Neuronal APOE4 alone is sufficient to drive tau pathology, neurodegeneration, and neuroinflammation in an Alzheimer's disease mouse model

Abstract

Apolipoprotein E4 (APOE4), the strongest genetic risk factor for late-onset Alzheimer's disease (AD), exacerbates tau tangles, amyloid plaques, neurodegeneration, and neuroinflammation-the pathological hallmarks of AD. While astrocytes are the primary producers of APOE in the CNS, neurons increase APOE expression under stress and aging. Prior work established that neuronal APOE4 is essential for AD pathogenesis, but whether it is sufficient to drive disease remained unknown. We generated a PS19 tauopathy mouse model selectively expressing APOE4 in neurons. Neuronal APOE4 alone proved sufficient to promote pathological tau accumulation and propagation, neurodegeneration, and neuroinflammation to levels comparable to a tauopathy model with human APOE4 knocked-in globally. Single-nucleus RNA sequencing further revealed similar transcriptomic changes in neurons and glia of both models. Together, these findings demonstrate that neuronal APOE4 alone can initiate and propagate AD pathologies, underscoring its pivotal role in disease pathogenesis and its potential as a therapeutic target.

Extended Data Fig. 1|. Schematic of APOE4 minigene construct and validation of the PS19/NSE-E4 mouse model with APOE selectively expressed by neurons.

a, Schematic of APOE4 minigene inserted under the NSE promoter. b,c, Representative images of the cell-type specificity of APOE expression in the hippocampus of 10-month-old PS19/E4 and PS19/NSE-E4 mice. In b, APOE is depicted in red, co-labeled with microglia (IBA1) in blue, astrocytes (GFAP) in green, and neurons (NeuN) in white (scale bar, 20 μm). In c, APOE is depicted in red, co-labeled with oligodendocytes (Olig2) in green and neurons (NeuN) in white (scale bar, 20 μm).

Extended Data Fig. 2|. Neuronal APOE4-promoted gliosis is correlated with neurodegeneration.

a, Representative western blot images with microglia IBA1 antibody. GAPDH served as a loading control. b, Representative western blot images with astrocyte GFAP antibody. Membrane in a was stripped and stained for GFAP and GAPDH. GAPDH served as a loading control. c,d, Quantifications of IBA1 (c) and GFAP (d) levels in hippocampal lysates of PS19/E3 (n = 10), PS19/E4 (n = 10), and PS19/NSE-E4 (n = 10) mice. Immunoblot levels were normalized to GAPDH first and then to those of PS19/E4 mice. e,f, Correlation of IBA1 microglia levels and DG layer thickness in PS19/E4 mice (e) and PS19/NSE-E4 mice (f). g,h, Correlation of IBA1 microglia levels and CA1 layer thickness in PS19/E4 mice (g) and PS19/NSE-E4 mice (h). i,j, Correlation of CD68 microglia levels and DG layer thickness in PS19/E4 mice (i) and PS19/NSE-E4 mice (j). k,l, Correlation of CD68 microglia levels and CA1 layer thickness in PS19/E4 mice (k) and PS19/NSE-E4 mice (l). m,n, Correlation of GFAP astrocyte levels and DG layer thickness in PS19/E4 mice (m) and PS19/NSE-E4 mice (n). o,p, Correlation of GFAP astrocyte levels and CA1 layer thickness in PS19/E4 mice (o) and PS19/NSE-E4 mice (p). q,r, Correlation of S100β astrocyte levels and DG layer thickness in PS19/E4 mice (q) and PS19/NSE-E4 mice (r). s,t, Correlation of S100β astrocyte levels and CA1 layer thickness in PS19/E4 mice (s) and PS19/NSE-E4 mice (t). In f-t, PS19/E3, n = 24; PS19/E4, n = 25; and PS19/NSE-E4, n = 18. Correlations utilized Pearson correlation analysis to determine significance.

Extended Data Fig. 3|. Neuronal APOE4 alone drives additional pathological endpoints in the in vivo Tau propagation model.

