non‐significant genes. Figure S3: CD31 knockdown attenuates elevated histone lactylation levels in the hippocampus of 5xFAD mice. (A, B) Immunofluorescence staining shows increased lactylation of histone H3K14la (A) and H4K12la (B) in the hippocampus of 5xFAD mice, which is markedly reduced following CD31 knockdown, bar = 50 µm. Table S1: Antibodies employed in this study. Table S2: Information for human brain samples.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1C. Oboudiyat , H. Glazer , A. Seifan , et al., “Alzheimer's Disease,” Seminars in Neurology 33, no. 4 (2013): 313–329.24234352 10.1055/s-0033-1359319 · doi ↗ · pubmed ↗
- 2P. Scheltens , B. De Strooper , M. Kivipelto , et al., “Alzheimer's Disease,” Lancet 397, no. 10284 (2021): 1577–1590.33667416 10.1016/S 0140-6736(20)32205-4PMC 8354300 · doi ↗ · pubmed ↗
- 3G. Caligiuri , “CD 31 as a Therapeutic Target in Atherosclerosis,” Circulation Research 126, no. 9 (2020): 1178–1189.32324506 10.1161/CIRCRESAHA.120.315935 · doi ↗ · pubmed ↗
- 4P. Lertkiatmongkol , D. Liao , H. Mei , et al., “Endothelial Functions of Platelet/Endothelial Cell Adhesion Molecule‐1 (CD 31),” Current Opinion in Hematology 23, no. 3 (2016): 253–259.27055047 10.1097/MOH.0000000000000239 PMC 4986701 · doi ↗ · pubmed ↗
- 5G. Morris , B. K. Puri , L. Olive , et al., “Endothelial Dysfunction in Neuroprogressive Disorders‐Causes and Suggested Treatments,” BMC Medicine 18, no. 1 (2020): 305.33070778 10.1186/s 12916-020-01749-w PMC 7570030 · doi ↗ · pubmed ↗
- 6Y. Zhao , H. Jin , K. Lei , et al., “Oridonin Inhibits Inflammation of Epithelial Cells via Dual‐Targeting of CD 31 Keap 1 to Ameliorate Acute Lung Injury,” Frontiers in Immunology 14 (2023): 1163397.37090710 10.3389/fimmu.2023.1163397 PMC 10116055 · doi ↗ · pubmed ↗
- 7H. M. Nielsen , E. Londos , L. Minthon , et al., “Soluble Adhesion Molecules and Angiotensin‐Converting Enzyme in Dementia,” Neurobiology of Disease 26, no. 1 (2007): 27–35.17270454 10.1016/j.nbd.2006.11.011 · doi ↗ · pubmed ↗
- 8M. K. Singh‐Bains , V. Linke , M. D. R. Austria , et al., “Altered Microglia and Neurovasculature in the Alzheimer's Disease Cerebellum,” Neurobiology of Disease 132 (2019): 104589.31454549 10.1016/j.nbd.2019.104589 · doi ↗ · pubmed ↗