Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 107 of the 107 references it cites were found in a registry or an index; 44 of the 54 sources its claims cite were checked for support.

Lymphoid gene expression supports neuroprotective microglia function

Open access

Abstract

Abstract Microglia, the innate immune cells of the brain, play a defining role in the progression of Alzheimer’s disease (AD) 1 . The microglial response to amyloid plaques in AD can range from neuroprotective to neurotoxic 2 . Here we show that the protective function of microglia is governed by the transcription factor PU.1, which becomes downregulated following microglial contact with plaques. Lowering PU.1 expression in microglia reduces the severity of amyloid disease pathology in mice and is linked to the expression of immunoregulatory lymphoid receptor proteins, particularly CD28, a surface receptor that is critical for T cell activation 3,4 . Microglia-specific deficiency in CD28, which is expressed by a small subset of plaque-associated PU.1 low microglia, promotes a broad inflammatory microglial state that is associated with increased amyloid plaque load. Our findings indicate that PU.1 low CD28-expressing microglia may operate as suppressive microglia that mitigate the progression of AD by reducing the severity of neuroinflammation. This role of CD28 and potentially other lymphoid co-stimulatory and co-inhibitory receptor proteins in governing microglial responses in AD points to possible immunotherapy approaches for treating the disease by promoting protective microglial functions.

Overview

CoChat Summary

## TL;DR

The paper argues that a plaque-associated PU.1-low microglial state is neuroprotective in Alzheimer’s disease and is linked to expression of lymphoid receptor genes, especially CD28. It shows that plaque contact downregulates PU.1 through SYK–PLCγ2-dependent signaling, and that disrupting CD28 in microglia shifts them toward inflammation and higher plaque burden.

## Key findings

  • In the 5xFAD mouse model and in human AD tissue, a distinct PU.1-low microglial subpopulation preferentially co-localized with amyloid plaques and increased with disease progression.
  • Lowering PU.1 in microglia reduced the severity of amyloid disease pathology in mice; the paper also reports that microglia-specific deficiency in CD28 promoted a broad inflammatory microglial state associated with increased amyloid plaque load.
  • Plaque-associated PU.1-low microglia expressed immunoregulatory lymphoid receptor and signaling genes, including CD28, Pdcd1, Cd274, Cd5, Ctla2a, Cd52, Lat2, and Cd72, and these transcripts were also found in microglial ribosome-associated RNA.
  • Ex vivo ligands that engage TREM2 and CLEC7A induced PU.1 downregulation, and this effect was attenuated by phospholipase C inhibition; PLC activation alone was sufficient to downregulate PU.1.
  • Microglia-specific inactivation of SYK or PLCγ2 reduced the number of PU.1-low microglia and prevented their CSF1R-independent survival at plaques.

## Methods

  • Multi-modal design combining human post-mortem brain immunohistochemistry, mouse genetics in the 5xFAD AD model, ex vivo primary microglia culture, in vitro perturbations, and transcriptomic/epigenomic profiling.
  • Mouse cohorts included microglia-specific Spi1/PU.1 manipulation, Cd28, Syk, and Plcg2 conditional knockouts, plus 5xFAD crosses; human tissue included 3 post-mortem frontal cortex AD samples for immunohistochemistry.
  • Main analyses included single-cell and single-nucleus RNA-seq, TRAP sequencing, MERFISH spatial transcriptomics, ATAC-seq, proteomics, western blotting, and behavioral/survival assays, with standard group comparisons and clustering/differential-expression pipelines described in Methods.

## Limitations

  • The paper states that it did not recruit participants and used only 3 post-mortem human brain samples for immunohistochemistry; broader human validation is therefore limited by sample availability.
  • It states that sample size justification was not provided; instead, sample sizes were chosen pragmatically based on expected effect size, prior literature, and practical constraints.
  • Evident from the text: several key human and mouse analyses rely on de-identified post-mortem tissue and model systems, so causality in human AD is not directly established by the human data alone.

Model-written summary of a shortened version of the paper’s full text (it is longer than we read). Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1038/s41586-025-09662-z

