SNX8 Enhances Non-amyloidogenic APP Trafficking and Attenuates Aβ Accumulation and Memory Deficits in an AD Mouse

Abstract

Dysregulation of various APP trafficking components in the endosome has been previously implicated in Alzheimer's disease (AD). Although single nucleotide polymorphisms within the gene locus encoding the endosomal component, SNX8 have been previously associated with AD, how SNX8 levels are altered and its contribution to AD onset is currently unknown. Here, we observe decreased expression of SNX8 in human AD and AD mouse brain. SNX8 predominantly localized to early and late endosomes, where SNX8 overexpression enhanced total APP levels, cell surface APP distribution and consequent soluble APPα cleavage. SNX8 depletion resulted in elevated β-amyloid (Aβ) levels, while SNX8 overexpression reduced Aβ levels in cells and in an APP/PS1 AD mouse model. Importantly, SNX8 overexpression rescued cognitive impairment in APP/PS1 mice. Together, these results implicate a neuroprotective role for SNX8 in enhancing non-amyloidogenic APP trafficking and processing pathways. Given that endosomal dysfunction is an early event in AD, restoration of dysfunctional endosomal components such as SNX8 may be beneficial in future therapeutic strategies.

Keywords: APP trafficking; Alzheimer’s disease; SNX8; sAPPα; sorting nexins; β-amyloid; β-amyloid precursor protein.

FIGURE 1

SNX8 levels are reduced in human AD and APP/PS1 AD mouse brain. (A) Equal protein quantities in lysates from paired AD patient brains (AD) and age- and sex-matched non-dementia controls (Ctrl) were subjected to Western blotting with antibodies to detect SNX8, phosphorylated tau (PHF1) and β-actin. SNX8 protein intensities were quantified by densitometry, and normalized to respective controls (set to 1.0), n = 5, ^∗p < 0.05, Mann–Whitney U test. (B,C) Equal protein quantities of lysates from hippocampal (B) or cortical (C) tissues of APP/PS1 AD mice and wild type (WT) littermates at 3 and 9 months of age were subjected to Western blotting with antibodies against SNX8, APP (the 369 antibody recognizing both mouse and human APP was used in B; the 6E10 antibody recognizing human APP only was used in C) and GAPDH. SNX8 protein intensities were quantified by densitometry, and normalized to respective controls (set to 1.0), n = 4, ^∗p < 0.05, Mann–Whitney U test.

FIGURE 2

Characterizing SNX8 Expression and Localization. (A) Human and mouse SNX8 protein sequences were aligned and compared. Conserved amino acid residues are highlighted in red/yellow, weak similarities are highlighted in green, and gaps are marked with dashed lines (“–”). (B) Equal protein quantities from various C57BL/6 mouse (2-month-old) tissues were analyzed for SNX8 and GAPDH. SNX8 band intensities were quantified by densitometry, normalized to GAPDH; and compared to levels detected in heart (set to 1.0), n = 3. (C) Equal protein quantities from cultured primary neurons, astrocytes and microglia were analyzed for SNX8, NeuN (neuron marker), Iba1 (microglia marker), GFAP (astrocyte marker), and β-actin levels. SNX8 levels were quantified by densitometry, normalized to those of β-actin, and then normalized to levels detected in neurons (set to 1.0), n = 4. (D) HeLa cells co-transfected with SNX8-myc and EGFP-tagged Rab4, Rab5, or Rab7 constructs were immunostained for myc (to detect SNX8, red) and stained with DAPI (blue). HeLa cells transfected with SNX8-myc were immunostained for myc (SNX8, red) and Giantin (green), and stained with DAPI (blue). Images were acquired by confocal microscopy. Rab4, Rab5, Rab7, and Giantin staining is depicted in green. Zoom images are magnified panels from regions in merged images. Scale bars, 10 μm.

FIGURE 3

SNX8 Modulates APP, Aβ, and sAPPα Levels. (A,B) HEK-swAPP cells were transfected with control (Ctrl) and SNX8 constructs for 48 h (A), or transfected with scrambled control (SC) and SNX8 shRNA vectors (shSNX8) for 72 h (B). Cell lysates were subjected to Western blotting for APP, PS1-NTF, SNX8, and β-actin. Aβ and sAPPα in conditioned media were analyzed by Western blotting (sAPPα) or immunoprecipitation-Western blotting (Aβ). APP, Aβ, PS1-NTF, and sAPPα band intensities were quantified by densitometry, and normalized to respective controls (set to 1.0 arbitrary unit, A.U.). n = 4, ns: not significant, ^∗p < 0.05, Mann–Whitney U test. (C,D) Aβ₄₀ and Aβ₄₂ levels in conditioned media from cells overexpressing SNX8 (C) and SNX8 knockdown (D) were determined by ELISA and compared to respective controls. n = 4, ^∗p < 0.05, Mann–Whitney U test.

