Intraneuronal aggregation of the β-CTF fragment of APP (C99) induces Aβ-independent lysosomal-autophagic pathology

Abstract

Endosomal-autophagic-lysosomal (EAL) dysfunction is an early and prominent neuropathological feature of Alzheimers's disease, yet the exact molecular mechanisms contributing to this pathology remain undefined. By combined biochemical, immunohistochemical and ultrastructural approaches, we demonstrate a link between EAL pathology and the intraneuronal accumulation of the β-secretase-derived βAPP fragment (C99) in two in vivo models, 3xTgAD mice and adeno-associated viral-mediated C99-infected mice. We present a pathological loop in which the accumulation of C99 is both the effect and causality of impaired lysosomal-autophagic function. The deleterious effect of C99 was found to be linked to its aggregation within EAL-vesicle membranes leading to disrupted lysosomal proteolysis and autophagic impairment. This effect was Aβ independent and was even exacerbated when γ-secretase was pharmacologically inhibited. No effect was observed in inhibitor-treated wild-type animals suggesting that lysosomal dysfunction was indeed directly linked to C99 accumulation. In some brain areas, strong C99 expression also led to inflammatory responses and synaptic dysfunction. Taken together, this work demonstrates a toxic effect of C99 which could underlie some of the early-stage anatomical hallmarks of Alzheimer's disease pathology. Our work also proposes molecular mechanisms likely explaining some of the unfavorable side-effects associated with γ-secretase inhibitor-directed therapies.

Keywords: Aggregation; Alzheimer; Autophagy; C99; Lysosomes; Triple-transgenic mouse; γ-Secretase inhibition.

Fig. 1

C99 accumulates in enlarged cathepsin B- and lamp1-positive structures. a Co-immunohistochemical staining of C99 with 82E1 (green) and α-cathepsin B (red) in 3-month-old 3xTgAD mice. Note the co-localization (merged image with DAPI staining) of the two labelings and the presence of enlarged (arrowheads) cathepsin B structures in 82E1-positive cells as compared to normal-sized cathepsin B structures (arrows) in 82E1-negative cells. Scale bar 20 μm. b Co-immunohistochemical staining with FCA18 (red) and α-lamp1 (green) shows the localization of the FCA18-associated staining within enlarged lamp1 structures in both 14-month-old 2xTgAD (2AD) and 3xTgAD (3AD) mice. Scale bar 10 μm. c Electron microphotographs were taken from the subiculum of 14-month-old non-transgenic (nonTg), 2AD or 3AD mice. Upper panels correspond to low-magnification images of representative neurons from nonTg, 2AD or 3xTgAD mice (scale bar 5 μm). The lower images illustrate examples of lysosomal-dense bodies (<0.2 μm, left image), or of bigger membrane-limited structures of homogenous density corresponding to autolysosomes (>0.5 μm, middle panels) and giant autolysosomes (>1.5 μm, middle and right panels)

Fig. 2

In vitro analysis shows that APP-CTFs are degraded by cathepsins through autophagy. a, b SH-APPswe cells were treated with NH₄Cl, leupeptin (Leu), pepstatin A (PepA) or PADK and analyzed by western blot using α-APPct. Arrows correspond to a non-identified 25 kDa APP-CTF fragment. Bars in b correspond to quantification of C99 and C83 immunoreactivities obtained in a and expressed relative to expressions measured in DMSO-treated cells normalized to actin. Data are represented as mean ± SEM, as determined by ANOVA one-way Dunnett post hoc test, **p < 0.01 and *p < 0.05 indicate significant differences relative to control cells (n = 10–16, from four independent experiments). c Immunocytochemical analysis using α-APPct and α-lamp1 on H₂O or NH₄Cl treated SH-APPswe cells. Note the labeling of α-APPct in enlarged α-lamp1 vesicles in merged images (arrow). d SH-APPswe cells were infected with lenti-cathepsin B (CatB) or empty vector (mock) virus at different concentrations (1, 5 or 10 μl of a stock corresponding to 3.75 × 10⁸ TU/ml) and CatB and APP-CTF levels were analyzed after 48 h by western blot. α-cathepsin B revealed immature (proCatB) and mature (mCatB) cathepsin B. e, f CatB and APP-CTF levels were analyzed by western blot in CatB stable cell lines after six passages. Bars in f represent the quantification of mCatB and C83 and C99 immunoreactivities obtained in e, each expressed as respective expressions measured in mock cells. Data are represented as mean ± SEM, Mann–Whitney, ***p < 0.001 and **p < 0.01 (n = 7 from three independent experiments). g Recombinant C100-flag was incubated in the absence or presence of increasing concentrations of purified CatB at 37 °C during 0 (To, 50 ng) or 60 min (5,10 or 50 ng) and C100 levels were detected by western blot using α-Flag antibody. h, i SH-APPswe cells were treated with smer-28 (smer), PADK or bafilomycin A1 (BafA1), and APP-CTF levels were analyzed by western blot using α-APPct. Arrow corresponds to a non-identified 25 kDa APP-CTF fragment. Bars in i correspond to quantification of C83 and C99 immunoreactivities obtained in h and expressed as the percentage of the expressions in DMSO-treated cells normalized to actin. Data are represented as mean ± SEM, as determined by ANOVA one-way Dunnett post hoc test, **p < 0.01 and *p < 0.05 indicate significant differences relative to control cells n = 8, from three independent experiments