Cites

107 of 107 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Fig. 1 Plaque-associated microglia include PU.1low subpopulation. a, Uniform manifold approximation and projections (UMAPs) of ex vivo isolated forebrain microglia from 8-month-old wild-type and 5xFAD mice (two males per group) show distinct PU.1 expression states (PU.1low, PU.1med and PU.1high) among DAM12 (red). b,c, PU.1 protein is reduced in a subset of plaque-associated microglia in the cortex of 6-month-old 5xFAD mice (n = 5; one female and four males) (b) and in the frontal cortex of individuals with AD (79–91 years; n = 3; one female and two males) (c). Bar graphs show the proportion of PU.1low cells in plaque-associated microglia versus distal microglia, analysed using a paired two-tailed t-test. d,e, Plaque-associated PU.1low microglia are resistant to CSF1R inhibition (CSF1Ri). d, UMAP visualizations of lineage-traced cortical microglia (TRAP28) from 6-month-old control (one female) or CSF1Ri-fed 5xFAD mice (two males), analysed using single-nucleus sequencing (DAM outlined with dotted line). The pie chart shows PU.1low fraction within DAM. e, Reduced PU.1 protein expression in CSF1Ri-resistant cortical microglia in the cortices of 6-month-old CSF1Ri (n = 4; one female and three males) versus control diet-fed (n = 8 mice; six females and two males) 5xFAD mice. Bar graph shows the proportion of PU.1low microglia, analysed using an unpaired two-tailed t-test. f, Microglial SYK/PLCγ2 signalling promotes the plaque-associated PU.1low state. Representative images and quantification of PU.1 expression in cortical microglia from 6-month-old 5xFAD;Cx3cr1CreErt2/+ (n = 5; three females and two males), 5xFAD–SYK-KO (5xFAD;Cx3cr1CreErt2/+;Sykfl/fl; n = 3; one female and two males) and 5xFAD–PLCγ2-KO mice (5xFAD;Cx3cr1CreErt2/+;Plcg2fl/fl; n = 3; two females and one male). Bar graphs show plaque-normalized microglial numbers and PU.1low microglia fractions, as determined by ordinary one-way ANOVA with multiple comparisons. Microglial nuclei identified using Imaris (b,f) and CellProfiler (e) in dotted circles. Bar graphs show the mean ± s.e.m. with individual points. Scale bars, 20 μm (b,c,f), 40 μm (d), 10 μm (e). Illustrations in a, d and f were created using BioRender (https://biorender.com). Source data
Fig. 1 Plaque-associated microglia include PU.1low subpopulation. a, Uniform manifold approximation and projections (UMAPs) of ex vivo isolated forebrain microglia from 8-month-old wild-type and 5xFAD mice (two males per group) show distinct PU.1 expression states (PU.1low, PU.1med and PU.1high) among DAM12 (red). b,c, PU.1 protein is reduced in a subset of plaque-associated microglia in the cortex of 6-month-old 5xFAD mice (n = 5; one female and four males) (b) and in the frontal cortex of individuals with AD (79–91 years; n = 3; one female and two males) (c). Bar graphs show the proportion of PU.1low cells in plaque-associated microglia versus distal microglia, analysed using a paired two-tailed t-test. d,e, Plaque-associated PU.1low microglia are resistant to CSF1R inhibition (CSF1Ri). d, UMAP visualizations of lineage-traced cortical microglia (TRAP28) from 6-month-old control (one female) or CSF1Ri-fed 5xFAD mice (two males), analysed using single-nucleus sequencing (DAM outlined with dotted line). The pie chart shows PU.1low fraction within DAM. e, Reduced PU.1 protein expression in CSF1Ri-resistant cortical microglia in the cortices of 6-month-old CSF1Ri (n = 4; one female and three males) versus control diet-fed (n = 8 mice; six females and two males) 5xFAD mice. Bar graph shows the proportion of PU.1low microglia, analysed using an unpaired two-tailed t-test. f, Microglial SYK/PLCγ2 signalling promotes the plaque-associated PU.1low state. Representative images and quantification of PU.1 expression in cortical microglia from 6-month-old 5xFAD;Cx3cr1CreErt2/+ (n = 5; three females and two males), 5xFAD–SYK-KO (5xFAD;Cx3cr1CreErt2/+;Sykfl/fl; n = 3; one female and two males) and 5xFAD–PLCγ2-KO mice (5xFAD;Cx3cr1CreErt2/+;Plcg2fl/fl; n = 3; two females and one male). Bar graphs show plaque-normalized microglial numbers and PU.1low microglia fractions, as determined by ordinary one-way ANOVA with multiple comparisons. Microglial nuclei identified using Imaris (b,f) and CellProfiler (e) in dotted circles. Bar graphs show the mean ± s.e.m. with individual points. Scale bars, 20 μm (b,c,f), 40 μm (d), 10 μm (e). Illustrations in a, d and f were created using BioRender (https://biorender.com). Source dataCC BY · PubMed Central
Fig. 2 Upregulation of lymphoid genes in PU.1low plaque-associated microglia. a, The volcano plot shows genes upregulated in PU.1low (red) versus PU.1high (blue) DAM in 8-month-old wild-type and 5xFAD mice (two males per group), as determined by Wilcoxon rank-sum test. The dot plot highlights selected myeloid/microglia-associated and lymphoid-lineage-associated genes in PU.1low and PU.1high DAM. b,c, PU.1low lymphoid gene-expressing cortical microglia localize near amyloid plaques. Independently repeated twice. b, Representative MERFISH images of plaque-associated and distal microglia from an 8-month-old male 5xFAD brain. c, UMAP of spatial microglia distribution to the nearest plaque using MERFISH. The volcano plot shows differentially expressed genes in plaque-associated microglia (red; less than 15 μm to plaques) versus distal microglia (blue; greater than 15 μm to plaques), as determined by Wilcoxon rank-sum test. d, Lymphoid protein-encoding RNAs associate with ribosomes in lineage-traced microglia. The heat map (z-score) shows selected lymphoid lineage genes identified using microglia-specific TRAP sequencing28 from the cortices of 6-month-old 5xFAD mice (Cx3cr1CreErt2/+;5xFAD;Eef1a1LSL.eGFPL10a/+) versus control (Cx3cr1CreErt2/+;Eef1a1LSL.eGFPL10a/+ mice; n = 6 per group; three females and three males), as determined using DESeq2. Each column represents an individual mouse. e, PU.1low lymphoid gene-expressing microglia are present in individuals with AD. The dot plots show scaled average expression of selected genes in human microglia15 organized into homeostatic, lymphoid+ DAM and lymphoid– DAM clusters. Scale bar, 5 μm. Illustrations in c and d were created using BioRender (https://biorender.com). Max., maximum; Min., minimum.
Fig. 2 Upregulation of lymphoid genes in PU.1low plaque-associated microglia. a, The volcano plot shows genes upregulated in PU.1low (red) versus PU.1high (blue) DAM in 8-month-old wild-type and 5xFAD mice (two males per group), as determined by Wilcoxon rank-sum test. The dot plot highlights selected myeloid/microglia-associated and lymphoid-lineage-associated genes in PU.1low and PU.1high DAM. b,c, PU.1low lymphoid gene-expressing cortical microglia localize near amyloid plaques. Independently repeated twice. b, Representative MERFISH images of plaque-associated and distal microglia from an 8-month-old male 5xFAD brain. c, UMAP of spatial microglia distribution to the nearest plaque using MERFISH. The volcano plot shows differentially expressed genes in plaque-associated microglia (red; less than 15 μm to plaques) versus distal microglia (blue; greater than 15 μm to plaques), as determined by Wilcoxon rank-sum test. d, Lymphoid protein-encoding RNAs associate with ribosomes in lineage-traced microglia. The heat map (z-score) shows selected lymphoid lineage genes identified using microglia-specific TRAP sequencing28 from the cortices of 6-month-old 5xFAD mice (Cx3cr1CreErt2/+;5xFAD;Eef1a1LSL.eGFPL10a/+) versus control (Cx3cr1CreErt2/+;Eef1a1LSL.eGFPL10a/+ mice; n = 6 per group; three females and three males), as determined using DESeq2. Each column represents an individual mouse. e, PU.1low lymphoid gene-expressing microglia are present in individuals with AD. The dot plots show scaled average expression of selected genes in human microglia15 organized into homeostatic, lymphoid+ DAM and lymphoid– DAM clusters. Scale bar, 5 μm. Illustrations in c and d were created using BioRender (https://biorender.com). Max., maximum; Min., minimum.CC BY · PubMed Central
Fig. 3 PU.1 levels regulate lymphoid gene expression in microglia. a, Microglial PU.1 protein expression (green) correlates with Spi1 gene copy number in PU.1-low, PU.1-wt and PU.1-high mice. Bar graph shows the mean PU.1 intensity within the cortical microglia in 6-month-old control (Cx3cr1CreErt2/+; n = 3; one female and two males), PU.1-low (Cx3cr1CreErt2/+;Spi1fl/+; n = 3; one female and two males) and PU.1-high (Cx3cr1CreErt2/+;R26LSL.F-Spi1/+; n = 3 females) mice, as determined by ordinary one-way ANOVA with multiple comparisons. Microglial nuclei identified by CellProfiler are outlined (dotted circles). b, PU.1low microglia upregulate lymphoid-lineage genes. The heat map shows differentially expressed genes (DESeq2; z-scored variance-stabilizing transformation (VST)) from cortical microglia-specific TRAP sequencing from 6-month-old PU.1-low (n = 6; four females and two males), PU.1-wt (n = 6; three females and three males) and PU.1-high (n = 4; three females and one male) mice. c,d, PU.1low microglia display T-cell-like chromatin accessibility profiles in vivo. c, Profile plots (top) and heat maps (bottom) show normalized genome-wide chromatin accessibility by assay for transposase-accessible chromatin sequencing (ATAC-seq) (reads per kilobase per million mapped reads (RPKM)) from the microglia of PU.1-low, PU.1-wt and PU.1-high mice (9–10 months old; n = 3 males per genotype) and splenic T cells (n = 2 males) from wild-type mice. Data are centered on ±4 kb around differentially accessible (DESeq2) peaks in PU.1-low compared with wild-type microglia. Peaks are ranked by signal intensity using the same minimum and maximum ranges for plots and heat maps. d, Genome browser views (Integrative Genomics Viewer) of Cd28 and Cd22 gene loci in T cells versus PU.1-low, PU.1-wt and PU.1-high microglia. Bracketed numbers indicate the minimum and maximum y-axis values for each sample. Scale bar, 5 μm. Illustrations in a were created using BioRender (https://biorender.com). wt, wild type. Source data
Fig. 3 PU.1 levels regulate lymphoid gene expression in microglia. a, Microglial PU.1 protein expression (green) correlates with Spi1 gene copy number in PU.1-low, PU.1-wt and PU.1-high mice. Bar graph shows the mean PU.1 intensity within the cortical microglia in 6-month-old control (Cx3cr1CreErt2/+; n = 3; one female and two males), PU.1-low (Cx3cr1CreErt2/+;Spi1fl/+; n = 3; one female and two males) and PU.1-high (Cx3cr1CreErt2/+;R26LSL.F-Spi1/+; n = 3 females) mice, as determined by ordinary one-way ANOVA with multiple comparisons. Microglial nuclei identified by CellProfiler are outlined (dotted circles). b, PU.1low microglia upregulate lymphoid-lineage genes. The heat map shows differentially expressed genes (DESeq2; z-scored variance-stabilizing transformation (VST)) from cortical microglia-specific TRAP sequencing from 6-month-old PU.1-low (n = 6; four females and two males), PU.1-wt (n = 6; three females and three males) and PU.1-high (n = 4; three females and one male) mice. c,d, PU.1low microglia display T-cell-like chromatin accessibility profiles in vivo. c, Profile plots (top) and heat maps (bottom) show normalized genome-wide chromatin accessibility by assay for transposase-accessible chromatin sequencing (ATAC-seq) (reads per kilobase per million mapped reads (RPKM)) from the microglia of PU.1-low, PU.1-wt and PU.1-high mice (9–10 months old; n = 3 males per genotype) and splenic T cells (n = 2 males) from wild-type mice. Data are centered on ±4 kb around differentially accessible (DESeq2) peaks in PU.1-low compared with wild-type microglia. Peaks are ranked by signal intensity using the same minimum and maximum ranges for plots and heat maps. d, Genome browser views (Integrative Genomics Viewer) of Cd28 and Cd22 gene loci in T cells versus PU.1-low, PU.1-wt and PU.1-high microglia. Bracketed numbers indicate the minimum and maximum y-axis values for each sample. Scale bar, 5 μm. Illustrations in a were created using BioRender (https://biorender.com). wt, wild type. Source dataCC BY · PubMed Central
Fig. 4 PU.1low microglia exhibit neuroprotective phenotypes. a,b, UMAP (a) and bar graph (b) show microglial subpopulations and their distribution on the basis of MERFISH data in wild-type (n = 2 males), 5xFAD (n = 3 males) and 5xFAD–PU.1-low (n = 2 males) mice (8–10 months old). c–f, Representative cortical immunofluorescence and quantification in 6-month-old mice, analysed using an unpaired two-tailed t-test. c, Plaque-associated microglia in 5xFAD (n = 6; three females and three males) and 5xFAD–PU.1-low (n = 5; three females and two males) mice. d, CD28+ microglia in 5xFAD (n = 6; three females and three males) and 5xFAD–PU.1-low (n = 5; four females and one male) mice. e, MX1–GFP+ microglia in 5xFAD;Mx1–GFP (n = 4; two females and two males) and 5xFAD–PU.1-low;Mx1–GFP (n = 3; two females and one male) mice. f, Plaque composition (6E10+ soluble amyloid-β relative to ThioS+ dense-core plaques) in the subiculum of 5xFAD (n = 7; two females and five males) and 5xFAD–PU.1-low (n = 5; two females and three males) mice. g, Seeding of AT8+ phosphorylated Tau (p-Tau) in 10-month-old mouse brains injected with human Tau aggregates. Bar graphs show percentage of AT8+ area in 5xFAD (n = 12; seven females and five males) and 5xFAD–PU.1-low (n = 9; four females and five males) mice, analysed using an unpaired two-tailed t-test. h–k, Synaptic analyses were performed in 6-month-old mice, and physiological and behavioural analyses were performed in 9–10-month-old mice, using ordinary one-way ANOVA with multiple comparisons. h, VGLUT2 and Bassoon protein density in layer V cortex of control (n = 10; four females and six males), 5xFAD (VGLUT2 (n = 17; eight females and nine males); Bassoon (n = 14; six females and eight males)) and 5xFAD–PU.1-low (n = 6; four females and two males) mice. i, Late-phase LTP in CA1 hippocampal neurons of control (n = 12 slices from five mice; three females and two males), 5xFAD (n = 13 slices from six mice; four females and two males) and 5xFAD–PU.1-low (n = 11 slices from five mice; one female and four males) mice. j, Open-field thigmotaxis (centre/periphery) in control (n = 15; nine females and six males), 5xFAD (n = 12; six females and six males) and 5xFAD–PU.1-low (n = 13; seven females and six males) mice. k, Novel object recognition scores in control (n = 10; five females and five males), 5xFAD (n = 11; six females and five males) and 5xFAD–PU.1-low (n = 15; seven females and eight males) mice. l, Survival curves for wild-type (n = 113; 49 females and 64 males), 5xFAD (n = 81; 38 females and 43 males), 5xFAD–PU.1-low (n = 42; 19 females and 23 males) and 5xFAD–PU.1-high (n = 34; 22 females and 12 males) mice, analysed using Kaplan–Meier curve with a simple log-rank Mantel–Cox test. Bar graphs show the mean ± s.e.m. with individual points. Scale bars, 20 μm (c–f,h), 1 mm (g). Illustration in g was created using BioRender (https://biorender.com). Source data