a, Representative images of human Tau (HT7) immunolabeling of the ipsilateral hippocampus of APOE3, APOE4, and NSE-E4 mice, 10 weeks post unilateral injection of mutant human Tau virus (scale bar, 500 μm). b, Quantifications of the percent HT7 coverage area in the ipsilateral hippocampus of these mice. c, Quantifications of the number of HT7-positive soma on the contralateral side normalized to the percent HT7 coverage area of the ipsilateral side of these mice. d, Representative images of p-Tau (AT8) immunolabeling of the ipsilateral hippocampus of APOE3, APOE4, and NSE-E4 mice, 10 weeks post unilateral injection of mutant human Tau virus (scale bar, 500 μm). e, Quantifications of the percent AT8 coverage area in the ipsilateral hippocampus of these mice. f, Quantifications of the number of AT8-positive soma on the contralateral side normalized to the HT7 percent coverage area of the ipsilateral side of these mice. g, Representative images of human Tau (HT7) and p-Tau (AT8) immunolabeling of the ipsilateral entorhinal cortex (EC) of injected APOE3, APOE4, and NSE-E4 mice (scale bar, 200 μm). h,i, Quantifications of the number of AT8-positive cells (h) and HT7-positive cells (i) in the ipsilateral EC of these mice. j, Representative images of human Tau (HT7) immunolabeling of the contralateral EC of injected APOE3, APOE4, and NSE-E4 mice (scale bar, 200 μm). k, Quantifications of the number of HT7-positive cells in the contralateral EC of these mice. l, Representative hippocampal cross-sections of injected APOE3, APOE4, and NSE-E4 mice, stained with Sudan Black to enhance hippocampal visualization (scale bar, 2000 μm). m,n, Quantifications of the contralateral (m) and ipsilateral (n) hippocampal volumes of these mice. For quantifications in b,c,e,f,h,i,k,m,n, APOE3 n = 8; APOE4 n = 8; and NSE-E4 n = 8; data is expressed as mean ± s.e.m and was assessed via one-way analysis of variance (ANOVA) with Tukey’s post hoc multiple comparisons test.

Extended Data Fig. 4|. snRNA-seq identification of cell clusters.

a, Dot-plot depicting normalized average expression of selected cell identity marker genes for all 33 unique hippocampal cell clusters from mice at 10 months of age. The size of the dots is proportional to the percentage of cells expressing a given gene. Average expression is on a colored scale (lower expression, blue; higher expression, red). b, Assigned identity of 33 distinct cell types. In neuron, inhibitory neuron; Ex neuron, excitatory neuron; oligo, oligodendrocyte.

Extended Data Fig. 5|. Characteristics of neuronal APOE4-vulnerable subiculum neuron cluster 22.

a, Dot-plot of normalized average expression of marker genes and genes of interest for selected subiculum neuron clusters. The size of the dots is proportional to the percentage of cells expressing a given gene. Average expression is on a colored scale (lower expression, blue; higher expression, red). b, Volcano plot of the differentially-expressed (DE) genes between PS19/E4 and PS19/E3 in cluster 22 subiculum neurons. c, Volcano plot of the DE genes between PS19/NSE-E4 and PS19/E3 in cluster 22 subiculum neurons. Up- or down-regulated genes shared between b and c are indicated in red font. d, Volcano plot of the DE genes between subiculum neuron cluster 22 and all other subiculum clusters. For all volcano plots in b-d, Dashed lines represent log₂ fold change threshold of 0.1 and p value threshold of 0.05. The unadjusted p values and log₂ fold change values used were generated from the gene-set enrichment analysis using the two-sided Wilcoxon rank-sum test as implemented in the FindMarkers function of the Seurat package.

Extended Data Fig. 6|. Neuronal APOE4 induces mild astrocyte changes.

a, Genotype-split UMAP highlighting astrocyte clusters 15 and 23 across each genotype group. b, Box plot of the proportion of cells from each sample for astrocyte clusters 15 and 23 (PS19/E3, n = 4; PS19/E4, n = 4; and PS19/NSE-E4, n = 4). From bottom to top, the hinges of the box plots correspond to the 25^th, 50^th, and 75^th percentiles. The upper and lower whiskers of the box plot extend to the largest and smallest values, respectively, though no further than 1.5 × IQR from the nearest hinge. IQR, interquartile range, or distance between the 25^th and 75^th percentiles. The log odds ratios (LOR) are the mean ± s.e.m estimates of LOR for these clusters, which represents the change in the log odds of cells per sample from each mouse belonging to the respective clusters relative to log odds of cells per sample from PS19/E3 mice. c, Volcano plot of the differentially-expressed (DE) genes between PS19/E4 and PS19/E3 in cluster 15 astrocytes. d, Volcano plot of the DE genes between PS19/NSE-E4 and PS19/E3 in cluster 15 astrocytes. Up- or down-regulated genes shared between c and d are indicated in red font. e, Volcano plot of the DE genes between PS19/E4 and PS19/E3 in cluster 23 astrocytes. f, Volcano plot of the DE genes between PS19/NSE-E4 and PS19/E3 in cluster 23 astrocytes. Up- or down-regulated genes shared between e and f are indicated in red font. g, Dot-plot of normalized average expression of marker genes and genes of interest for astrocyte clusters. The size of the dots is proportional to the percentage of cells expressing a given gene. Average expression is on a colored scale (lower expression, blue; higher expression, red). Unadjusted p values in b are from fits to a GLMM_AM; association tests were two-sided. In c-f, dashed lines represent log₂ fold change threshold of 0.1 and p value threshold of 0.05; the unadjusted p values and log₂ fold change values used were generated from the differential expression analysis using the two-sided Wilcoxon rank-sum test as implemented in the FindMarkers function of the Seurat package.