Fig. 3

In vivo treatment of 3xTgAD mice with the γ-secretase inhibitor ELND006 (D6) leads to increased APP-CTFs localized to large cathepsin B-positive structures and to autophagic impairment. 5-month-old non-transgenic (nonTg) or 3xTgAD males were treated daily with D6 or vehicle by oral gavage during 1 month. a Immunostaining with FCA18 in the subiculum (see insert) of 3xTgAD vehicle- (AD-CT) or D6-treated (AD-D6) mice. Scale bar 125 and 20 μm, respectively. b, c RIPA-soluble (sol.) fractions from hippocampi of AD-CT or AD-D6 mice were analyzed for βAPP, APP-CTF, LC3-I, LC3-II and p62 expression by western blot. LC3-II was revealed after long time exposure of the same blot revealed for LC3-I. RIPA insoluble acid formic retrieved fractions (is) were analyzed for APP-CTF expression. Bars in c, are the mean ± SEM of 12 animals of each treatment and represent the quantification of βAPP, C83, C99 and p62 immunoreactivities expressed as the percentage measured in AD-CT mice normalized to actin, and the LC3-II to LC3-I ratio normalized to AD-CT. Statistical analysis was performed by Mann–Whitney and ***p < 0.001 and **p < 0.01 indicate significant differences relative to AD-CT. d, e RIPA-soluble fractions from hippocampi of non-transgenic vehicle- (nonTg-CT) or D6-treated (nonTg-D6) mice were analyzed for APP-CTF, LC3-I, LC3-II and p62 expression by western blot. Right lane corresponds to AD-D6 mice. Bars in e are the mean ± SEM of six animals of each treatment and represent the quantification of C83 and p62 immunoreactivities expressed as the percentage measured in nonTg-CT mice. Statistical analysis was performed by Mann–Whitney and ***p < 0.001 indicate significant differences relative to nonTg-CT. f Images correspond to co-immunohistochemical staining with 82E1 and α-cathepsin B in subiculum of AD-D6 mice. Scale bar 20 μm

Fig. 4

In vivo treatment of 3xTgAD mice with the γ-secretase inhibitor ELND006 (D6) leads to autophagic pathology and intraneuronal damage. Electron microphotographs of neuronal somas and neuropil from 5-month-old 3xTgAD males treated with D6 (AD-D6) (b, d, e, h) or vehicle (AD-CT) (a, c) during 1 month. Neuronal perikarya of AD-CT and AD-D6 mice contained dense lysosomal bodies (blue arrows) (a, b) and lipid-containing autophagic vacuoles (red arrows) (a), that are more abundant and enlarged in AD-D6 mice (red arrows) (b, e, h). The AD-D6 mice also displayed multiple large vesicles filled with heterogenous material (d–h—red arrowheads) and multilamellar bodies (d, g—blue ML).The neuropil of AD-CT mice presented many normal appearing synaptic contacts characterized by typical synaptic post-densities (PSDs, yellow arrows in c) and normal mitochondria (black arrows). In contrast, the neuropil in AD-D6 mice displayed very few normal appearing synaptic contacts (d—yellow arrows), many damaged mitochondria (black arrows, f, g) and electron-lucent areas (h—blue stars). Scale bars are 2 μm in c, d, e, g, 5 μm in a, b, f, and 10 μm in h. i Quantification of autophagic structures in slices from AD-CT (black bars) and AD-D6 (grey bars) mice. The autophagic structures (AVs) were classified in three (1–3) distinct groups, 1 corresponding to small dense AVs (blue arrows, less than 1 μm), 2 to a mix of larger sized and more or less lipid-containing AVs (red arrows) and 3 to large vesicles containing membranous material and multilamellar bodies (red arrowheads). Bars correspond to the average number of AVs per neuron (per cross section) and a count of 40–50 neurons per mouse (2 mice for each condition). Data are represented as mean ± SEM and statistical analysis was performed using the Mann–Whitney test and ***p < 0.001 indicate significant differences relative to AD-CT