Fig. 4 PU.1low microglia exhibit neuroprotective phenotypes. a,b, UMAP (a) and bar graph (b) show microglial subpopulations and their distribution on the basis of MERFISH data in wild-type (n = 2 males), 5xFAD (n = 3 males) and 5xFAD–PU.1-low (n = 2 males) mice (8–10 months old). c–f, Representative cortical immunofluorescence and quantification in 6-month-old mice, analysed using an unpaired two-tailed t-test. c, Plaque-associated microglia in 5xFAD (n = 6; three females and three males) and 5xFAD–PU.1-low (n = 5; three females and two males) mice. d, CD28+ microglia in 5xFAD (n = 6; three females and three males) and 5xFAD–PU.1-low (n = 5; four females and one male) mice. e, MX1–GFP+ microglia in 5xFAD;Mx1–GFP (n = 4; two females and two males) and 5xFAD–PU.1-low;Mx1–GFP (n = 3; two females and one male) mice. f, Plaque composition (6E10+ soluble amyloid-β relative to ThioS+ dense-core plaques) in the subiculum of 5xFAD (n = 7; two females and five males) and 5xFAD–PU.1-low (n = 5; two females and three males) mice. g, Seeding of AT8+ phosphorylated Tau (p-Tau) in 10-month-old mouse brains injected with human Tau aggregates. Bar graphs show percentage of AT8+ area in 5xFAD (n = 12; seven females and five males) and 5xFAD–PU.1-low (n = 9; four females and five males) mice, analysed using an unpaired two-tailed t-test. h–k, Synaptic analyses were performed in 6-month-old mice, and physiological and behavioural analyses were performed in 9–10-month-old mice, using ordinary one-way ANOVA with multiple comparisons. h, VGLUT2 and Bassoon protein density in layer V cortex of control (n = 10; four females and six males), 5xFAD (VGLUT2 (n = 17; eight females and nine males); Bassoon (n = 14; six females and eight males)) and 5xFAD–PU.1-low (n = 6; four females and two males) mice. i, Late-phase LTP in CA1 hippocampal neurons of control (n = 12 slices from five mice; three females and two males), 5xFAD (n = 13 slices from six mice; four females and two males) and 5xFAD–PU.1-low (n = 11 slices from five mice; one female and four males) mice. j, Open-field thigmotaxis (centre/periphery) in control (n = 15; nine females and six males), 5xFAD (n = 12; six females and six males) and 5xFAD–PU.1-low (n = 13; seven females and six males) mice. k, Novel object recognition scores in control (n = 10; five females and five males), 5xFAD (n = 11; six females and five males) and 5xFAD–PU.1-low (n = 15; seven females and eight males) mice. l, Survival curves for wild-type (n = 113; 49 females and 64 males), 5xFAD (n = 81; 38 females and 43 males), 5xFAD–PU.1-low (n = 42; 19 females and 23 males) and 5xFAD–PU.1-high (n = 34; 22 females and 12 males) mice, analysed using Kaplan–Meier curve with a simple log-rank Mantel–Cox test. Bar graphs show the mean ± s.e.m. with individual points. Scale bars, 20 μm (c–f,h), 1 mm (g). Illustration in g was created using BioRender (https://biorender.com). Source dataCC BY · PubMed Central
Fig. 5 Microglial CD28 regulates inflammatory activation and amyloid plaque burden in 5xFAD mice. a,UMAPs show CD28+ microglia in 5xFAD DAM. Single-cell RNA-seq of forebrain microglia from 3-month-old, 6-month-old and 8-month-old control and 5xFAD mice (3 months (n = 1 control and n = 2 5xFAD), 6 months (n = 1 per group) and 8 months (n = 2 per group); all males); DAM outlined (dotted line). Dot and feature plots show microglial Cd28 expression. b, Inducible microglia-specific deletion of CD28 in 5xFAD mice. Representative images and quantification of CD28+ microglia in 6-month-old 5xFAD (n = 3; one female and two males) and 5xFAD;CD28-KO (5xFAD;Cx3cr1CreErt2/+;Cd28fl/fl; n = 3; two females and one male) mice were analysed using an unpaired two-tailed t-test. c–e, Microglial CD28 influences plaque load and microglial activation. c, Microglial area and plaque density (% ThioS) in dentate gyrus of 3-month-old and 6-month-old 5xFAD (3 months (n = 6; one female and five males) and 6 months (n = 5; two females and three males)) and 5xFAD;CD28-KO mice (3 months (n = 5; two females and three males) and 6 months (n = 7; four females and three males) were analysed using two-way ANOVA with multiple comparisons. d, Single-cell sequencing of forebrain microglia from 5xFAD and 5xFAD;CD28-KO mice. Significant gene ontologies (Fisher’s exact test) of differentially regulated genes using pseudobulk analysis (Wilcoxon rank-sum test) in 3- and 6-month-old mice by genotype (n = two 3-month-old male mice and n = one 6-month-old male mouse per group). e, UMAP (two 3-month-old male mice per group) and dot plots (control and 5xFAD (as in a) and 5xFAD;CD28-KO (two 3-month-old male mice and one 6-month-old male mouse) show expression of interferon-response genes. f,g, Human PU.1-lowering allele associates with increased lymphoid gene-expressing microglia. f, The heat map shows scaled average expression of indicated genes in surveillance clusters (Mic. 2, 3 and 4) and lipid-associated/DAM clusters (Mic. 12 and 13) in human microglia17. g, PU.1 protective allele carriers have lower SPI1 expression, whereas risk allele carriers show higher SPI1 expression in AD microglia. Normalized SPI1 expression and fraction of CD28+ Mic. 13 population per individual versus cognitive diagnosis and PU.1 variant (rs1057233 A/A = risk (n = 61 not impaired; n = 64 AD); G/A and G/G = protective (n = 83 not impaired; n = 98 AD)) were analysed using two-way ANOVA with Tukey post hoc test. Bar graphs with individual data points show the mean ± s.e.m. Scale bars, 10 μm (b), 100 μm (c). Illustrations in b were created using BioRender (https://biorender.com). Mic., microglia. Source data
Fig. 5 Microglial CD28 regulates inflammatory activation and amyloid plaque burden in 5xFAD mice. a,UMAPs show CD28+ microglia in 5xFAD DAM. Single-cell RNA-seq of forebrain microglia from 3-month-old, 6-month-old and 8-month-old control and 5xFAD mice (3 months (n = 1 control and n = 2 5xFAD), 6 months (n = 1 per group) and 8 months (n = 2 per group); all males); DAM outlined (dotted line). Dot and feature plots show microglial Cd28 expression. b, Inducible microglia-specific deletion of CD28 in 5xFAD mice. Representative images and quantification of CD28+ microglia in 6-month-old 5xFAD (n = 3; one female and two males) and 5xFAD;CD28-KO (5xFAD;Cx3cr1CreErt2/+;Cd28fl/fl; n = 3; two females and one male) mice were analysed using an unpaired two-tailed t-test. c–e, Microglial CD28 influences plaque load and microglial activation. c, Microglial area and plaque density (% ThioS) in dentate gyrus of 3-month-old and 6-month-old 5xFAD (3 months (n = 6; one female and five males) and 6 months (n = 5; two females and three males)) and 5xFAD;CD28-KO mice (3 months (n = 5; two females and three males) and 6 months (n = 7; four females and three males) were analysed using two-way ANOVA with multiple comparisons. d, Single-cell sequencing of forebrain microglia from 5xFAD and 5xFAD;CD28-KO mice. Significant gene ontologies (Fisher’s exact test) of differentially regulated genes using pseudobulk analysis (Wilcoxon rank-sum test) in 3- and 6-month-old mice by genotype (n = two 3-month-old male mice and n = one 6-month-old male mouse per group). e, UMAP (two 3-month-old male mice per group) and dot plots (control and 5xFAD (as in a) and 5xFAD;CD28-KO (two 3-month-old male mice and one 6-month-old male mouse) show expression of interferon-response genes. f,g, Human PU.1-lowering allele associates with increased lymphoid gene-expressing microglia. f, The heat map shows scaled average expression of indicated genes in surveillance clusters (Mic. 2, 3 and 4) and lipid-associated/DAM clusters (Mic. 12 and 13) in human microglia17. g, PU.1 protective allele carriers have lower SPI1 expression, whereas risk allele carriers show higher SPI1 expression in AD microglia. Normalized SPI1 expression and fraction of CD28+ Mic. 13 population per individual versus cognitive diagnosis and PU.1 variant (rs1057233 A/A = risk (n = 61 not impaired; n = 64 AD); G/A and G/G = protective (n = 83 not impaired; n = 98 AD)) were analysed using two-way ANOVA with Tukey post hoc test. Bar graphs with individual data points show the mean ± s.e.m. Scale bars, 10 μm (b), 100 μm (c). Illustrations in b were created using BioRender (https://biorender.com). Mic., microglia. Source dataCC BY · PubMed Central
Extended Data Fig. 1 Downregulation of PU.1 is plaque signal-driven and associated with CSF1R-independent microglia survival (related to Fig. 1). (a) Increased PU.1low microglia abundance during AD disease progression. PU.1 protein expression in cortical microglia is shown. Scale bar: 20 μm. Microglial nuclei, as identified by Imaris, are indicated in dotted circles. The bar graphs show the size of PU.1low microglia subpopulation in the cortex of 3-month-old (n = 5 mice, 2♀, 3♂), 6-month-old (n = 5 mice, 3♀, 2♂), 9-month-old (n = 3 mice, 3♀), and 12-month-old (n = 3 mice, 1♀, 2♂) 5xFAD mice—ordinary one-way ANOVA with multiple comparisons. (b) Spatial definition of plaque-associated microglia (closest microglia processes within <5 μm of ThioS+ plaque surface) versus distal microglia. (c) Strategy for microglia depletion using CSF1R inhibitor (CSF1Ri, PLX5622, 1200 ppm diet) for ~6 weeks starting at 4.5 months of age in control and 5xFAD mice. (d) CSF1R-independent microglia survival at the plaque. Representative immunofluorescence images are shown. CSF1Ri treatment ablates >80% of distal microglia while >60% of plaque-associated microglia remain. Bar graphs show: Left: the total number of all microglia per mm3 in the cortex of 6-month-old control diet- and CSF1Ri diet-fed wild-type (left) or 5xFAD mice (right). Middle: the number of distal (left) and plaque-associated (right) microglia per mm3. Right: the average number of plaque-associated microglia normalized to plaque numbers in control or CSF1Ri-treated 5xFAD mice. Wild-type mice: n = 7/group, 4♀, 3♂. 5xFAD mice: n = 8 mice/group, 5♀, 3♂—unpaired two-tailed t-test. (e) SYK/PLCγ2 are required for PU.1low microglia at the plaque. Representative immunofluorescence images are shown. Bar graphs show the average PU.1 protein intensity in microglia in the cortex of 6-month-old control (Cx3cr1CreErt2/+, n = 4 mice, 2♀, 2♂), SYK-KO (Cx3cr1CreErt2/+; Sykfl/fl, n = 3 mice, 2♀, 1♂), and PLCγ2-KO mice (Cx3cr1CreErt2/+; Plcg2fl/fl, n = 3 mice, 3♂), as well as in 5xFAD (5xFAD; Cx3cr1CreErt2/+, n = 5 mice, 3♀, 2♂), 5xFAD-SYK-KO (5xFAD; Cx3cr1CreErt2/+; Sykfl/fl, n = 3 mice, 1♀, 2♂), and 5xFAD; PLCγ2-KO mice (5xFAD; Cx3cr1CreErt2/+; Plcg2fl/fl, n = 3 mice, 2♀, 1♂)—ordinary one-way ANOVA with multiple comparisons. (f) CSF1R-independent survival of plaque-associated PU.1low microglia depends on PLCγ2 activity. Bar graph shows the number of plaque-associated microglia per plaque and number of distal microglia per mm3 in CSF1Ri-treated 5xFAD (n = 4 mice, 2♀, 2♂), CSF1Ri- and PLCi-treated 5xFAD (pharmacological PLC inhibition: PLC inhibitor, 8.7 mg/kg U73122, n = 6 mice, 2♀, 4♂) and CSF1Ri-treated 5xFAD; PLCγ2-KO mice (genetic microglia-specific PLCγ2 deletion: n = 4 mice, 3♀, 1♂)—ordinary one-way ANOVA with multiple comparisons. The dotted line indicates the expected number of distal microglia in the cortex of 5xFAD mice (without CSF1Ri treatment). Bar graphs with individual data points show mean ± SEM. Illustrations in b, c, e and f were created using BioRender (https://biorender.com). Source data
Extended Data Fig. 1 Downregulation of PU.1 is plaque signal-driven and associated with CSF1R-independent microglia survival (related to Fig. 1). (a) Increased PU.1low microglia abundance during AD disease progression. PU.1 protein expression in cortical microglia is shown. Scale bar: 20 μm. Microglial nuclei, as identified by Imaris, are indicated in dotted circles. The bar graphs show the size of PU.1low microglia subpopulation in the cortex of 3-month-old (n = 5 mice, 2♀, 3♂), 6-month-old (n = 5 mice, 3♀, 2♂), 9-month-old (n = 3 mice, 3♀), and 12-month-old (n = 3 mice, 1♀, 2♂) 5xFAD mice—ordinary one-way ANOVA with multiple comparisons. (b) Spatial definition of plaque-associated microglia (closest microglia processes within <5 μm of ThioS+ plaque surface) versus distal microglia. (c) Strategy for microglia depletion using CSF1R inhibitor (CSF1Ri, PLX5622, 1200 ppm diet) for ~6 weeks starting at 4.5 months of age in control and 5xFAD mice. (d) CSF1R-independent microglia survival at the plaque. Representative immunofluorescence images are shown. CSF1Ri treatment ablates >80% of distal microglia while >60% of plaque-associated microglia remain. Bar graphs show: Left: the total number of all microglia per mm3 in the cortex of 6-month-old control diet- and CSF1Ri diet-fed wild-type (left) or 5xFAD mice (right). Middle: the number of distal (left) and plaque-associated (right) microglia per mm3. Right: the average number of plaque-associated microglia normalized to plaque numbers in control or CSF1Ri-treated 5xFAD mice. Wild-type mice: n = 7/group, 4♀, 3♂. 5xFAD mice: n = 8 mice/group, 5♀, 3♂—unpaired two-tailed t-test. (e) SYK/PLCγ2 are required for PU.1low microglia at the plaque. Representative immunofluorescence images are shown. Bar graphs show the average PU.1 protein intensity in microglia in the cortex of 6-month-old control (Cx3cr1CreErt2/+, n = 4 mice, 2♀, 2♂), SYK-KO (Cx3cr1CreErt2/+; Sykfl/fl, n = 3 mice, 2♀, 1♂), and PLCγ2-KO mice (Cx3cr1CreErt2/+; Plcg2fl/fl, n = 3 mice, 3♂), as well as in 5xFAD (5xFAD; Cx3cr1CreErt2/+, n = 5 mice, 3♀, 2♂), 5xFAD-SYK-KO (5xFAD; Cx3cr1CreErt2/+; Sykfl/fl, n = 3 mice, 1♀, 2♂), and 5xFAD; PLCγ2-KO mice (5xFAD; Cx3cr1CreErt2/+; Plcg2fl/fl, n = 3 mice, 2♀, 1♂)—ordinary one-way ANOVA with multiple comparisons. (f) CSF1R-independent survival of plaque-associated PU.1low microglia depends on PLCγ2 activity. Bar graph shows the number of plaque-associated microglia per plaque and number of distal microglia per mm3 in CSF1Ri-treated 5xFAD (n = 4 mice, 2♀, 2♂), CSF1Ri- and PLCi-treated 5xFAD (pharmacological PLC inhibition: PLC inhibitor, 8.7 mg/kg U73122, n = 6 mice, 2♀, 4♂) and CSF1Ri-treated 5xFAD; PLCγ2-KO mice (genetic microglia-specific PLCγ2 deletion: n = 4 mice, 3♀, 1♂)—ordinary one-way ANOVA with multiple comparisons. The dotted line indicates the expected number of distal microglia in the cortex of 5xFAD mice (without CSF1Ri treatment). Bar graphs with individual data points show mean ± SEM. Illustrations in b, c, e and f were created using BioRender (https://biorender.com). Source dataCC BY · PubMed Central
