Annexin A6 membrane repair protein protects against amyloid-induced dystrophic neurites and tau phosphorylation in Alzheimer's disease model mice

Abstract

In Alzheimer's disease, accumulation of amyloid-β (Aβ) peptide is thought to cause formation of neurofibrillary tangles composed of hyperphosphorylated tau protein, which correlates with neuronal loss and cognitive impairment, but the mechanism linking Aβ and tau pathologies is unknown. Dystrophic neurites, which surround Aβ plaques and accumulate phosphorylated tau and other proteins, may play a role in seeding and spreading of pathologic tau. Here, we investigate the novel hypothesis that improved membrane repair capacity decreases dystrophic neurite formation by protecting axons from Aβ-induced membrane damage. Using a ratiometric calcium sensor and a FRET-based calpain cleavage sensor, we demonstrate that dystrophic neurites in 5XFAD mice have elevated resting calcium levels and calpain activity because of putative membrane damage. Annexin A6, a plasma membrane repair in muscle and neurons, is present at plasma membrane of neurons and dystrophic neurites in murine and human brains. Overexpression of annexin A6 in brains of 5XFAD mice decreased size and quantity of dystrophic neurites and accumulation of phospho-tau181, an early biomarker of amyloid pathology. Phospho-tau231, another early amyloid biomarker, and phosphorylated of tau kinases, c-jun N-terminal kinase (JNK) and Calmodulin Kinase II (CaMKII) accumulate in dystrophic neurites in the brains of amyloid pathology mice and humans with AD, suggesting that dystrophic neurites are sites of active tau phosphorylation. Overexpression of dominant-negative annexin A6 in 5XFAD mice increased dystrophic neurites and phospho-tau181. Intracerebral injection of recombinant annexin A6 in 5XFAD and APP-NLGF knock-in mice resulted in localization of recombinant A6 to membranes of dystrophic neurites, suggesting therapeutic potential of recombinant annexin A6 for AD. In conclusion, dystrophic neurites have Aβ-induced membrane damage characterized by calcium elevation, calpain activation, and accumulation of tau kinases and phosphorylated tau. Overexpression of annexin A6 reduces dystrophic neurites and phospho-tau accumulation, suggesting that annexin A6-mediated membrane repair may represent a novel therapeutic approach for AD.

Keywords: Alzheimer’s disease; Amyloid pathology; Calcium dysregulation; Dystrophic neurites; Membrane repair; Tau phosphorylation.

Fig. 1

Resting calcium levels are elevated in dystrophic neurites of 5XFAD mice. A Schematic of ratiometric calcium sensor GCaMP6s-ERK-TdTomato. B 300 μm live slices of 5XFAD (N = 5) mice injected with AAV PHP.eB syn-GCaMP6s-ERK-TdT were imaged on a Nikon A1R confocal multiphoton microscope at 25x. Upper panel of B shows an example of a neuronal soma with low GCaMP6s fluorescence (arrow) and dystrophic neurites with higher GCamP6s fluorescence (dotted circles). Lower panels show MethoxyX04 (MeX) marking dense core amyloid plaques and the fluorescence of GCaMP6 s (middle panel) or TdTomato (lower panel) independently. Imaging of brain slices from 5XFAD mice without GCaMP6s-ERK-TdTomato expression indicated that bright signals from plaques and puncta are autofluorescence of live tissue (Suppl. Fig. 1C). C, D For each z stack (n = 6–9 per mouse, from 3 different slices, 5 mice), the ratio of GCaMP6s:TdT was quantified in neuronal soma and dystrophic neurites using FIJI. C All neuronal soma and dystrophic neurites from one representative z stack are shown, indicating a significant increase in ratio in dystrophic neurites, but with wide variation (image shown in Sup. Fig 1D). Dotted black line represents median, solid black lines represents quartiles, and red points indicate individual cells or dystrophic neurites. D The average GCaMP6s:TdT ratio per mouse (n = 6–9 z-stacks per mouse, from 3 different slices, 5 mice) is shown, demonstrating a significantly increased GCaMP6s:TdT ratio, but with decreased variation. Two tailed Students t test, ****p < 0.0001, ***p < 0.001. Red points represent individual mice, with mean and SEM