Fig. 5

γ-Secretase inhibitor treatment triggers autophagic dysfunction in APPswe-expressing SHSY-5Y but not mock-transfected cells. a SH-APPswe cells were treated with ELND006 (D6) and analyzed for APP-CTF levels by western blot. APP-CTFs were detected using α-APPct, 4G8 or 6E10 (see diagram for epitope recognition) and compared to C99-flag expression in HEK transfected cells. b Co-immunocytochemical staining of vehicle (Veh) or D6-treated APPswe cells using α-APPct and α-lamp1 shows a high overlap of these labelings. Scale bar 2 μm. c Vehicle (V) or D6-treated SH-mock or SH-APPswe cells were stained with Lysotracker red and DAPI. Scale bar 5 μm. d Cathepsin B (CatB) activity was monitored in vitro from microsomal fractions of SH-APPswe cells treated with vehicle (V), D6, DAPT (DT) or NH₄Cl (NH4). Data are represented as mean ± SEM, as determined by ANOVA one-way Dunnett post hoc test, ***p < 0.001 indicate significant differences relative to control cells (vehicle-treated cells), n = 8–14 from 2 independent experiments. e, f Stably CatB expressing mock or APPswe cells were treated with vehicle (V), D6, DAPT (DT) or NH₄Cl and western blot analysis was used to determine the levels of βAPP and APP-CTFs, LC3-I, LC3-II, p62, CatB and actin. Besides the detection of the 30 kDa immunoreactivity corresponding to mature CatB, the CatB antibody revealed immunoreactivity at about 12 kDa most probably corresponding to a degradation product of this enzyme (CatB-dg.). f Quantification of immunoreactivities obtained in e for APPswe cells were normalized to actin and expressed as the percent of expression levels obtained in control treated cells. g SH-mock or SH-APPswe cells were transfected with LC3-GFP, treated with vehicle or D6 and was visualized by fluorescence microscopy. Scale bar 5 μm. h Electron microphotographs show representative cells from each condition. Scale bars are 1 or 2 μm, as indicated

Fig. 6

In vivo adeno-associated viral vector (AAV)-mediated expression of C99 in wild-type mice leads to the accumulation of aggregated C99 in enlarged cathepsin B-positive structures. 2-month-old AAV-empty, AAV-GFP or AAV-C99 were analyzed by western blot (a) or immunohistochemistry (b). a RIPA-soluble fractions from forebrain hemispheres were analyzed for APP-CTF expression using either α-APPct or 6E10. b GFP fluorescence was observed in AVV-GFP mice using a confocal microscope and C99 expression was visualized after immunohistochemical staining with α-APPct and using AAV-empty mice as a negative control. The lower panel shows higher magnification images at the level of the boxed areas highlighting the different subcellular labeling of GFP and C99. c Bars correspond to ELISA measurement of Aβ40 and Aβ42 in soluble (sol) and insoluble (is) fractions, respectively, and are represented as mean ± SEM n = 6 animals. d Images from immunohistochemical analysis of C99-associated expression in C99-AAV mice at the level of the subiculum using α-APPct, FCA18, 6E10, 4G8 or NU1. Scale bar 25 μm. e High-magnification images of co-immunolabeling with α-APPct and 4G8 illustrating the intraneuronal membrane-associated (arrow) and punctiform (arrowhead) stainings. Scale bar 5 μm. f, g Co-labeling of 82E1 and cathepsin B (f) or 4G8 and lamp1 (g) shows that C99 localizes to enlarged EAL-associated structures (arrowhead in merged images). Scale bar 25 and 5 μm, respectively. h Quantitative analysis of lamp1 structures in the subiculum of AAV-empty or AAV-C99 mice. Data are represented as the relative number per cell (counts (a.u)/cell) and average size of lamp1-positive puncta expressed as arbitrary units (a.u). Data are represented as mean ± SEM and are from two AAV-empty and two AAV-C99 mice, three brain slices for each animal and six images each slice. Images show representative images of lamp1 staining