Extended Data Fig. 2 Signal-induced downregulation of PU.1 in microglia (related to Fig. 1). (a) Experimental design (left), representative images (middle), and quantification (right) of Spi1 mRNA (measured by single-molecule in situ hybridization) and PU.1 protein expression (measured by immunofluorescence) in cultured primary microglia treated with vehicle or 1 mg/ml of phosphatidylserine/phosphatidylcholine (PS/PC) liposomes for either 12 h or 3 days. Spi1 mRNA: red, PU.1 protein: green. Scale bar: 50 μm. The bar graphs show Spi1 mRNA expression (left) and PU.1 protein expression (right) normalized to the mean of the control for each time point—an unpaired two-tailed t-test was used. Quantification shown is from 2 independent experiments (12 h: experiment 1: n = 4 wells/group, experiment 2: n = 4 wells/group; total: n = 8 wells/group. 3 days: experiment 1: n = 4 wells/group, experiment 2: n = 3 wells/control, n = 2 wells/liposome-treated; total: n = 7 wells/control group, n = 6 wells/liposome-treated group). (b) Experimental design (top left), representative immunofluorescence images (bottom left), and quantification (right) of PU.1 intensity in cultured primary microglia treated with vehicle or PS/PC liposomes at 0.1 or 1 mg/ml for 3 days. Scale bar: 50 μm. The bar graph shows quantification of PU.1 protein expression in microglia nuclei with treatment of vehicle control (n = 6 wells), 0.1 mg/ml PS/PC liposomes (n = 3 wells), and 1 mg/ml PS/PC (n = 3 wells), normalized to the mean of vehicle control—ordinary one-way ANOVA with multiple comparisons. The experiment was independently repeated. (c) Liposome-induced PU.1 downregulation is reversible. Experimental design (top left), representative immunofluorescence images (bottom left), and quantification (right) of PU.1 levels in cultured primary microglia treated with vehicle, PS/PC liposomes at 1 mg/ml for 3 days, and received washout (4 days). CD11B+ microglia: red, PU.1: green, DAPI+ nuclei: blue. Scale bar: 50 μm. The bar graph shows PU.1 levels in microglial nuclei in vehicle-treated, 3-day liposome-treated (1 mg/ml), or liposome-treated followed by washout (n = 3 wells/group)—ordinary one-way ANOVA with multiple comparisons. (d) CLEC7A agonist-induced PU.1 downregulation. Experimental design (top), representative immunofluorescence images (bottom left), and quantification (bottom right) of PU.1 expression in cultured primary microglia treated with vehicle or CLEC7A agonist, 50 μg/mL pustulan for 3 days—unpaired two-tailed t-test (n = 3 wells/group). (e) Experimental design (top left), representative immunofluorescence images of PU.1 expression (right), and quantification (bottom left) of cultured primary microglia treated with vehicle or 0.1 mg/ml liposomes for 3 days following a 1-h pre-treatment of PLC inhibitor (PLCi, 0.1 μM U73122) or vehicle DMSO, or only PLC activator (PLCa, 40 μM m-3M3FBS). CD11B+ microglia: red, PU.1: green, DAPI+ nuclei: blue. Scale bar: 50 μm. The bar graphs show quantification of PU.1 expression in microglia nuclei normalized to vehicle control. Left for PLCi: n = 4 wells/group—ordinary one-way ANOVA with multiple comparisons. The PLCi experiment was independently repeated. The PLCa experiment: Data are pooled from two independent experiments (experiment 1: n = 3/group; experiment 2: n = 4/group)—unpaired two-tailed t-test. (f) PLC inhibitor treatment alone does not change baseline PU.1 expression. Experimental design (top), representative immunofluorescence images (bottom left), and quantification (bottom right) of PU.1 expression in cultured primary microglia treated with vehicle (n = 4 wells) or PLC inhibitor (PLCi, 0.1 μM U73122) for 3 days—an unpaired two-tailed t-test was used (n = 4 wells/group). Bar graphs with individual data points show mean ± SEM. Illustrations in a–f were created using BioRender (https://biorender.com). Source data
Extended Data Fig. 2 Signal-induced downregulation of PU.1 in microglia (related to Fig. 1). (a) Experimental design (left), representative images (middle), and quantification (right) of Spi1 mRNA (measured by single-molecule in situ hybridization) and PU.1 protein expression (measured by immunofluorescence) in cultured primary microglia treated with vehicle or 1 mg/ml of phosphatidylserine/phosphatidylcholine (PS/PC) liposomes for either 12 h or 3 days. Spi1 mRNA: red, PU.1 protein: green. Scale bar: 50 μm. The bar graphs show Spi1 mRNA expression (left) and PU.1 protein expression (right) normalized to the mean of the control for each time point—an unpaired two-tailed t-test was used. Quantification shown is from 2 independent experiments (12 h: experiment 1: n = 4 wells/group, experiment 2: n = 4 wells/group; total: n = 8 wells/group. 3 days: experiment 1: n = 4 wells/group, experiment 2: n = 3 wells/control, n = 2 wells/liposome-treated; total: n = 7 wells/control group, n = 6 wells/liposome-treated group). (b) Experimental design (top left), representative immunofluorescence images (bottom left), and quantification (right) of PU.1 intensity in cultured primary microglia treated with vehicle or PS/PC liposomes at 0.1 or 1 mg/ml for 3 days. Scale bar: 50 μm. The bar graph shows quantification of PU.1 protein expression in microglia nuclei with treatment of vehicle control (n = 6 wells), 0.1 mg/ml PS/PC liposomes (n = 3 wells), and 1 mg/ml PS/PC (n = 3 wells), normalized to the mean of vehicle control—ordinary one-way ANOVA with multiple comparisons. The experiment was independently repeated. (c) Liposome-induced PU.1 downregulation is reversible. Experimental design (top left), representative immunofluorescence images (bottom left), and quantification (right) of PU.1 levels in cultured primary microglia treated with vehicle, PS/PC liposomes at 1 mg/ml for 3 days, and received washout (4 days). CD11B+ microglia: red, PU.1: green, DAPI+ nuclei: blue. Scale bar: 50 μm. The bar graph shows PU.1 levels in microglial nuclei in vehicle-treated, 3-day liposome-treated (1 mg/ml), or liposome-treated followed by washout (n = 3 wells/group)—ordinary one-way ANOVA with multiple comparisons. (d) CLEC7A agonist-induced PU.1 downregulation. Experimental design (top), representative immunofluorescence images (bottom left), and quantification (bottom right) of PU.1 expression in cultured primary microglia treated with vehicle or CLEC7A agonist, 50 μg/mL pustulan for 3 days—unpaired two-tailed t-test (n = 3 wells/group). (e) Experimental design (top left), representative immunofluorescence images of PU.1 expression (right), and quantification (bottom left) of cultured primary microglia treated with vehicle or 0.1 mg/ml liposomes for 3 days following a 1-h pre-treatment of PLC inhibitor (PLCi, 0.1 μM U73122) or vehicle DMSO, or only PLC activator (PLCa, 40 μM m-3M3FBS). CD11B+ microglia: red, PU.1: green, DAPI+ nuclei: blue. Scale bar: 50 μm. The bar graphs show quantification of PU.1 expression in microglia nuclei normalized to vehicle control. Left for PLCi: n = 4 wells/group—ordinary one-way ANOVA with multiple comparisons. The PLCi experiment was independently repeated. The PLCa experiment: Data are pooled from two independent experiments (experiment 1: n = 3/group; experiment 2: n = 4/group)—unpaired two-tailed t-test. (f) PLC inhibitor treatment alone does not change baseline PU.1 expression. Experimental design (top), representative immunofluorescence images (bottom left), and quantification (bottom right) of PU.1 expression in cultured primary microglia treated with vehicle (n = 4 wells) or PLC inhibitor (PLCi, 0.1 μM U73122) for 3 days—an unpaired two-tailed t-test was used (n = 4 wells/group). Bar graphs with individual data points show mean ± SEM. Illustrations in a–f were created using BioRender (https://biorender.com). Source dataCC BY · PubMed Central
Extended Data Fig. 3 PU.1low plaque-associated microglia express lymphoid genes (related to Fig. 2). (a, b) PU.1low plaque-associated microglia show lymphoid gene expression. Single-cell mRNA sequencing of forebrain microglia from 8-month-old wild-type (2♂) and 5xFAD mice (2♂) from Figs. 1a and 2a, Supplementary Fig. 1, and Supplementary Table 1. (a) Top: Isolation of microglia by fluorescence-activated cell sorting (FACS) followed by single-cell mRNA sequencing. Bottom: The balloon plot shows differentially regulated gene ontologies (GOs, Fisher’s exact test) in the PU.1low and PU.1high cortical disease-associated microglia (DAM) subpopulations. (b) Scatter plot shows the correlation of upregulated genes in cortical PU.1high (red) vs. PU.1low microglia (blue) between two independent experiments (n = 1 mouse/group per experiment) using FindMarkers() analysis in Seurat—Wilcoxon rank sum test. 558 genes are significantly upregulated and 445 genes are significantly downregulated in PU.1low microglia in two independent experiments, while 0 genes are regulated in opposing directions. (c) Plaque-associated, CSF1R-independent PU.1low microglia show lymphoid gene expression. Single-nuclei sequencing of cortical microglia from 6-month-old control (♀) and two CSF1Ri-treated 5xFAD mice (2♂ pooled). Left: Isolation of lineage-traced microglial nuclei from the cortex by fluorescence-activated nuclear sorting (FANS) followed by single-nuclei sequencing. Right: The volcano plot shows the differentially up- (red) and down- (blue) regulated genes in cortical microglia control and CSF1Ri-treated mice from Fig. 1d by FindMarkers() function in Seurat—Wilcoxon rank sum test. Please also see Supplementary Fig. 2 and Supplementary Table 3. (d) PU.1low microglia-expressed lymphoid-associated signaling molecules (ligands, surface receptors, and downstream signaling proteins) in T-cells are shown. Adapted from “T-cell and B-cell signaling” and “TCR downstream signaling” generated by BioRender.com. (e) Heatmap shows log2 (fold change) of selected differentially expressed genes in the PU.1low DAM compared to the PU.1high DAM, two independent experiments: (1) single-cell sequencing of forebrain microglia from wild-type and 5xFAD mice (Figs. 1a, 2a, Extended Data Fig. 3a) and (2) single-nuclei sequencing of cortical microglia from control diet- and CSF1Ri diet-fed 5xFAD mice (Fig. 1d, Extended Data Fig. 3c). The color indicates log2 (fold change). (f) Top: Schematic shows multiplexed error-robust fluorescent in situ hybridization (MERFISH) within the brain of the 5xFAD model (8-month-old, ♂). Bottom: The representative image shows the identification of cells and plaques in the cortex. DAPI+ nuclei: red, 6E10+ amyloid β plaques: blue, Meg3+ neurons: brown, segmented plaques: blue dots. Scale bar: 1 mm. (g) Representative images show different cell types as well as a PU.1low plaque-associated, and PU.1high distal microglia. Macrophage-identifying gene Mrc1: dark blue. Endothelial cell identifying gene Cldn5: orange. T cell identifying genes Cd3e and Cd8a: green. Microglia homeostatic genes Cx3cr1, Tmem119, P2ry12, Csf1r, Hexb, Aif1, Tnf, Trem2, Itgam, C1qa, C1qb, C1qc, Sall1, Selplg: red. Neuron-identifying gene Meg3: pink. Oligodendrocyte-identifying genes Mog, Sox10, and Mag: brown. Lymphoid genes Pdcd1, Cd274, Cd28, Cd72, Tnfsf13b, Cd4: cyan. DAM identifying genes Itgax, Clec7a, Igf1, Cst7, Spp1: yellow. Spi1 (PU.1): white. Amyloid β (6E10): blue. DAPI+ nuclei: gray. Marker genes are shown in Supplementary Table 4. Illustrations in a, c, d and f were created using BioRender (https://biorender.com).
Extended Data Fig. 3 PU.1low plaque-associated microglia express lymphoid genes (related to Fig. 2). (a, b) PU.1low plaque-associated microglia show lymphoid gene expression. Single-cell mRNA sequencing of forebrain microglia from 8-month-old wild-type (2♂) and 5xFAD mice (2♂) from Figs. 1a and 2a, Supplementary Fig. 1, and Supplementary Table 1. (a) Top: Isolation of microglia by fluorescence-activated cell sorting (FACS) followed by single-cell mRNA sequencing. Bottom: The balloon plot shows differentially regulated gene ontologies (GOs, Fisher’s exact test) in the PU.1low and PU.1high cortical disease-associated microglia (DAM) subpopulations. (b) Scatter plot shows the correlation of upregulated genes in cortical PU.1high (red) vs. PU.1low microglia (blue) between two independent experiments (n = 1 mouse/group per experiment) using FindMarkers() analysis in Seurat—Wilcoxon rank sum test. 558 genes are significantly upregulated and 445 genes are significantly downregulated in PU.1low microglia in two independent experiments, while 0 genes are regulated in opposing directions. (c) Plaque-associated, CSF1R-independent PU.1low microglia show lymphoid gene expression. Single-nuclei sequencing of cortical microglia from 6-month-old control (♀) and two CSF1Ri-treated 5xFAD mice (2♂ pooled). Left: Isolation of lineage-traced microglial nuclei from the cortex by fluorescence-activated nuclear sorting (FANS) followed by single-nuclei sequencing. Right: The volcano plot shows the differentially up- (red) and down- (blue) regulated genes in cortical microglia control and CSF1Ri-treated mice from Fig. 1d by FindMarkers() function in Seurat—Wilcoxon rank sum test. Please also see Supplementary Fig. 2 and Supplementary Table 3. (d) PU.1low microglia-expressed lymphoid-associated signaling molecules (ligands, surface receptors, and downstream signaling proteins) in T-cells are shown. Adapted from “T-cell and B-cell signaling” and “TCR downstream signaling” generated by BioRender.com. (e) Heatmap shows log2 (fold change) of selected differentially expressed genes in the PU.1low DAM compared to the PU.1high DAM, two independent experiments: (1) single-cell sequencing of forebrain microglia from wild-type and 5xFAD mice (Figs. 1a, 2a, Extended Data Fig. 3a) and (2) single-nuclei sequencing of cortical microglia from control diet- and CSF1Ri diet-fed 5xFAD mice (Fig. 1d, Extended Data Fig. 3c). The color indicates log2 (fold change). (f) Top: Schematic shows multiplexed error-robust fluorescent in situ hybridization (MERFISH) within the brain of the 5xFAD model (8-month-old, ♂). Bottom: The representative image shows the identification of cells and plaques in the cortex. DAPI+ nuclei: red, 6E10+ amyloid β plaques: blue, Meg3+ neurons: brown, segmented plaques: blue dots. Scale bar: 1 mm. (g) Representative images show different cell types as well as a PU.1low plaque-associated, and PU.1high distal microglia. Macrophage-identifying gene Mrc1: dark blue. Endothelial cell identifying gene Cldn5: orange. T cell identifying genes Cd3e and Cd8a: green. Microglia homeostatic genes Cx3cr1, Tmem119, P2ry12, Csf1r, Hexb, Aif1, Tnf, Trem2, Itgam, C1qa, C1qb, C1qc, Sall1, Selplg: red. Neuron-identifying gene Meg3: pink. Oligodendrocyte-identifying genes Mog, Sox10, and Mag: brown. Lymphoid genes Pdcd1, Cd274, Cd28, Cd72, Tnfsf13b, Cd4: cyan. DAM identifying genes Itgax, Clec7a, Igf1, Cst7, Spp1: yellow. Spi1 (PU.1): white. Amyloid β (6E10): blue. DAPI+ nuclei: gray. Marker genes are shown in Supplementary Table 4. Illustrations in a, c, d and f were created using BioRender (https://biorender.com).CC BY · PubMed Central
