Silencing Apoe with divalent-siRNAs improves amyloid burden and activates immune response pathways in Alzheimer's disease

Abstract

Introduction: The most significant genetic risk factor for late-onset Alzheimer's disease (AD) is APOE4, with evidence for gain- and loss-of-function mechanisms. A clinical need remains for therapeutically relevant tools that potently modulate APOE expression.

Methods: We optimized small interfering RNAs (di-siRNA, GalNAc) to potently silence brain or liver Apoe and evaluated the impact of each pool of Apoe on pathology.

Results: In adult 5xFAD mice, siRNAs targeting CNS Apoe efficiently silenced Apoe expression and reduced amyloid burden without affecting systemic cholesterol, confirming that potent silencing of brain Apoe is sufficient to slow disease progression. Mechanistically, silencing Apoe reduced APOE-rich amyloid cores and activated immune system responses.

Discussion: These results establish siRNA-based modulation of Apoe as a viable therapeutic approach, highlight immune activation as a key pathway affected by Apoe modulation, and provide the technology to further evaluate the impact of APOE silencing on neurodegeneration.

Keywords: Alzheimer's Disease; Apoe; RNAi; neurodegeneration; oligonucleotide therapeutics; siRNA.

FIGURE 1

Silencing brain Apoe reduces AD neuropathology in APP/PSEN1 mice with no effect on serum cholesterol. (A,B) Experimental timeline. Di‐siRNA dose is 237 μg. Di‐siRNA^NTC shown in grey, and di‐siRNA^APOE shown in green. (C,D) Apoe mRNA (C), and APOE protein (D) expression in the hippocampus, cortex, and liver 2 months post administration of di‐siRNA^NTC or di‐siRNA^APOE. (E) Serum HDL and LDL cholesterol in treated and control mice (n = 8 to 9 per group). (F) APP6E10 plaques in the cortex 2 months post‐injection in treated mice (left) and controls (right). (G) X‐34‐positive plaques in cortex 2 months post‐treatment in treated mice (left) and controls (right). (H–I) Sex‐stratified quantification of APP6E10‐positive cortex plaques (H), and insoluble Aβ40 fibrils (I) and insoluble Aβ 42 fibrils (J) in cortex samples. (K–M) Sex‐stratified quantification of X‐34‐positive plaques in the cortex (K), soluble Aβ40 fibrils (L) and soluble Aβ42 fibrils (M) in cortex samples. mRNA evaluated using QuantiGene. APOE protein quantified with WES ProteinSimple; statistical analysis was one‐way ANOVA or t‐test using GraphPad Prism. Error bars are SD. Middle bars are median. Aβ, amyloid beta; AD, Alzheimer's disease; di‐siRNA, divalent small interfering RNA; HDL, high‐density lipoprotein; LDL, low‐density lipoprotein. *p < 0.05, **p < 0.01, ***p < 0.002, ****p < 0.0001.

FIGURE 2

Silencing Apoe after the onset of pathologic changes improves amyloid burden in 5xFAD mice. (A) Experimental design. (B) Schematic showing antibody binding sites for APP6E10 and Aβ42‐specific antibody (D9A3A). APP6E10 will react with Aβ and soluble APPα (sAPP α). (C) mRNA expression throughout the brain 2 weeks and 2 months post administration of di‐siRNA^APOE. Controls (di‐siRNA^NTC) are shown in grey and treated (di‐siRNA^APOE) are shown in green. (D) Amyloid deposition using APP6E10 in controls (left) and treated (right) animals. Scale bar: 1000 μm. (E) Aβ−42 using D9A3A in controls (left) and treated (right) animals. Scale bar: 1000 μm. (F) Zoom of cortical regions from panel D showing neuronal amyloid reactivity in both groups and reduction in amyloid plaque deposition in di‐siRNA^APOE treated animals. Scale bar: 1000 μm. (G) Zoom of cortical regions from panel E showing Aβ42 plaque reduction in di‐siRNA^APOE treated animals. Scale bar: 1000 μm. (H,I) Quantification of plaque burden measured as percent area, separated by sex using APP6E10 (H) and Aβ42 antibodies (I). (J,K) MSD quantification of insoluble amyloid in the cortex (J) and hippocampus (K). Statistics: t‐test per brain region; sex separated; n = 9–10 per sex per group. Timepoint: 17 weeks old, 2 months post‐injection. Aβ, amyloid beta; di‐siRNA, divalent small interfering RNA; MSD, Meso Scale Discovery; NTC, non‐targeting (vehicle) controls. *p < 0.0332, **p < 0.0021, ***p < 0.0002, ****p < 0.0001.