Fig. 7

γ-Secretase inhibitor treatment of AAV-C99 expressing mice leads to increased accumulation of aggregated APP-CTFs within lysosomal-autophagic vesicles. AAV-C99 and AAV-empty injected mice were treated daily for 12 days with ELND006 (D6) or vehicle (V). a RIPA-soluble fractions of forebrain hemispheres were analyzed by western blot for APP-CTF expression using α-APPct or 6E10. b ELISA analysis of Aβ40 levels in RIPA-soluble- (sol) and formic acid retrieved (is) hemisphere fractions. Data are represented as mean ± SEM. Statistics were performed with the Mann–Whitney test by comparing soluble and insoluble fractions separately,**p < 0.01 (n = 6). c Images from immunofluorescence staining of brain slices from vehicle- (AAV-C99V) or D6-treated (AAV-C99 D6) AAV-C99 mice at the level of the subiculum using either NU1 or α-APPct. The lower panels correspond to high-magnification of the boxed areas. Scale bar is 125 and 20 μm, respectively. d Immunohistochemical staining with 4G8 of brain slices from AAV-C99V or AAV-C99 D6 mice using peroxidase/DAB development (brown staining). Scale bar is 20 μm. e Co-immunostaining of α-APPct and α-synaptophysin (α-Syn) revealed a high overlap within the subiculum of AAV-C99-D6 mice, but not in D6-treated AAV-empty mice (left panel). Scale bar is 5 μm. f Immunohistochemical staining of brain slices of AAV-C99 V (upper panels) or AAV-C99 D6 (lower panels) mice using α-APPct, 4G8 and NU1. Middle and right panels correspond to medium- and high-magnification images of NU1 labeling at the level of the boxed areas. Note the vesicular membrane-like staining of NU1 in right panel (arrowheads). Nuclei were stained with DAPI. Scale bar is 2.5 μm

Fig. 8

γ-Secretase inhibitor treatment of AAV-C99 expressing mice leads to exacerbated autophagic dysfunction, inflammatory responses and synaptic dysfunction. a Brain slices at the levels of the subiculum from vehicle- (Veh) or D6-treated AAV-C99 mice were immunostained with 82E1 and α-CatB. Scale bar 25 μm. b–d Western blot analysis of RIPA-soluble fractions from AAV-C99 injected mice (b, c) or AAV-GFP mice (d) treated with D6 or vehicle (Veh). Brains were analyzed for APP-CTF (using either α-APPct or 6E10) or LC3-I/LC3-II expression. Bars in c correspond to the quantitative analysis of LC3-II, expressed as the LC3-II/LC3-I ratio, and are relative to control (AAV-C99-Veh). Mann–Whitney test, p < 0.001, n = 8. Right lane corresponds to AAV-C99 mice. e, f Immunostaining with 4G8 or NU1 of D6-treated AAV-C99 brains using fluorescence (e) or peroxidase-DAB labeling (f). Note the extracellular staining (white arrows in e and black arrows in f) surrounding pycnotic nuclei visualized by DAPI (e, white arrowheads) or Cresyl violet (f, black arrowheads). Scale bars 50 or 10 μm, respectively. g, h Immunostaining with 4G8, Iba1 (g) or GFAP (h) reveals both microglial and astrocytic activation in D6-treated mice within brain regions expressing important APP-CTF levels. Scale bar 300 and 50 μm, respectively. i Basal synaptic transmission in AAV-empty, vehicle-treated AAV-C99 or D6-treated AAV-C99 mice (n = 2–3 slices per mouse from 6 to 7 mice per group. All values are mean ± SEM. j Field excitatory postsynaptic potential (fEPSP) slopes in subiculum (n = 2–3 slices per mouse from 6 to 7 mice per group). k Summary graph of LTP magnitudes calculated 40 to 60 min after high-frequency stimulation from graphs in (j) with statistical analysis (*p < 0.05; one-way ANOVA with the Tukey’s post hoc test). Error bars represent SEM