Extended Data Fig. 4 Generation of PU.1-low and PU.1-high mice (related to Fig. 3). (a) Gene targeting approach110,111 for the generation of the inducible PU.1-high R26LSL.F-Spi1/+ mice. Brackets indicate the two homology arms mediating recombination in embryonic stem cells. F-Spi1 (green) encodes the FLAG-tagged PU.1 protein sequence, and loxP (red) sites are indicated by arrowheads. Cre-mediated removal of the loxP-flanked STOP codon leads to F-Spi1 expression. Right: Representative genotyping gel images of R26+/+ wild-type, R26LSL.F-Spi1/+ heterozygous, and R26LSL.F-Spi1/LSL.F-Spi1 homozygous mice are shown. Independently repeated >800 times. (b) PU.1 protein expression in PU.1-low, PU.1-wild-type (PU.1-wt), and PU.1-high mice. Representative Western blot (left) and quantification of protein levels (right). Representative fluorescence-activated cell sorting (FACS) plots for microglia isolation are shown in Supplementary Fig. 6b. The upper band in PU.1-high mice indicates the FLAG-tagged PU.1 protein expressed from the R26 locus. The bar graphs show PU.1 level normalized to microglial marker IBA1, IBA1 level normalized to Histone H3 in the cortex of 6-month-old PU.1-low (Cx3cr1CreErt2/+;Spi1fl/+, n = 5 mice, 3♀, 2♂), PU.1-wt (Cx3cr1CreErt2/+, n = 5 mice, 2♀, 3♂), and PU.1-high mice (Cx3cr1CreErt2/+;R26LSL.F-Spi1/+, n = 3 mice, 3♀)—ordinary one-way ANOVA with multiple comparisons. (c) PU.1-low and PU.1-high mice show changes in microglia numbers (lower in PU.1-low) and morphology (increased volume in PU.1-high) in the healthy brain. Representative immunofluorescence images (left) and quantification of microglia morphologies (right). CD11B+ microglia: red, DAPI+ nuclei: blue. The bar graphs show the number of microglia (left) and microglial cell size (volume, right) in the cortex of 6-month-old PU.1-low (n = 4 mice, 1♀, 3♂), PU.1-wt (n = 4 mice, 3♀, 1♂), and PU.1-high mice (n = 4 mice, 3♀, 1♂), as calculated by Imaris—ordinary one-way ANOVA with multiple comparisons. (d-f) PU.1-low and PU.1-high microglia show distinct changes in gene expression in the healthy brain, with PU.1-low inducing lymphoid and PU.1-high inducing pro-inflammatory and stress response patterns. Microglia-specific translating ribosome affinity purification (TRAP) sequencing from 6-month-old PU.1-low (n = 6 mice, 4♀, 2♂), PU.1-wt (n = 6 mice, 3♀, 3♂), and PU.1-high mice (n = 4 mice, 3♀, 1♂). (d) Experimental design (left) and MA plots (right, representing log-ratio (M) on the y-axis and mean average (A) on the x-axis) show gene expression changes (red, upregulated; blue, downregulated by DESeq2) shown in Supplementary Table 6. (e) Representative genome browser views (in Integrative Genomics Viewer, IGV) of normalized read counts of Cd28 ribosome-associated mRNA in PU.1-low, PU.1-wt, and PU.1-high microglia. Numbers in brackets indicate the minimum and maximum values of the y-axis shown for each gene. (f) The balloon plot shows enrichment of gene ontologies (Fisher’s exact test) in differentially regulated genes associated with PU.1 level-dependent and independent gene sets by DESeq2. p-value (color) and combined score (size) are shown. (g-i) PU.1-low microglia show overall microglia-specific chromatin accessibility, with PU.1 expression levels governing distinct accessibility patterns. ATAC sequencing of microglia from PU.1-low, PU.1-wt, and PU.1-high microglia from the brain (n = 3 mice/group) and T-cells from wild-type spleen (n = 2 mice). 9- to 10-month-old ♂ mice were used. 50,000 cells used per sample. (g) Bottom: Isolation of PU.1-low, PU.1-wild-type (PU.1-wt), and PU.1-high microglia from the brain and wild-type T cells from the spleen by FACS, followed by ATAC sequencing. Top: Principal component analysis (PCA) of the 5,000 most differential ATAC sequencing peaks in microglia and T cells (left) or between microglia of the different PU.1 genotypes (right). (h) Heatmap shows all differentially up- and downregulated peaks (called by MACS2) between PU.1-low, PU.1-wt, and PU.1-high microglia identified by DESeq2 (PU.1-low vs. PU.1-wt, PU.1-high vs. PU.1-wt, and PU.1-low vs. PU.1-high). Colors in the main heatmap (left) show scaled read counts (variance stabilizing transformations, vst) per sample. Colors on the side heatmap (right) show significantly up or downregulated peaks between indicated samples shown in Supplementary Table 7. (i) HOMER DNA-motif enrichment analysis of peaks with reduced and increased accessibility in PU.1-low microglia identifies enrichment in peaks containing the PU.1 DNA binding motif (down), and ETS1, MEF2D, and IRF1 (up). Bar graphs with individual data points show mean ± SEM. Illustrations in d and g were created using BioRender (https://biorender.com). Source data
Extended Data Fig. 4 Generation of PU.1-low and PU.1-high mice (related to Fig. 3). (a) Gene targeting approach110,111 for the generation of the inducible PU.1-high R26LSL.F-Spi1/+ mice. Brackets indicate the two homology arms mediating recombination in embryonic stem cells. F-Spi1 (green) encodes the FLAG-tagged PU.1 protein sequence, and loxP (red) sites are indicated by arrowheads. Cre-mediated removal of the loxP-flanked STOP codon leads to F-Spi1 expression. Right: Representative genotyping gel images of R26+/+ wild-type, R26LSL.F-Spi1/+ heterozygous, and R26LSL.F-Spi1/LSL.F-Spi1 homozygous mice are shown. Independently repeated >800 times. (b) PU.1 protein expression in PU.1-low, PU.1-wild-type (PU.1-wt), and PU.1-high mice. Representative Western blot (left) and quantification of protein levels (right). Representative fluorescence-activated cell sorting (FACS) plots for microglia isolation are shown in Supplementary Fig. 6b. The upper band in PU.1-high mice indicates the FLAG-tagged PU.1 protein expressed from the R26 locus. The bar graphs show PU.1 level normalized to microglial marker IBA1, IBA1 level normalized to Histone H3 in the cortex of 6-month-old PU.1-low (Cx3cr1CreErt2/+;Spi1fl/+, n = 5 mice, 3♀, 2♂), PU.1-wt (Cx3cr1CreErt2/+, n = 5 mice, 2♀, 3♂), and PU.1-high mice (Cx3cr1CreErt2/+;R26LSL.F-Spi1/+, n = 3 mice, 3♀)—ordinary one-way ANOVA with multiple comparisons. (c) PU.1-low and PU.1-high mice show changes in microglia numbers (lower in PU.1-low) and morphology (increased volume in PU.1-high) in the healthy brain. Representative immunofluorescence images (left) and quantification of microglia morphologies (right). CD11B+ microglia: red, DAPI+ nuclei: blue. The bar graphs show the number of microglia (left) and microglial cell size (volume, right) in the cortex of 6-month-old PU.1-low (n = 4 mice, 1♀, 3♂), PU.1-wt (n = 4 mice, 3♀, 1♂), and PU.1-high mice (n = 4 mice, 3♀, 1♂), as calculated by Imaris—ordinary one-way ANOVA with multiple comparisons. (d-f) PU.1-low and PU.1-high microglia show distinct changes in gene expression in the healthy brain, with PU.1-low inducing lymphoid and PU.1-high inducing pro-inflammatory and stress response patterns. Microglia-specific translating ribosome affinity purification (TRAP) sequencing from 6-month-old PU.1-low (n = 6 mice, 4♀, 2♂), PU.1-wt (n = 6 mice, 3♀, 3♂), and PU.1-high mice (n = 4 mice, 3♀, 1♂). (d) Experimental design (left) and MA plots (right, representing log-ratio (M) on the y-axis and mean average (A) on the x-axis) show gene expression changes (red, upregulated; blue, downregulated by DESeq2) shown in Supplementary Table 6. (e) Representative genome browser views (in Integrative Genomics Viewer, IGV) of normalized read counts of Cd28 ribosome-associated mRNA in PU.1-low, PU.1-wt, and PU.1-high microglia. Numbers in brackets indicate the minimum and maximum values of the y-axis shown for each gene. (f) The balloon plot shows enrichment of gene ontologies (Fisher’s exact test) in differentially regulated genes associated with PU.1 level-dependent and independent gene sets by DESeq2. p-value (color) and combined score (size) are shown. (g-i) PU.1-low microglia show overall microglia-specific chromatin accessibility, with PU.1 expression levels governing distinct accessibility patterns. ATAC sequencing of microglia from PU.1-low, PU.1-wt, and PU.1-high microglia from the brain (n = 3 mice/group) and T-cells from wild-type spleen (n = 2 mice). 9- to 10-month-old ♂ mice were used. 50,000 cells used per sample. (g) Bottom: Isolation of PU.1-low, PU.1-wild-type (PU.1-wt), and PU.1-high microglia from the brain and wild-type T cells from the spleen by FACS, followed by ATAC sequencing. Top: Principal component analysis (PCA) of the 5,000 most differential ATAC sequencing peaks in microglia and T cells (left) or between microglia of the different PU.1 genotypes (right). (h) Heatmap shows all differentially up- and downregulated peaks (called by MACS2) between PU.1-low, PU.1-wt, and PU.1-high microglia identified by DESeq2 (PU.1-low vs. PU.1-wt, PU.1-high vs. PU.1-wt, and PU.1-low vs. PU.1-high). Colors in the main heatmap (left) show scaled read counts (variance stabilizing transformations, vst) per sample. Colors on the side heatmap (right) show significantly up or downregulated peaks between indicated samples shown in Supplementary Table 7. (i) HOMER DNA-motif enrichment analysis of peaks with reduced and increased accessibility in PU.1-low microglia identifies enrichment in peaks containing the PU.1 DNA binding motif (down), and ETS1, MEF2D, and IRF1 (up). Bar graphs with individual data points show mean ± SEM. Illustrations in d and g were created using BioRender (https://biorender.com). Source dataCC BY · PubMed Central
Extended Data Fig. 5 PU.1 regulates lymphoid gene expression in microglia in vivo and in vitro, in BV2 cells, in primary mouse microglia, and human iMgls (related to Fig. 3). (a) Microglia in vivo: Schematic showing microglia and T cell isolation (left) and Western blot with quantification for CD28 levels (normalized CD28 quantification shown below the blot) in ex vivo acutely isolated microglia from the forebrain of 12-month-old PU.1-low, PU.1-wild-type (PU.1-wt), and PU.1-high mice (n = 3 pooled mice, 1♀, 2♂ each) as well as T cells isolated from the spleen of a wild-type control (♂). Representative FACS plots for microglia isolation are shown in Supplementary Fig. 1c. Representative FACS plots for T cell isolation are shown in Supplementary Fig. 6b. (b-c) BV2 cells: control cells treated with scrambled small interfering (si) RNAs (scramble) and PU.1-knockdown (PU.1-KD) cells treated with siRNAs against Spi1, as well as control cells transfected with empty vector (empty) and PU.1-overexpressor (PU.1-OE) cells transfected with a vector containing the Spi1 transgene from Pimenova et al.39. (b) Representative protein expression analysis using Western blot (left) with experimental design (top) and quantification (right) of CD28 and PU.1 levels in scramble and PU.1-KD cells (n = 4 wells/group), as well as in empty and PU.1-OE cells (n = 3 wells/group) are shown—unpaired two-tailed t-test was used. (c) Experimental design (left) and bar graphs (right) show protein levels (n = 4 wells/group) identified by proteomics analysis—unpaired two-tailed t-test. Differentially expressed proteins are shown in Supplementary Table 8. (d) Primary microglia: Experimental design (left), representative Western blot (middle), and quantification (right) of CD28 and PU.1 levels in primary microglia treated with siRNAs against Spi1 (PU.1) or Cd28 or control (scramble, Scr, n = 3 wells/group)—ordinary one-way ANOVA with multiple comparisons used. (e-f) Human iMgls: (e) Experimental design (top), RNA expression quantification by qPCR (bottom left), and representative protein expression analysis using Western blot (bottom right) of SPI1/PU.1 level in human hematopoietic progenitor cell-(HPC)-derived microglial-like cells (iMgls) treated with siRNAs against SPI1/PU.1 (siSPI1) or control scramble siRNA (siSCR). The bar graph shows qPCR-based quantification of SPI1 normalized to GAPDH in siSCR and siSPI1-targeted iMgls (n = 4 independent experiments)—paired two-tailed t-test. Please see Table 2 for demographic information. (f) Heatmap shows z-scored vsts of the differentially regulated genes in siSCR and siSPI1-targeted iMgls identified by RNA sequencing and DESeq2 (n = 6 independent experiments), shown in Supplementary Table 9. Bar graphs with individual data points show mean ± SEM. Illustrations in a–e were created using BioRender (https://biorender.com). Source data
Extended Data Fig. 5 PU.1 regulates lymphoid gene expression in microglia in vivo and in vitro, in BV2 cells, in primary mouse microglia, and human iMgls (related to Fig. 3). (a) Microglia in vivo: Schematic showing microglia and T cell isolation (left) and Western blot with quantification for CD28 levels (normalized CD28 quantification shown below the blot) in ex vivo acutely isolated microglia from the forebrain of 12-month-old PU.1-low, PU.1-wild-type (PU.1-wt), and PU.1-high mice (n = 3 pooled mice, 1♀, 2♂ each) as well as T cells isolated from the spleen of a wild-type control (♂). Representative FACS plots for microglia isolation are shown in Supplementary Fig. 1c. Representative FACS plots for T cell isolation are shown in Supplementary Fig. 6b. (b-c) BV2 cells: control cells treated with scrambled small interfering (si) RNAs (scramble) and PU.1-knockdown (PU.1-KD) cells treated with siRNAs against Spi1, as well as control cells transfected with empty vector (empty) and PU.1-overexpressor (PU.1-OE) cells transfected with a vector containing the Spi1 transgene from Pimenova et al.39. (b) Representative protein expression analysis using Western blot (left) with experimental design (top) and quantification (right) of CD28 and PU.1 levels in scramble and PU.1-KD cells (n = 4 wells/group), as well as in empty and PU.1-OE cells (n = 3 wells/group) are shown—unpaired two-tailed t-test was used. (c) Experimental design (left) and bar graphs (right) show protein levels (n = 4 wells/group) identified by proteomics analysis—unpaired two-tailed t-test. Differentially expressed proteins are shown in Supplementary Table 8. (d) Primary microglia: Experimental design (left), representative Western blot (middle), and quantification (right) of CD28 and PU.1 levels in primary microglia treated with siRNAs against Spi1 (PU.1) or Cd28 or control (scramble, Scr, n = 3 wells/group)—ordinary one-way ANOVA with multiple comparisons used. (e-f) Human iMgls: (e) Experimental design (top), RNA expression quantification by qPCR (bottom left), and representative protein expression analysis using Western blot (bottom right) of SPI1/PU.1 level in human hematopoietic progenitor cell-(HPC)-derived microglial-like cells (iMgls) treated with siRNAs against SPI1/PU.1 (siSPI1) or control scramble siRNA (siSCR). The bar graph shows qPCR-based quantification of SPI1 normalized to GAPDH in siSCR and siSPI1-targeted iMgls (n = 4 independent experiments)—paired two-tailed t-test. Please see Table 2 for demographic information. (f) Heatmap shows z-scored vsts of the differentially regulated genes in siSCR and siSPI1-targeted iMgls identified by RNA sequencing and DESeq2 (n = 6 independent experiments), shown in Supplementary Table 9. Bar graphs with individual data points show mean ± SEM. Illustrations in a–e were created using BioRender (https://biorender.com). Source dataCC BY · PubMed Central