Fig. 2

Calpain activity is elevated in dystrophic neurites of 5XFAD mice. A Schematic of calpain sensor CFP–PLFAAR–YFP. Arrow indicates calpain cleavage site. B, C At 6–7 months of age, 5XFAD mice and non-transgenic littermates (n=3 each genotype) received a retro-orbital injection of AAV PHP.eB syn-CFP–PLFAAR–YFP. After 5 weeks of expression, brains were collected and immunoblot analysis performed, which indicated significantly elevated ratio of cleaved to full-length CFP–PLFAAR–YFP in 5XFAD mice compared to non-Tg mice. D, E Using FIJI, FRET and YFP intensity were measured in neuronal soma (arrows) and dystrophic neurites (dotted circles). E The ratio of FRET:YFP was significantly decreased in dystrophic neurites compared to neuronal cell bodies, indicating an increase in calpain activity in dystrophic neurites. For (E), 3–4 random cortical fields were used per mouse (n = 2) containing 2–11 neuronal soma and 9–64 dystrophic neurites per image for a total of 46 cell bodies and 245 dystrophic neurites. Each red data point represents the ratio in a cortical field. Dotted black line represents median and solid black lines represent quartiles

Fig. 3

Annexin A6 localizes to neuronal plasma membrane in mouse and human brain. A Annexin A6 crystal structure. Arrows labeled “Ca⁺⁺” point at green dots that represent bound calcium molecules. B Schematic of annexin A6 repair cap complex following membrane injury. PS, phosphatidylserine; PIP2, phosphatidylinositol biphosphate. C, D Wild-type (WT) mouse (C) and cognitively normal human (D) brain sections immunostained for annexin A6 (red), NeuN (gray), Iba1 (green), GFAP (green), and DAPI (blue)

Fig. 4

Overexpressed and endogenous A6 localizes to dystrophic neurite membranes. A Cortical section of 5XFAD mouse transduced with AAV expressing A6-GFP driven by the neuron-specific synapsin promoter and immunostained for A6-GFP (anti-GFP, green), BACE1 (red), Aβ42 (white), and DAPI (blue) shows A6-GFP localized to membranes of neuron soma and BACE1+ DNs (arrows); *plaques; Bar=10μm. B, C 5XFAD (B) and APP-NLGF (C) cortical mouse brain sections immunostained for annexin A6 (red), BACE1 (green), APP (green), Aβ (blue), NeuN (gray), and DAPI (blue). We used the APP-NLGF knock-in mouse model of amyloid pathology to validate results obtained with 5XFAD mice. D, E Human AD hippocampal brain sections immunostained for annexin A6 (red), BACE1 (green), MeX04 (amyloid stain; blue), and NeuN (green). *plaque cores; double arrows, DNs; arrows, neurons. Bars=10μm (A), 25μm B-E)

Fig. 5

Annexin A6-GFP reduces dystrophic neurites without affecting Aβ deposits or glia in 5XFAD brain. A Coronal brain sections of 5XFAD mice transduced with AAV expressing GFP alone or A6-GFP driven by the neuron-specific synapsin promoter and immunostained for Aβ42 (red), LAMP1 (green), and DAPI (blue). Note the smaller LAMP1+ dystrophic neurite halo surrounding the amyloid plaque in A6-GFP expressing 5XFAD mice (far right). Bars=500 μm, left panels; bars=10 μm, right panels. B LAMP1/Aβ42 ratio of amyloid plaques in cortex (left) and hippocampus (right) of 5XFAD mice expressing GFP alone (n = 4) or A6-GFP (n = 4) binned by plaque core area with the indicated ranges in μm². Expression of A6-GFP significantly reduced LAMP1/Aβ42 ratio in smaller, faster growing plaques in the size ranges of 0–50 and 50–200 μm². ***, p < 0.001. C %Aβ42+ (left graph plots) and %LAMP1+ (right graph plots) areas are unchanged and significantly reduced, respectively, in 5XFAD mice expressing A6-GFP compared to GFP alone. *p < 0.05. D Endogenous GFP fluorescence was imaged using confocal microscopy, and ImageJ used to quantify size of GFP+ DNs. Average area of GFP+ DNs is significantly reduced in 5XFAD mice expressing A6-GFP compared to GFP alone. *, p<0.05. E Ratios of Iba1 (left graph plots) and GFAP (right graph plots) to amyloid as measured by MeX04 staining within 15 μm of plaque cores are unchanged in 5XFAD mice expressing A6-GFP compared to GFP alone, indicating A6-GFP causes no change in microglia or astrocytes near plaques. In all graphs, each point represents an individual animal