FIGURE 4

SNX8 Modulates APP Stability. (A) SH-SY5Y cells were transfected with SNX8-myc (SNX8) or control (Ctrl) vectors for 24 h. APP mRNA levels were determined by quantitative real-time PCR, normalized to β-actin levels, and compared to controls (set to 1.0), n = 4. ns: not significant, Mann–Whitney U test. (B) After transfection with SNX8-myc (SNX8) or control vectors for 24 h, SH-SY5Y cells were incubated with 500 μM cycloheximide (CHX) for the time indicated. Equal protein quantities from lysates were subjected to Western blotting. APP levels were quantified by densitometry, normalized to those at time point zero (set to 1.0), and compared to respective controls, n = 4, ^∗p < 0.05, two-way ANOVA. (C) HEK293T cells were transfected with APP-myc for 24 h, then transfected with scrambled control (SC) or SNX8 shRNA (shSNX8) vectors for 48 h. SNX8 and APP mRNA levels were determined by quantitative real-time PCR, normalized to β-actin levels, and compared to respective controls (set to 1.0), n = 4, ^∗p < 0.05, ns: not significant, Mann–Whitney U test. (D) After transfection with APP-myc for 24 h and scrambled control (SC) or SNX8 shRNA (shSNX8) vectors for 48 h, HEK293T cells were incubated with 500 μM CHX for the time indicated. Equal protein quantities from lysates were subjected to Western blotting. APP levels were quantified by densitometry, normalized to those at zero time point (set to 1.0), and compared to respective controls, n = 4, ^∗p < 0.05, two-way ANOVA.

FIGURE 5

SNX8 Regulates APP Trafficking to the Cell Surface. (A) HEK293T cells were co-transfected with APP-myc and mCherry control (mCherry-Ctrl) or mCherry-SNX8 vectors for 24 h. Cells were permeabilized, fixed, immunostained with antibodies against myc (to detect APP, red-pseudo color) and Giantin (to detect Golgi, green-pseudo color), and stained with DAPI (blue). Images were acquired by fluorescence microscopy. mCherry was shown in magenta (pseudo color). Scale bars, 10 μm. (B) HEK293T cells were co-transfected with APP-myc and SNX8-myc (SNX8) or control (Ctrl) vectors for 24 h, and subjected to cell surface biotinylation. Biotinylated cell surface components were precipitated from cell lysates with streptavidin-agarose beads. Cell surface biotin-labeled APP, PS1-NTF, and inputs from the cell lysates were detected by Western blotting. Cell surface APP (or PS1-NTF) and total APP (or PS1-NTF) intensities were quantified by densitometry, and the ratio of cell surface APP (or PS1-NTF) levels to total APP (or PS1-NTF) levels was compared to the control ratio (set to 1.0). n = 4, ^∗p < 0.05, ns: not significant, Mann–Whitney U test. (C) HEK293T cells were transfected with APP-myc and scrambled control (SC) or SNX8 shRNA (shSNX8) vectors for 48 h. After cell surface biotinylation, biotinylated proteins were precipitated for Western blotting. Cell surface APP and total APP intensities were quantified by densitometry, and the ratio of cell surface APP levels to total APP levels was compared to the control ratio (set to 1.0). n = 4, ^∗p < 0.05, Mann–Whitney U test.

FIGURE 6

SNX8 Overexpression Improves Short-Term Memory and Reduces Aβ₄₂ in APP/PS1 Mice. (A) A workflow diagram for AAV-SNX8 injection and subsequent analysis. (B) Equal protein amounts of hippocampal lysates from APP/PS1 mice injected with AAV-SNX8 or controls (AAV-Ctrl) were subjected to Western blotting for APP, SNX8, and GAPDH. APP and SNX8 band intensities were quantified by densitometry, and normalized to those of GAPDH for comparison (samples injected with AAV-Ctrl were set to 1.0 arbitrary unit, A.U.). AAV-SNX8, n = 8; AAV-Ctrl, n = 5. ^∗p < 0.05, ^∗∗p < 0.01, Mann–Whitney U test. (C) APP/PS1 mice injected with AAV-SNX8 and controls were evaluated for memory-related behavior in Y maze tests. Spontaneous alternations were determined between experimental groups. AAV-SNX8, n = 8; AAV-Ctrl, n = 5. ^∗p < 0.05, Mann–Whitney U test. (D) Aβ₄₂ levels in hippocampal lysates from injected mice were analyzed by ELISA. AAV-SNX8, n = 8; AAV-Ctrl, n = 5. ^∗p < 0.05, Mann–Whitney U test.