Extended Data Fig. 6 Downregulation of PU.1 in mouse microglia and BV2 cells is associated with an increase in SYK/PLCγ2 signaling (related to Fig. 3). (a, b) Experimental design (left), representative protein analysis by Western blot (middle), and quantification (right). The bar graph shows SYK and PLCγ2 phosphorylation levels normalized to total SYK and PLCγ2 protein levels—unpaired two-tailed t-test. (a) ex vivo microglia: Microglia acutely isolated from the forebrain of 6-month-old PU.1-wild-type (PU.1-wt, n = 3 mice, 3♀) and PU.1-low (n = 4 mice, 4♀); as well as PU.1-wt (n = 3 mice, 2♀, 1♂), and PU.1-high mice (n = 3 mice, 1♀, 2♂). Protein extracts equivalent to 120,000 microglia were loaded for Western blot per sample. Representative FACS plots for microglia isolation are shown in Supplementary Fig. 6b. (b) BV2 cells: PU.1-knockdown (PU.1-KD) cells treated with small interfering (si) RNAs against Spi1 with associated control cells treated with scrambled siRNAs (scramble), as well as PU.1-overexpressor (PU.1-OE) cells transfected with vector containing Spi1 transgene with associated control cells transfected with empty vector (empty) from Pimenova et al.39. n = 7 wells in scramble and PU.1-KD; n = 6 wells in empty, and n = 5 wells in PU.1-OE—pooled from two independently performed experiments. Bar graphs with individual data points show mean ± SEM. Illustrations in a and b were created using BioRender (https://biorender.com). Source data
Extended Data Fig. 6 Downregulation of PU.1 in mouse microglia and BV2 cells is associated with an increase in SYK/PLCγ2 signaling (related to Fig. 3). (a, b) Experimental design (left), representative protein analysis by Western blot (middle), and quantification (right). The bar graph shows SYK and PLCγ2 phosphorylation levels normalized to total SYK and PLCγ2 protein levels—unpaired two-tailed t-test. (a) ex vivo microglia: Microglia acutely isolated from the forebrain of 6-month-old PU.1-wild-type (PU.1-wt, n = 3 mice, 3♀) and PU.1-low (n = 4 mice, 4♀); as well as PU.1-wt (n = 3 mice, 2♀, 1♂), and PU.1-high mice (n = 3 mice, 1♀, 2♂). Protein extracts equivalent to 120,000 microglia were loaded for Western blot per sample. Representative FACS plots for microglia isolation are shown in Supplementary Fig. 6b. (b) BV2 cells: PU.1-knockdown (PU.1-KD) cells treated with small interfering (si) RNAs against Spi1 with associated control cells treated with scrambled siRNAs (scramble), as well as PU.1-overexpressor (PU.1-OE) cells transfected with vector containing Spi1 transgene with associated control cells transfected with empty vector (empty) from Pimenova et al.39. n = 7 wells in scramble and PU.1-KD; n = 6 wells in empty, and n = 5 wells in PU.1-OE—pooled from two independently performed experiments. Bar graphs with individual data points show mean ± SEM. Illustrations in a and b were created using BioRender (https://biorender.com). Source dataCC BY · PubMed Central
Extended Data Fig. 7 PU.1 levels govern microglia transcriptional responses in 5xFAD mice: microglia single nuclei analysis (related to Fig. 4). (a-f) Single-nuclei mRNA sequencing of cortical microglia from a 6-month-old 5xFAD (♀), 5xFAD;PU.1-low (♂), and 5xFAD;PU.1-high mouse (♂). (a) Left: Isolation of lineage-traced microglial nuclei by fluorescence-activated nuclear sorting (FANS) followed by single-nuclei sequencing of cortical microglia from 6-month-old control, 5xFAD, and 5xFAD;PU.1-low mice. Please see Supplementary Fig. 2b for the gating strategy. Right: Uniform manifold approximation and projection (UMAP) visualizations show microglia assigned into clusters that are manually annotated based on the marker genes. (b) The dot plot shows the scaled average expression per cluster (color) and the percentage of cells (size) from each DAM subcluster. (c) UMAP visualizations show microglia assigned into clusters that are manually annotated based on the marker genes shown in Supplementary Table 10. (d) The bar graphs show the change in the percentage of different microglia subpopulations in the three genotypes. (e, f) Feature plots show the expression of indicated lymphoid (e) and interferon genes (f) and the distribution of cells expressing the genes. Illustrations in a were created using BioRender (https://biorender.com). Source data
Extended Data Fig. 7 PU.1 levels govern microglia transcriptional responses in 5xFAD mice: microglia single nuclei analysis (related to Fig. 4). (a-f) Single-nuclei mRNA sequencing of cortical microglia from a 6-month-old 5xFAD (♀), 5xFAD;PU.1-low (♂), and 5xFAD;PU.1-high mouse (♂). (a) Left: Isolation of lineage-traced microglial nuclei by fluorescence-activated nuclear sorting (FANS) followed by single-nuclei sequencing of cortical microglia from 6-month-old control, 5xFAD, and 5xFAD;PU.1-low mice. Please see Supplementary Fig. 2b for the gating strategy. Right: Uniform manifold approximation and projection (UMAP) visualizations show microglia assigned into clusters that are manually annotated based on the marker genes. (b) The dot plot shows the scaled average expression per cluster (color) and the percentage of cells (size) from each DAM subcluster. (c) UMAP visualizations show microglia assigned into clusters that are manually annotated based on the marker genes shown in Supplementary Table 10. (d) The bar graphs show the change in the percentage of different microglia subpopulations in the three genotypes. (e, f) Feature plots show the expression of indicated lymphoid (e) and interferon genes (f) and the distribution of cells expressing the genes. Illustrations in a were created using BioRender (https://biorender.com). Source dataCC BY · PubMed Central
Extended Data Fig. 8 PU.1 levels regulate microglia transcriptional responses in 5xFAD mice: microglia-specific TRAP analysis (related to Fig. 4). (a-c) Microglia-specific translating ribosome affinity purification (TRAP) sequencing from 6-month-old control (n = 6 mice, 3♀, 3♂), 5xFAD (n = 6 mice, 3♀, 3♂), and 5xFADPU.1-low mice (1♀, 3♂). (a) Experimental design (left) and MA plots (right, representing log-ratio (M) on the y-axis and mean average (A) on the x-axis) show gene expression changes (red, upregulated; blue, downregulated) by DESeq2, shown in Supplementary Table 11. (b) Heatmap shows z-scored variance stabilizing transformations (vsts, color) of genes selected by overlapping and non-overlapping genes between pairwise comparisons (5xFAD;PU.1-low vs. 5xFAD and 5xFAD vs. control). (c) The balloon plot shows significantly enriched gene ontologies (Fisher’s exact test) associated with significantly regulated PU.1 level-dependent and independent gene sets (DESeq2). p-value (color) and combined score (size) are shown. Illustrations in a were created using BioRender (https://biorender.com).
Extended Data Fig. 8 PU.1 levels regulate microglia transcriptional responses in 5xFAD mice: microglia-specific TRAP analysis (related to Fig. 4). (a-c) Microglia-specific translating ribosome affinity purification (TRAP) sequencing from 6-month-old control (n = 6 mice, 3♀, 3♂), 5xFAD (n = 6 mice, 3♀, 3♂), and 5xFADPU.1-low mice (1♀, 3♂). (a) Experimental design (left) and MA plots (right, representing log-ratio (M) on the y-axis and mean average (A) on the x-axis) show gene expression changes (red, upregulated; blue, downregulated) by DESeq2, shown in Supplementary Table 11. (b) Heatmap shows z-scored variance stabilizing transformations (vsts, color) of genes selected by overlapping and non-overlapping genes between pairwise comparisons (5xFAD;PU.1-low vs. 5xFAD and 5xFAD vs. control). (c) The balloon plot shows significantly enriched gene ontologies (Fisher’s exact test) associated with significantly regulated PU.1 level-dependent and independent gene sets (DESeq2). p-value (color) and combined score (size) are shown. Illustrations in a were created using BioRender (https://biorender.com).CC BY · PubMed Central
Extended Data Fig. 9 PU.1low microglia display neuroprotective phenotypes and reduce amyloid load and Tau spreading in the brain of 5xFAD mice (related to Fig. 4). (a) 5xFAD-PU.1-low mice show similar numbers of distal microglia. The bar graphs show the density of distal and total microglia in the cortex of 6-month-old 5xFAD (n = 6 mice, 3♀, 3♂) and 5xFAD-PU.1-low mice (n = 5 mice, 3♀, 2♂)—unpaired two-tailed t-test. (b) CD28 protein expression. Western blot shows CD28 levels in ex vivo microglia acutely isolated from the forebrain of 12-month-old control (n = 3 pooled mice, 1♀, 2♂) and 5xFAD; PU.1-low mice (n = 3 pooled mice,1♀, 2♂). Normalized CD28 quantification is shown below the blot. (c-e) Microglia in 5xFAD-PU.1-low mice show reduced neutral lipid droplet load and complement C1Q expression. (c) Western blot (left) with quantification (right). The bar graph shows the C1QA level normalized to TUBB3 per sample in the hippocampi of 6-month-old control (n = 6 mice, 6♂), 5xFAD (n = 4 mice, 4♂), and 5xFAD-PU.1-low mice (n = 3 mice, 3♂)—ordinary one-way ANOVA with multiple comparisons was used. (d-e) Representative immunofluorescence images (left) with quantification (right) from the cortex of 6-month-old mice—unpaired two-tailed t-test. Scale bar: 10 μm. (d) CLEC7A+ disease-associated microglia: green, TMEM119+ homeostatic microglia: cyan, C1QA: red. The bar graph shows the percent C1QA in the cortex of 5xFAD (n = 8 mice, 5♀, 3♂) and 5xFAD-PU.1-low mice (n = 6 mice, 4♀, 2♂). (e) CD11B+ microglia: red, BODIPY 493/503: green, ThioS+ dense-core plaques: blue. The bar graph shows the BODIPY+ volume within microglia normalized to microglia volume in 5xFAD (n = 3 mice, 2♀, 1♂) and 5xFAD-PU.1-low mice (n = 4 mice, 2♀, 2♂). (f) 5xFAD-PU.1-low mice show a reduced soluble amyloid load. The bar graphs show the percent area covered by 6E10+ soluble Aβ or ThioS+ dense-core plaques in the subiculum from 6-month-old 5xFAD (n = 7 mice, 2♀, 5♂) and 5xFAD; PU.1-low mice (n = 5 mice, 2♀, 3♂)—unpaired two-tailed t-test. Representative images are shown in Fig. 4f. (g) 5xFAD-PU.1-low mice show a reduced spread of p-Tau following intracerebral injection of human pathogenic Tau isolated from the brain (schematic, top). Bar graphs (bottom) show the percent AT8+ phosphorylated Tau (p-Tau) area and HJ3.4+ amyloid β in the indicated brain regions of 10-month-old 5xFAD (n = 12 mice, 7♀, 5♂) and 5xFAD; PU.1-low mice (n = 9 mice, 4♀, 5♂) (h) LTP: The line graph shows the time course of long-term synaptic plasticity (LTP) quantified as a percentage change in excitatory postsynaptic potential (EPSP) slope from baseline (10-min period of a pre-and 60-min period of a post-100 Hz stimulation) in CA1 neurons of the hippocampus in 9-10-month-old control (n = 12 slices from 5 mice, 3♀, 2♂), 5xFAD (n = 13 slices from 6 mice, 4♀, 2♂), and 5xFAD; PU.1-low mice (n = 11 slices from 5 mice, 1♀, 4♂)—two-way ANOVA with repeated measures, Ptime = 0.0183, Pgenotype = 0.1031. (i) Open field analysis: Experimental design (left) and quantification (right) of the open field experiment. The area identified as “center” is marked by a red square. The bar graph shows the total distance traveled (m), time spent in the center (seconds), and the periphery (seconds) in a 10-min-long open field assay from 9–10 month-old control (n = 15 mice, 9♀, 6♂), 5xFAD (n = 12 mice, 6♀, 6♂), and 5xFAD; PU.1-low (n = 13 mice, 7♀, 6♂)—ordinary one-way ANOVA with multiple comparisons. (j) Novel object recognition: The schematic (top) and quantification of the novel object recognition experiment in a 3-room box. The bar graphs show the distance traveled during 20-min habituation, 10-min training, and 10-min testing; the total time spent sniffing any object during training and testing; and the “recognition score” calculated as the percent time spent sniffing the constant object to the total time spent sniffing either object during training from 9-10-month-old control (n = 10 mice, 5♀, 5♂), 5xFAD (n = 11 mice, 6♀, 5♂), and 5xFAD; PU.1-low mice (n = 15 mice, 7♀, 8♂)—ordinary one-way ANOVA with multiple comparisons. No sex differences are observed in behavioral analyses, and trends between genotypes were preserved in sex-specific analyses. Bar graphs with individual data points show mean ± SEM. Illustrations in g, i and j were created using BioRender (https://biorender.com). Source data
Extended Data Fig. 9 PU.1low microglia display neuroprotective phenotypes and reduce amyloid load and Tau spreading in the brain of 5xFAD mice (related to Fig. 4). (a) 5xFAD-PU.1-low mice show similar numbers of distal microglia. The bar graphs show the density of distal and total microglia in the cortex of 6-month-old 5xFAD (n = 6 mice, 3♀, 3♂) and 5xFAD-PU.1-low mice (n = 5 mice, 3♀, 2♂)—unpaired two-tailed t-test. (b) CD28 protein expression. Western blot shows CD28 levels in ex vivo microglia acutely isolated from the forebrain of 12-month-old control (n = 3 pooled mice, 1♀, 2♂) and 5xFAD; PU.1-low mice (n = 3 pooled mice,1♀, 2♂). Normalized CD28 quantification is shown below the blot. (c-e) Microglia in 5xFAD-PU.1-low mice show reduced neutral lipid droplet load and complement C1Q expression. (c) Western blot (left) with quantification (right). The bar graph shows the C1QA level normalized to TUBB3 per sample in the hippocampi of 6-month-old control (n = 6 mice, 6♂), 5xFAD (n = 4 mice, 4♂), and 5xFAD-PU.1-low mice (n = 3 mice, 3♂)—ordinary one-way ANOVA with multiple comparisons was used. (d-e) Representative immunofluorescence images (left) with quantification (right) from the cortex of 6-month-old mice—unpaired two-tailed t-test. Scale bar: 10 μm. (d) CLEC7A+ disease-associated microglia: green, TMEM119+ homeostatic microglia: cyan, C1QA: red. The bar graph shows the percent C1QA in the cortex of 5xFAD (n = 8 mice, 5♀, 3♂) and 5xFAD-PU.1-low mice (n = 6 mice, 4♀, 2♂). (e) CD11B+ microglia: red, BODIPY 493/503: green, ThioS+ dense-core plaques: blue. The bar graph shows the BODIPY+ volume within microglia normalized to microglia volume in 5xFAD (n = 3 mice, 2♀, 1♂) and 5xFAD-PU.1-low mice (n = 4 mice, 2♀, 2♂). (f) 5xFAD-PU.1-low mice show a reduced soluble amyloid load. The bar graphs show the percent area covered by 6E10+ soluble Aβ or ThioS+ dense-core plaques in the subiculum from 6-month-old 5xFAD (n = 7 mice, 2♀, 5♂) and 5xFAD; PU.1-low mice (n = 5 mice, 2♀, 3♂)—unpaired two-tailed t-test. Representative images are shown in Fig. 4f. (g) 5xFAD-PU.1-low mice show a reduced spread of p-Tau following intracerebral injection of human pathogenic Tau isolated from the brain (schematic, top). Bar graphs (bottom) show the percent AT8+ phosphorylated Tau (p-Tau) area and HJ3.4+ amyloid β in the indicated brain regions of 10-month-old 5xFAD (n = 12 mice, 7♀, 5♂) and 5xFAD; PU.1-low mice (n = 9 mice, 4♀, 5♂) (h) LTP: The line graph shows the time course of long-term synaptic plasticity (LTP) quantified as a percentage change in excitatory postsynaptic potential (EPSP) slope from baseline (10-min period of a pre-and 60-min period of a post-100 Hz stimulation) in CA1 neurons of the hippocampus in 9-10-month-old control (n = 12 slices from 5 mice, 3♀, 2♂), 5xFAD (n = 13 slices from 6 mice, 4♀, 2♂), and 5xFAD; PU.1-low mice (n = 11 slices from 5 mice, 1♀, 4♂)—two-way ANOVA with repeated measures, Ptime = 0.0183, Pgenotype = 0.1031. (i) Open field analysis: Experimental design (left) and quantification (right) of the open field experiment. The area identified as “center” is marked by a red square. The bar graph shows the total distance traveled (m), time spent in the center (seconds), and the periphery (seconds) in a 10-min-long open field assay from 9–10 month-old control (n = 15 mice, 9♀, 6♂), 5xFAD (n = 12 mice, 6♀, 6♂), and 5xFAD; PU.1-low (n = 13 mice, 7♀, 6♂)—ordinary one-way ANOVA with multiple comparisons. (j) Novel object recognition: The schematic (top) and quantification of the novel object recognition experiment in a 3-room box. The bar graphs show the distance traveled during 20-min habituation, 10-min training, and 10-min testing; the total time spent sniffing any object during training and testing; and the “recognition score” calculated as the percent time spent sniffing the constant object to the total time spent sniffing either object during training from 9-10-month-old control (n = 10 mice, 5♀, 5♂), 5xFAD (n = 11 mice, 6♀, 5♂), and 5xFAD; PU.1-low mice (n = 15 mice, 7♀, 8♂)—ordinary one-way ANOVA with multiple comparisons. No sex differences are observed in behavioral analyses, and trends between genotypes were preserved in sex-specific analyses. Bar graphs with individual data points show mean ± SEM. Illustrations in g, i and j were created using BioRender (https://biorender.com). Source dataCC BY · PubMed Central