Fig. 6

p-tau181, p-tau231, and phosphorylated tau kinases increase in 5XFAD, APP-NLGF, and human AD dystrophic neurites, while annexin A6-GFP expression decreases p-tau181. A 5XFAD (left) and 5XFAD;Tau^-/- (right) cortex (ctx) sections immunostained for p-tau181 (red), LAMP1 (green), and DAPI (blue). In 5XFAD brain, p-tau181 accumulates in LAMP1+ DNs in contact with amyloid plaques. Note that 5XFAD;Tau^-/- brain forms amyloid plaques and LAMP1+ DNs, but that 5XFAD;Tau^-/- DNs lack p-tau181 accumulation. B p-tau181:Thiazine red (ThR; amyloid stain) ratio in 5XFAD mice transduced with AAV expressing GFP or A6-GFP driven by the neuron-specific synapsin promoter. Note that p-tau181:ThR ratio is significantly decreased in 5XFAD mice expressing A6-GFP. C APP-NLGF cortex section immunostained for p-tau181 (red), LAMP1 (green), and NeuN (blue). The accumulation of p-tau181 in DNs of APP-NLGF appears very similar to that of 5XFAD mice. D 5XFAD (left) and 5XFAD;Tau^-/- (right) cortex sections immunostained for p-tau231 (red), LAMP1 (green), and NeuN (blue). E–G 5XFAD cortex sections immunostained for p-tau181 (red), p-JNK (green), p-CaMKII (green), BACE1 (red), NeuN (blue), and DAPI (blue). H–J AD hippocampal sections immunostained for p-tau181 (red), total tau (Tau5, green), MeX04 (amyloid stain, blue), BACE1 (green), and APP (green). K AD hippocampal section immunostained for p-tau231 (green), p-JNK (red), APP (gray), MeX04 (blue), and DAPI (blue). *, plaque cores; arrows, DNs. Bars=10 μm. Images representative from n = 3 mice

Fig. 7

Dominant negative annexin N32 A6-GFP increases dystrophic neurites and p-tau181 in 5XFAD brain. A Cortical sections of 5XFAD mice transduced with AAV expressing annexin A6-GFP or dominant-negative A6 truncation, N32 A6-GFP driven by the neuron-specific synapsin promoter immunostained for GFP (green), p-tau181 (red), and MeX04 (amyloid stain, blue). Note the mislocalization of N32 A6-GFP away from the plasma membrane toward the cytoplasm and nucleus in N32 A6-GFP compared to A6-GFP expressing neurons. Bar=25 μm. B LAMP1:Aβ42 ratio of amyloid plaques in the cortices of A6-GFP and N32 A6-GFP expressing 5XFAD mice binned by plaque core area with the indicated ranges in μm². Smaller, faster growing plaques in the size ranges < 50 μm² (left) and 50–200 μm² (right) of N32 A6-GFP expressing mice had significantly increased LAMP1:Aβ42 ratio compared to that of A6-GFP expressing mice. C The p-tau181:MeX04 ratio of amyloid plaques in the cortices of A6-GFP and N32 A6-GFP expressing 5XFAD mice binned by plaque core area with the indicated ranges in μm² (<50 μm², left; 50–200 μm², right). As with LAMP1/Aβ42 ratio (B), p-tau181:MeX04 ratio was significantly increased in smaller, faster growing plaques of N32 A6-GFP compared to A6-GFP expressing mice. **p < 0.01, *p < 0.05. n = 4–6 mice

Fig. 8

Recombinant annexin A6 localizes to dystrophic neurites following intracerebral ventricular injection in 5XFAD and APP-NLGF mice. A-D Cortical sections from 5XFAD (A) and APP-NLGF (B) mice that received a single intracerebral ventricular (ICV) injection of recombinant annexin A6-HIS (3.3 mg/kg) or A6-GFP (0.9 mg/kg; C 5XFAD; D NLGF) and brains harvested 3 h later. Sections were immunostained for α-HIS (red), NeuN (blue), MeX04 (amyloid stain, blue), BACE1 (red), LAMP1 (green), and Aβ (3D6, blue). GFP fluorescence of A6-GFP was imaged in (C) and (D). Both recombinant A6-HIS and A6-GFP localized to membrane puncta and whole membranes on BACE1+ and LAMP1+ dystrophic neurites (arrows) in 5XFAD and APP-NLGF mice, indicating the presence of membrane damage. *, plaque cores; Bars=10μm. Representative from n = 2–5 mice per genotype, and protein injection. E Low magnification images of brain section from 5XFAD mouse after intracerebral ventricular (ICV) injection of recombinant annexin A6-HIS (3.3 mg/kg), immunostained with MethoxyX04 (blue) for amyloid, α-HIS (red), LAMP1 (green), and Iba1 (white) to demonstrate the spread of A6-HIS after three hours. White boxes indicate approximate areas at edge of injection spread selected for 60x confocal imaging in similar sections for A through D. Scale bar = 1000 μm