Fig. 1|. Neuronal APOE4 alone exacerbates Tau pathology in tauopathy mice.

a, Representative images of the hippocampus from 10-month-old PS19/E3, PS19/E4, and PS19/NSE-E4 mice, stained with p-Tau-specific monoclonal AT8 antibody (scale bar, 500 μm). b, Quantifications of percent p-Tau coverage of the hippocampus of PS19/E3, PS19/E4, and PS19/NSE-E4 mice. c, Representative images of the four levels of AT8 hippocampal staining patterns (scale bar, 500 μm). d, Stacked bar graph of the distribution of p-Tau staining patterns across PS19/E3, PS19/E4, and PS19/NSE-E4 mice. e, Quantification of p-Tau staining pattern severity across PS19/E3, PS19/E4, and PS19/NSE-E4 mice. f, Correlation of hippocampal p-Tau staining pattern and percent p-Tau coverage area. For quantifications in b,d,e,f, PS19/E3 n = 24; PS19/E4 n = 25; and PS19/NSE-E4 n = 18. In b,e,f, data is expressed as mean ± s.e.m and was assessed via one-way analysis of variance (ANOVA) with Tukey’s post hoc multiple comparisons test.

Fig. 2|. Neuronal APOE4 alone potentiates neurodegeneration and hippocampal atrophy in tauopathy mice.

a, Representative images of cross-sectional hippocampal brain tissue from 10-month-old PS19/E3, PS19/E4, and PS19/NSE-E4 mice, incubated in Sudan Black to enhance hippocampal visibility (scale bar, 1000 μm). b, Quantifications of hippocampal volume of PS19/E3, PS19/E4, and PS19/NSE-E4 mice. c,d, Representative images of the DG granule layer (scale bar, 200 μm) (c) and the CA1 pyramidal layer subfield (scale bar, 50 μm) (d) of PS19/E3, PS19/E4, and PS19/NSE-E4 mice, labeled with NeuN to capture neurons. e,f, Quantification of DG granule layer thickness (e) and CA1 pyramidal layer (f) of PS19/E3, PS19/E4, and PS19/NSE-E4 mice. g,h, Correlation of hippocampal volume and DG layer thickness in PS19/E4 mice (g) and PS19/NSE-E4 mice (h). i,j, Correlation of hippocampal volume and CA1 layer thickness in PS19/E4 mice (i) and PS19/NSE-E4 mice (j). For quantifications throughout, PS19/E3 n = 24; PS19/E4 n = 25; and PS19/NSE-E4 n = 18. In b,e,f, data is expressed as mean ± s.e.m and was assessed via one-way analysis of variance (ANOVA) with Tukey’s post hoc multiple comparisons test. Pearson correlation analysis was used to determine significance in g,h,i,j.

Fig. 3|. Neuronal APOE4 alone drives gliosis in tauopathy mice.

a, Representative images of hippocampal brain tissue from 10-month-old PS19/E3, PS19/E4, and PS19/NSE-E4 mice, immunolabeled with microglial marker IBA1. b, Quantifications of percent coverage area of IBA1. c, Representative images of hippocampal brain tissue from PS19/E3, PS19/E4, and PS19/NSE-E4 mice, immunolabeled with activated microglial marker CD68. d, Quantifications of percent coverage area of CD68. e, Representative images of hippocampal brain tissue from PS19/E3, PS19/E4, and PS19/NSE-E4 mice, immunolabeled with astrocytic marker GFAP. f, Quantifications of percent coverage area of GFAP. g, Representative images of hippocampal brain tissue from PS19/E3, PS19/E4, and PS19/NSE-E4 mice, immunolabeled with activated astrocytic marker S100β. h, Quantifications of percent coverage area of S100β. In a,c,e,g, Scale bars at 300 μm. i-l, Correlation of hippocampal volume and IBA1 (i), GFAP (j), CD68 (k), and S100β (l) in PS19/E4 mice (n = 25). m-p, Correlation of hippocampal volume and IBA1 (m), GFAP (n), CD68 (o), and S100β (p) in PS19/NSE-E4 mice (n = 18). Throughout b,d,f,h, PS19/E3, n = 24; PS19/E4, n = 25; and PS19/NSE-E4, n = 18; data is expressed as mean ± s.e.m and was assessed via one-way analysis of variance (ANOVA) with Tukey’s post hoc multiple comparisons test. Pearson correlation analysis was used to determine significance in i-p.