Extended Data Fig. 10 Plaque-associated signals induce microglial CD28 RNA and protein expression (related to Fig. 5). (a-b) Liposomes induce microglial CD28 RNA and protein expression in primary microglia in vitro: (a) Experimental design (left), qPCR quantification (middle left), representative Western blot (middle right), and Western blot quantification (right) of CD28 levels in primary microglia treated with 0.1 mg/ml liposomes for 3 days. The first bar graph shows Cd28 normalized to Gapdh in primary microglia treated with vehicle (n = 4 wells) or liposomes (n = 5 wells)—unpaired two-tailed t-test. The second bar graph shows CD28 levels normalized to ACTB in primary microglia treated with vehicle or liposomes (n = 3 wells/group)—unpaired two-tailed t-test. (b) Representative immunofluorescence images (left) with quantification (right) of CD28 in primary microglia treated with 0.1 mg/ml liposomes for 3 days. IBA1+ microglia: green, CD28: red, DAPI+ nuclei: blue. Scale bar: 20 μm. The bar graphs show normalized integrated CD28 intensity in liposome-treated (control: n = 6 wells, liposome: n = 7 wells) primary microglia—unpaired t-test. (c-d) The number of plaque-associated CD28+ cells increases with amyloid disease progression. (c) Cd28 RNA: Feature plots (right) show Cd28 expression in uniform manifold approximation and projection (UMAP) of single-cell sequencing of forebrain microglia from 3- and 8-month-old wild-type and 5xFAD mice (n = 2 mice/group except control at 3 months: n = 1 mouse, all ♂). (d) CD28 protein: Representative immunofluorescence images (left) with quantification (right) of CD28+ microglia in the subiculum. IBA1+ microglia: green, CD28: red, ThioS+ dense-core plaques: blue, DRAQ5+ nuclei: gray. Scale bar: 20 μm. The bar graphs show the fraction of CD28 + IBA1+ microglia in the subiculum of 3-month-old (n = 4 mice, 2♀, 2♂), 6-month-old (n = 4 mice, 1♀, 3♂), and 12-month-old (n = 4 mice, 3♀, 1♂) 5xFAD mice—ordinary one-way ANOVA with multiple comparisons. Bar graphs with individual data points show mean ± SEM. Illustrations in a were created using BioRender (https://biorender.com). Source data
Extended Data Fig. 10 Plaque-associated signals induce microglial CD28 RNA and protein expression (related to Fig. 5). (a-b) Liposomes induce microglial CD28 RNA and protein expression in primary microglia in vitro: (a) Experimental design (left), qPCR quantification (middle left), representative Western blot (middle right), and Western blot quantification (right) of CD28 levels in primary microglia treated with 0.1 mg/ml liposomes for 3 days. The first bar graph shows Cd28 normalized to Gapdh in primary microglia treated with vehicle (n = 4 wells) or liposomes (n = 5 wells)—unpaired two-tailed t-test. The second bar graph shows CD28 levels normalized to ACTB in primary microglia treated with vehicle or liposomes (n = 3 wells/group)—unpaired two-tailed t-test. (b) Representative immunofluorescence images (left) with quantification (right) of CD28 in primary microglia treated with 0.1 mg/ml liposomes for 3 days. IBA1+ microglia: green, CD28: red, DAPI+ nuclei: blue. Scale bar: 20 μm. The bar graphs show normalized integrated CD28 intensity in liposome-treated (control: n = 6 wells, liposome: n = 7 wells) primary microglia—unpaired t-test. (c-d) The number of plaque-associated CD28+ cells increases with amyloid disease progression. (c) Cd28 RNA: Feature plots (right) show Cd28 expression in uniform manifold approximation and projection (UMAP) of single-cell sequencing of forebrain microglia from 3- and 8-month-old wild-type and 5xFAD mice (n = 2 mice/group except control at 3 months: n = 1 mouse, all ♂). (d) CD28 protein: Representative immunofluorescence images (left) with quantification (right) of CD28+ microglia in the subiculum. IBA1+ microglia: green, CD28: red, ThioS+ dense-core plaques: blue, DRAQ5+ nuclei: gray. Scale bar: 20 μm. The bar graphs show the fraction of CD28 + IBA1+ microglia in the subiculum of 3-month-old (n = 4 mice, 2♀, 2♂), 6-month-old (n = 4 mice, 1♀, 3♂), and 12-month-old (n = 4 mice, 3♀, 1♂) 5xFAD mice—ordinary one-way ANOVA with multiple comparisons. Bar graphs with individual data points show mean ± SEM. Illustrations in a were created using BioRender (https://biorender.com). Source dataCC BY · PubMed Central
Extended Data Fig. 11 Microglial CD28 regulates amyloid burden in the brain of 5xFAD mice (related to Fig. 5). (a) Microglial Cd28 ablation in the healthy brain has no impact on microglia numbers or morphology. The experimental design (left), representative immunofluorescence images (middle), and quantification (right) of microglia in the cortex of CD28-KO mice. CD11B+ microglia: green, DAPI+ nuclei: blue. Scale bar: 20 μm. The bar graphs show the total microglia numbers per image (left) and microglial cell volume (right, μm3) in the subiculum of 3-month-old control (n = 5 mice, 1♀, 4♂) and CD28-KO mice (Cx3cr1CreErt2/+(Litt); Cd28fl/fl, n = 3 mice, 3♂)—unpaired two-tailed t-test. (b) Microglia-specific Cd28 ablation in 5xFAD mice prevents plaque-induced Cd28 expression. Experimental design (left), representative RNA in situ hybridization with immunofluorescence images (middle), and quantification (right) of Cd28+ microglia. CD11B+ microglia: green, Cd28 mRNA: red, DAPI+ nuclei: blue. Scale bar: 5 μm. The bar graph shows the percent of Cd28 + IBA1+ microglia in the subiculum of 6-month-old 5xFAD (n = 3 mice, 1♀, 2♂) and 5xFAD;CD28-KO mice (n = 3 mice, 2♀, 1♂)—unpaired two-tailed t-test. (c) Microglial Cd28 ablation in 5xFAD mice does not affect microglia numbers. Representative immunofluorescence images (left) with quantification (right) of microglia numbers. IBA1+ microglia: green, ThioS+ dense-core plaques: blue, DRAQ5+ nuclei: gray. The bar graph shows the microglia numbers in the dentate gyrus of 3- and 6-month-old 5xFAD (n = 6 mice at 3 months, 1♀, 5♂ and n = 5 mice at 6 months, 2♀, 3♂) and 5xFAD; CD28-KO mice (n = 5 mice at 3 months, 2♀, 3♂ and n = 7 mice at 6 months, 4♀, 3♂)—two-way ANOVA with multiple comparisons. Pgenotype = 0.3913, Page < 0.0001 (d) Microglial Cd28 ablation in 5xFAD mice alters plaque load and composition. Representative immunofluorescence images (left) with quantification (right) of plaque load in the subiculum. 6E10+ amyloid beta: red, ThioS+ dense-core plaques: blue. The bar graphs show 6E10/ThioS ratio, total 6E10 level, and total ThioS level, number of ThioS+ plaques, and volume of ThioS+ plaques in the dentate gyrus of 3- and 6-month-old 5xFAD (n = 6 mice at 3 months, 1♀, 5♂ and n = 5 mice at 6 months, 2♀, 3♂) and 5xFAD; CD28-KO mice (n = 5 mice at 3 months, 2♀, 3♂ and n = 7 mice at 6 months, 4♀, 3♂)—two-way ANOVA was used (ThioS: Pgenotype = 0.3077, Page = 0.0005; 6E10: Pgenotype = 0.6140, Page = 0.0024; ThioS/6E10: Pgenotype = 0.1121, Page = 0.0271; average ThioS plaque numbers per image per mouse: Pgenotype = 0.9603, Page < 0.0001; ThioS volume per image per animal: Pgenotype = 0.0273, Page < 0.0001). (e) Single-cell sequencing of CD45low CD11B+ forebrain microglia from 3- and 6-month-old wild-type, 5xFAD, and 5xFAD;CD28-KO mice (control: n = 1 at 3 and 6 months, 5xFAD and 5xFAD;CD28-KO: n = 2 mice at 3 months, 5xFAD and 5xFAD;CD28-KO: n = 1 mice at 6 months, all ♂). Left: Microglia isolation strategy for single-cell sequencing. Right: Volcano plot shows significantly up- (red) and downregulated (blue) genes between all microglia from 5xFAD and 5xFAD;CD28-KO mice identified by DESeq2 in pseudobulk analysis using FindMarkers() function using Wilcoxon Rank Sum test in Seurat, shown in Supplementary Table 12. (f) Representative RNA in situ hybridization with immunofluorescence images (left) with quantification (right) of Mx1 levels in microglia. IBA1+ microglia: green, Mx1 RNA: red, DAPI+ nuclei: blue. Scale bar: 10 μm. Dotted circles indicate microglial nuclei, some of which are shown in higher magnifications on the bottom. The bar graphs show the fraction of Mx1 + , IBA1+ microglia in the subiculum of 3-month-old wild-type (n = 5 mice, 1♀, 4♂), CD28-KO (n = 3 mice, 3♂), 5xFAD (n = 7 mice, 2♀, 5♂) and 5xFAD; CD28-KO mice (n = 5 mice, 2♀, 3♂)—two-way ANOVA was used Pdisease = 0.0004, Pgenotype = 0.1004—and 6-month-old 5xFAD (n = 6 mice, 3♀, 3♂) 5xFAD; CD28-KO (n = 7 mice, 4♀, 3♂)—unpaired two-tailed t test. Bar graphs with individual data points show mean ± SEM. Illustrations in a, b and e were created using BioRender (https://biorender.com). Source data
Extended Data Fig. 11 Microglial CD28 regulates amyloid burden in the brain of 5xFAD mice (related to Fig. 5). (a) Microglial Cd28 ablation in the healthy brain has no impact on microglia numbers or morphology. The experimental design (left), representative immunofluorescence images (middle), and quantification (right) of microglia in the cortex of CD28-KO mice. CD11B+ microglia: green, DAPI+ nuclei: blue. Scale bar: 20 μm. The bar graphs show the total microglia numbers per image (left) and microglial cell volume (right, μm3) in the subiculum of 3-month-old control (n = 5 mice, 1♀, 4♂) and CD28-KO mice (Cx3cr1CreErt2/+(Litt); Cd28fl/fl, n = 3 mice, 3♂)—unpaired two-tailed t-test. (b) Microglia-specific Cd28 ablation in 5xFAD mice prevents plaque-induced Cd28 expression. Experimental design (left), representative RNA in situ hybridization with immunofluorescence images (middle), and quantification (right) of Cd28+ microglia. CD11B+ microglia: green, Cd28 mRNA: red, DAPI+ nuclei: blue. Scale bar: 5 μm. The bar graph shows the percent of Cd28 + IBA1+ microglia in the subiculum of 6-month-old 5xFAD (n = 3 mice, 1♀, 2♂) and 5xFAD;CD28-KO mice (n = 3 mice, 2♀, 1♂)—unpaired two-tailed t-test. (c) Microglial Cd28 ablation in 5xFAD mice does not affect microglia numbers. Representative immunofluorescence images (left) with quantification (right) of microglia numbers. IBA1+ microglia: green, ThioS+ dense-core plaques: blue, DRAQ5+ nuclei: gray. The bar graph shows the microglia numbers in the dentate gyrus of 3- and 6-month-old 5xFAD (n = 6 mice at 3 months, 1♀, 5♂ and n = 5 mice at 6 months, 2♀, 3♂) and 5xFAD; CD28-KO mice (n = 5 mice at 3 months, 2♀, 3♂ and n = 7 mice at 6 months, 4♀, 3♂)—two-way ANOVA with multiple comparisons. Pgenotype = 0.3913, Page < 0.0001 (d) Microglial Cd28 ablation in 5xFAD mice alters plaque load and composition. Representative immunofluorescence images (left) with quantification (right) of plaque load in the subiculum. 6E10+ amyloid beta: red, ThioS+ dense-core plaques: blue. The bar graphs show 6E10/ThioS ratio, total 6E10 level, and total ThioS level, number of ThioS+ plaques, and volume of ThioS+ plaques in the dentate gyrus of 3- and 6-month-old 5xFAD (n = 6 mice at 3 months, 1♀, 5♂ and n = 5 mice at 6 months, 2♀, 3♂) and 5xFAD; CD28-KO mice (n = 5 mice at 3 months, 2♀, 3♂ and n = 7 mice at 6 months, 4♀, 3♂)—two-way ANOVA was used (ThioS: Pgenotype = 0.3077, Page = 0.0005; 6E10: Pgenotype = 0.6140, Page = 0.0024; ThioS/6E10: Pgenotype = 0.1121, Page = 0.0271; average ThioS plaque numbers per image per mouse: Pgenotype = 0.9603, Page < 0.0001; ThioS volume per image per animal: Pgenotype = 0.0273, Page < 0.0001). (e) Single-cell sequencing of CD45low CD11B+ forebrain microglia from 3- and 6-month-old wild-type, 5xFAD, and 5xFAD;CD28-KO mice (control: n = 1 at 3 and 6 months, 5xFAD and 5xFAD;CD28-KO: n = 2 mice at 3 months, 5xFAD and 5xFAD;CD28-KO: n = 1 mice at 6 months, all ♂). Left: Microglia isolation strategy for single-cell sequencing. Right: Volcano plot shows significantly up- (red) and downregulated (blue) genes between all microglia from 5xFAD and 5xFAD;CD28-KO mice identified by DESeq2 in pseudobulk analysis using FindMarkers() function using Wilcoxon Rank Sum test in Seurat, shown in Supplementary Table 12. (f) Representative RNA in situ hybridization with immunofluorescence images (left) with quantification (right) of Mx1 levels in microglia. IBA1+ microglia: green, Mx1 RNA: red, DAPI+ nuclei: blue. Scale bar: 10 μm. Dotted circles indicate microglial nuclei, some of which are shown in higher magnifications on the bottom. The bar graphs show the fraction of Mx1 + , IBA1+ microglia in the subiculum of 3-month-old wild-type (n = 5 mice, 1♀, 4♂), CD28-KO (n = 3 mice, 3♂), 5xFAD (n = 7 mice, 2♀, 5♂) and 5xFAD; CD28-KO mice (n = 5 mice, 2♀, 3♂)—two-way ANOVA was used Pdisease = 0.0004, Pgenotype = 0.1004—and 6-month-old 5xFAD (n = 6 mice, 3♀, 3♂) 5xFAD; CD28-KO (n = 7 mice, 4♀, 3♂)—unpaired two-tailed t test. Bar graphs with individual data points show mean ± SEM. Illustrations in a, b and e were created using BioRender (https://biorender.com). Source dataCC BY · PubMed Central
Extended Data Fig. 12 The neuroprotective PU.1low, lymphoid+/CD28+ microglia subpopulation expands in TIM-3-KO mice (related to Fig. 5). (a-d) Single-cell sequencing of human microglia from Kimura et al.49. (a) Uniform manifold approximation and projection (UMAP) visualizations (left). (b, c) Dot plot (b) and feature plots (c) show the expression of Cd28, Spi1, and selected PU.1low microglia expressed lymphoid genes (Cd28, Pdcd1, Cd274, Sox5, Dkk2, Prkcq, Btla) in UMAP visualizations. Marker genes for subsets are shown in Supplementary Table 13. (d) The bar graphs show the change in the percentage of microglia comprising the lymphoid + /Cd28+ populations (cluster 14) in 5xFAD (1%) and 5xFAD, TIM-3-KO mice (3%).
Extended Data Fig. 12 The neuroprotective PU.1low, lymphoid+/CD28+ microglia subpopulation expands in TIM-3-KO mice (related to Fig. 5). (a-d) Single-cell sequencing of human microglia from Kimura et al.49. (a) Uniform manifold approximation and projection (UMAP) visualizations (left). (b, c) Dot plot (b) and feature plots (c) show the expression of Cd28, Spi1, and selected PU.1low microglia expressed lymphoid genes (Cd28, Pdcd1, Cd274, Sox5, Dkk2, Prkcq, Btla) in UMAP visualizations. Marker genes for subsets are shown in Supplementary Table 13. (d) The bar graphs show the change in the percentage of microglia comprising the lymphoid + /Cd28+ populations (cluster 14) in 5xFAD (1%) and 5xFAD, TIM-3-KO mice (3%).CC BY · PubMed Central