FIGURE 3

Silencing Apoe primarily reduces plaque number. (A) Amyloid plaque size, and (B) amyloid plaque number measured using APP6E10 antibody. (C) Aβ42 plaque size, and (D) Aβ42 plaque number measured using Aβ42‐specific antibody (D9A3A) in 5xFAD treated and control mice, separated by sex. (E) MSD ELISA quantification of Aβ42 and Aβ40 in soluble brain fractions. Statistics: t‐test per brain region; sex separated. Timepoint: 17 weeks old, 2 months post‐injection (n = 9‐10 per sex per group). Di‐siRNA^NTC is shown in grey, and di‐siRNA^ApoE is shown in green. Aβ, amyloid beta; ELISA, enzyme‐linked immunosorbent assay; MSD, Meso Scale Discovery. *p < 0.0332, **p < 0.0021; ***p < 0.0002; ****p < 0.0001.

FIGURE 4

Apoe silencing decreases amyloid CTF production and removes structural support for amyloid plaques. Western blots (A) and quantification (B) showing a reduction in CTFs and Aβ42 in the cortex of treated compared to control animals. Western blots (C) and quantification (D) showing a reduction in APP and Aβ42 in the hippocampus of treated compared to control animals. N = 4–5 per group. Statistics: one‐way ANOVA. (E) RNA expression levels of APP, PSEN1, PSEN2, and BACE1 in the cortex of treated and control animals. (F,G) Co‐staining of amyloid plaques (APP6E10, green) and APOE (red) in control (F) and Apoe silenced (G) 5xFAD mice. Insets show amyloid (top, green) and APOE (red, bottom). Arrows denote early plaque with APOE present in the core while star denotes APOE poor plaque with altered structure. 63x, scale bar: 10 μm. Aβ, amyloid beta; APP, amyloid precursor protein; CTF, carboxy‐terminal fragment; di‐siRNA, divalent small interfering RNA; NTC, non‐targeting (vehicle) controls; RIPA, radioimmunopreciptation assay. *p < 0.0332, **p < 0.0021, ***p < 0.0002, ****p < 0.0001.

FIGURE 5

Apoe silencing activates the expression of immune‐related pathways. Co‐staining of amyloid plaques (APP, green) and microglia (IBA1, green) in control (A) and treated (B) 5xFAD mice. Insets show amyloid (top, green) and IBA1 (bottom, green). Representative images from corresponding cortex regions are shown. 40x, scale bar: 10 μm. (C) Quantification of IBA1+ cell clustering around amyloid plaques. Statistics: one‐way ANOVA. (D–F) Volcano plots showing differentially expressed genes after two weeks (D) and 2 months (E) of Apoe silencing in 5xFAD mice, and differentially expressed genes after 2 months of Apoe silencing in APP/PSEN1 mice (F). FDR set to < 0.05; shown in pink. Green denotes FDR < 0.05 and the presence of the seed match in the 3′UTR. Grey: not significant. Log2fold change cut‐off set to 0.5 (+/‐). (G,H) Gene set enrichment analysis on genes expressed in all conditions, clustered by gene set member semantic similarity. Labels indicate cluster themes. (I) Analysis of transcription factor target gene enrichment in upregulated genes at 2 months post‐treatment in 5XFAD, using the MAGIC package. (J) Comparison of gene overlap between Apoe silencing for 2 months in 5XFAD and marker genes of disease‐associated microglia from Keren‐Shaul et al., 2017. (K) Volcano plot of Pearson correlation between each gene in the “Disease‐Associated Microglia” marker gene set and Apoe in the 5XFAD mice 2 months post‐injection. Mice: 5xFAD, Age: 11 and 17 weeks old. Timepoints: 2 weeks (n = 9–10/group) and 2 months (n = 18–20/group) post‐injection. APP/PSEN1 mouse model: Age: 16 weeks old, 2 months post‐injection. N = 8–10 per group. di‐siRNA, divalent small interfering RNA; FDR, false discovery rate; NTC, non‐targeting (vehicle) controls.

FIGURE 6

Apoe silencing dynamically rescues gene expression alterations in 5xFAD mice. (A) Correlation between WGCNA clusters, treatment, and amyloid burden, measured as percent of brain area positive for amyloid IHC (see methods). (C–E) WGCNA module gene average expression z‐score and gene ontology results for modules that are significantly correlated with amyloid burden and treatment conditions (R > 0.3, p‐value < 0.05). 5xFAD mouse model; Age: 11 and 17 weeks old. Timepoints: 2 weeks post‐injection (n = 9–10/group), 2 months post‐injection (n = 18–20/group). IHC, immunohistochemistry; NTC, non‐targeting (vehicle) controls; WGCNA, weighted gene co‐expression network analysis.