Fig. 4|. Neuronal APOE4 alone stimulates intracellular HMGB1 translocation.

a, Representative images of CA1 pyramidal layer from 10-month-old PS19/E3, PS19/E4, and PS19/NSE-E4 mice, immunolabeled for HMGB1 (green), DAPI (blue), and NeuN (magenta) (scale bar, 20 μm). b, Quantifications of extranuclear HMGB1 fluorescent intensity in the CA1 of PS19/E3, PS19/E4, and PS19/NSE-E4 mice. c, Representative images of DG granule layer from PS19/E3, PS19/E4, and PS19/NSE-E4 mice, immunolabeled for HMGB1 (green), DAPI (blue), and NeuN (magenta) (scale bar, 20 μm). b, Quantifications of extranuclear HMGB1 fluorescent intensity in the DG of PS19/E3, PS19/E4, and PS19/NSE-E4 mice. In b,d, PS19/E3 n = 24; PS19/E4 n = 25; and PS19/NSE-E4 n = 18; data is expressed as mean ± s.e.m and was assessed via one-way analysis of variance (ANOVA) with Tukey’s post hoc multiple comparisons test.

Fig. 5|. Neuronal APOE4 alone promotes Tau propagation and subsequent pathologies.

a, Schematic depicting experimental design of in vivo Tau propagation study. b, Representative images of human Tau (HT7 monoclonal antibody) immunolabeling of the contralateral hippocampus of APOE3, APOE4, and NSE-E4 mice, 10 weeks post unilateral injection of mutant human Tau virus (scale bar, 500 μm). c, Quantifications of the number of HT7-positive soma in the contralateral hippocampus of these mice. d, Representative images of p-Tau (AT8) immunolabeling of the contralateral hippocampus of injected APOE3, APOE4, and NSE-E4 mice (scale bar, 500 μm). e, Quantifications of the number of AT8-positive soma in the contralateral hippocampus of these mice. f-g, Representative images of microglial (IBA1) immunolabeling of the ipsilateral (f) and contralateral (g) hippocampus of injected APOE3, APOE4, and NSE-E4 mice (scale bar, 500 μm). h,i, Quantifications of the percent area coverage of microglia (IBA1) in the ipsilateral (h) and contralateral (i) hippocampus of these mice. j,k, Representative images of astrocytic (GFAP) immunolabeling of the ipsilateral (j) and contralateral (k) hippocampus of injected APOE3, APOE4, and NSE-E4 mice (scale bar, 500 μm). l,m, Quantifications of the percent area coverage of astrocyte (GFAP) in the contralateral (l) and ipsilateral (m) hippocampus of these mice. n,o, Representative images of DG neuronal layers (NeuN) of the ipsilateral (n) and contralateral (o) hippocampus of injected APOE3, APOE4, and NSE-E4 mice (scale bar, 200 μm). p,q, Quantifications of the DG granule layer thickness in the contralateral (p) and ipsilateral (q) hippocampus of these mice. For quantifications in c,e,h,i,l,m,p,q, APOE3 n = 8; APOE4 n = 8; and NSE-E4 n = 8; data is expressed as mean ± s.e.m and was assessed via one-way analysis of variance (ANOVA) with Tukey’s post hoc multiple comparisons test.

Fig. 6|. Neuronal APOE4 alone elicits loss of vulnerable neuronal subtypes.