Verification

Claims the paper makes

25 Valid · 2 Not valid · 1 Not checked · 88 Own results and interpretations

35 more claims · 7 with their cited sources checked

Create a free account to see all 116 claims and to verify the 1 reference still open.

See all claims — free account
  • Probably not a substantial claim. 59
  • It may state two separate claims as one. 31
  • The quote only partly states this claim. 1

Create a free account to see these claims.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

18 more claims about this paper

Create a free account to see every claim other papers make about this one.

See all claims — free account

Citations

Cites

Show all 105 references

Research works printed but not linked: 1

  1. Samuels, J. D., Lukens, J. R. & Price, R. J. Emerging roles for ITAM and ITIM receptor signaling in microglial biology and Alzheimer’s disease-related amyloidosis. J. Neurochem. 168, 3558–3573 (2024).

Cited by

See all 32

Metadata

Authors
Ayata, P., Crowley, J. M., Challman, M. F., Sahasrabuddhe, V., Gratuze, M., Werneburg, S., Ribeiro, D., Hays, E. C., Durán-Laforet, V., Faust, T. E., Hwang, P., Mendes Lopes, F., Nikopoulou, C., Buchholz, S., Murphy, R. E., Mei, T., Pimenova, A. A., Romero-Molina, C., Garretti, F., Patel, T. A., De Sanctis, C., Ramirez Jimenez, A. V., Crow, M., Weiss, F. D., Ulrich, J. D., Marcora, E., Murray, J. W., Meissner, F., Beyer, A., Hasson, D., Crary, J. F., Schafer, D. P., Holtzman, D. M., Goate, A. M., Tarakhovsky, A., & Schaefer, A.
Affiliations
  • The Graduate Center, CUNY (US)
  • Icahn School of Medicine at Mount Sinai (US)
  • Centre National de la Recherche Scientifique (FR)
  • University of Massachusetts Chan Medical School (US)
  • University of Cologne (DE)
  • Genentech (US)
  • University of Bonn (DE)
  • Washington University in St. Louis (US)
  • Columbia University Irving Medical Center (US)
  • Tisch Cancer Institute (US)
  • Rockefeller University (US)
Corresponding author
  • Ayata, Pinar
  • Schaefer, Anne
Venue
Nature
Published
2025-11-05
Volume
648
Issue
8092
Pages
157-165
ISSN
  • 0028-0836 (ISSN-L)
  • 1476-4687
OpenAlex
W4415929059
Cited by
32
References
105
Citations by year
2026: 31 · 2025: 1
FWCI
11.42
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • Digital Commons@Becker (Washington University School of Medicine) (submittedVersion)
  • HAL AMU (submittedVersion, other-oa)
Funders
  • Max-Planck-Gesellschaft
  • National Institutes of Health (1s10od026880-01, 1s10od030463-01, UL1TR004419)
  • National Institute on Aging (U01AG058635, R01AG068281, R01AG068558, RF1AG068558, U01AG066757, RF1AG068281, RF1AG047644, RF1AG054011, R01AG085404, F30AG079492, F30AG079491, P30AG066444, P30AG066514, U19AG069701, R01AG072489)
  • National Institute of Mental Health (DP2MH100012, R01MH118329, R01MH113743)
  • National Institute on Drug Abuse (P01DA047233)
  • National Eye Institute (P30EY007003)
  • National Cancer Institute (P30CA196521)
  • National Institute of Neurological Disorders and Stroke (R01NS106721, R01NS117533, RF1NS090934)
  • National Center for Advancing Translational Sciences
Topics
  • Life Sciences › Neuroscience › Neurology › Neuroinflammation and Neurodegeneration Mechanisms
  • Life Sciences › Neuroscience › Cellular and Molecular Neuroscience › Alzheimer's disease research and treatments
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Inflammasome and immune disorders
Keywords
microglia, Alzheimer's disease, PU.1, CD28, amyloid plaques, neuroinflammation, neuroprotection, co-inhibitory receptors, microglial heterogeneity, immunotherapy
MeSH terms
Neuroprotection, Proto-Oncogene Protein Spi-1, Alzheimer Disease, Animals, Female, Humans, Male, Mice, Inbred C57BL, Proto-Oncogene Proteins, Trans-Activators, Microglia, CD28 Antigens, Disease Progression, Mice, Plaque, Amyloid

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Lymphoid gene expression supports neuroprotective microglia function. https://app.cochat.ai/papers/10.1038/s41586-025-09662-z
The paper
Ayata, P., Crowley, J. M., Challman, M. F., Sahasrabuddhe, V., Gratuze, M., Werneburg, S., Ribeiro, D., Hays, E. C., Durán-Laforet, V., Faust, T. E., Hwang, P., Mendes Lopes, F., Nikopoulou, C., Buchholz, S., Murphy, R. E., Mei, T., Pimenova, A. A., Romero-Molina, C., Garretti, F., . . . Schaefer, A. (2025). Lymphoid gene expression supports neuroprotective microglia function. Nature. https://doi.org/10.1038/s41586-025-09662-z

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.