a, UMAP plot of the 33 unique cell clusters identified in hippocampi from 10-month-old PS19/E3 (n = 4), PS19/E4 (n = 4), and PS19/NSE-E4 (n = 4) mice. b, Genotype-split UMAP highlighting neuronal clusters 5 and 22 across each genotype group. c, Heatmap plot of the log odds ratio (LOR) per unit change for each pathological parameter against each cell cluster. The LOR represents how changes in each pathological measurement (AT8 coverage area, hippocampal volume, etc.) are associated with the odds that a cell belongs to a given cluster versus not, while accounting for genotype and sample variability through random effects. Associations with pathologies are on a colored scale (negative associations, blue; positive associations, red). d, Box plot of the proportion of cells from each sample for clusters 5 and 22 (PS19/E3, n = 4; PS19/E4, n = 4; and PS19/NSE-E4, n = 4). From bottom to top, the hinges of the box plots correspond to the 25^th, 50^th, and 75^th percentiles. The upper and lower whiskers of the box plot extend to the largest and smallest values, respectively, though no further than 1.5 × IQR from the nearest hinge. IQR, interquartile range, or distance between the 25^th and 75^th percentiles. The LOR are the mean ± s.e.m estimates for these clusters, which represents the change in the log odds of cells per sample from each mouse belonging to the respective clusters relative to log odds of cells per sample from PS19/E3 mice. e, Feature plots illustrating the expression of SST and PV in inhibitory neuron cluster 5. f, Dot-plot of normalized average expression of marker genes and genes of interest for selected inhibitory neuron clusters. The size of the dots is proportional to the percentage of cells expressing a given gene. Average expression is on a colored scale (lower expression, blue; higher expression, red). g, Volcano plot of the differentially-expressed (DE) genes between PS19/E4 and PS19/E3 in cluster 5 inhibitory neurons. h, Volcano plot of the DE genes between PS19/NSE-E4 and PS19/E3 in cluster 5 inhibitory neurons. Up- or down-regulated genes shared between g and h are indicated in red font. i, Volcano plot of the DE genes between inhibitory neuron cluster 5 and all other inhibitory neuron clusters. In g-i, Dashed lines represent log₂ fold change threshold of 0.1 and an adjusted p value threshold of 0.05. The unadjusted P values and log₂ fold change values used were generated from the differential expression analysis using the two-sided Wilcoxon rank-sum test as implemented in the FindMarkers function of the Seurat package. In c, unadjusted p values are from fits to a GLMM_histopathology. In d, unadjusted p values are from fits to a GLMM_AM. Two-sided association tests were used. Ex neuron, excitatory neuron; In neuron, inhibitory neuron; oligo, oligodendrocyte.

Fig. 7|. Neuronal APOE4 alone produces disease-associated microglia and oligodendrocytes.

a, Genotype-split UMAP highlighting microglial clusters 8 and 13 and oligodendrocyte cluster 17 across each genotype group. b,c, Box plot of the proportion of cells from each sample for microglial clusters 8 and 13 (b) and oligodendrocyte cluster 17 (c) (PS19/E3, n = 4; PS19/E4, n = 4; and PS19/NSE-E4, n = 4). From bottom to top, the hinges of the box plots correspond to the 25^th, 50^th, and 75^th percentiles. The upper and lower whiskers of the box plot extend to the largest and smallest values, respectively, though no further than 1.5 × IQR from the nearest hinge. IQR, interquartile range, or distance between the 25^th and 75^th percentiles. The log odds ratios (LOR) are the mean ± s.e.m estimates of LOR for these clusters, which represents the change in the log odds of cells per sample from each mouse belonging to the respective clusters relative to log odds of cells per sample from PS19/E3 mice. d, Dot-plot of average expression of marker genes and genes of interest for microglial clusters. To best compare only two clusters, dot-plot values are presented without normalization. e, Dot-plot of normalized average expression of marker genes and genes of interest for oligodendrocyte clusters (e). For d,e, the size of the dots is proportional to the percentage of cells expressing a given gene. Average expression is on a colored scale (lower expression, blue; higher expression, red). f, Volcano plot of the differentially-expressed (DE) genes between PS19/E4 and PS19/E3 in cluster 8 microglia. g, Volcano plot of the DE genes between PS19/NSE-E4 and PS19/E3 in cluster 8 microglia. Up- or down-regulated genes shared between f and g are indicated in red font. h, Volcano plot of the differentially-expressed (DE) genes between PS19/E4 and PS19/E3 in cluster 13 microglia. i, Volcano plot of the DE genes between PS19/NSE-E4 and PS19/E3 in cluster 13 microglia. Up- or down-regulated genes shared between h and i are indicated in red font. j, Volcano plot of the DE genes between oligodendrocyte cluster 5 and all other oligodendrocyte clusters. k, Volcano plot of the DE genes between PS19/E4 and PS19/E3 in cluster 17 oligodendrocytes. l, Volcano plot of the DE genes between PS19/NSE-E4 and PS19/E3 in cluster 17 oligodendrocytes. Up- or down-regulated genes shared between k and l are indicated in red font. In f-l, dashed lines represent log₂ fold change threshold of 0.1 and adjusted p value threshold of 0.05. The adjusted p values and log₂ fold change values used were generated from the differential expression analysis using the two-sided Wilcoxon rank-sum test as implemented in the FindMarkers function of the Seurat package. Unadjusted p values in b,c are from fits to a GLMM_AM; association tests were two-sided.