Mitochondrial structure and function are altered in a cellular model of familial AD
We analyzed mitochondrial structure and function in the neuroblastoma cells expressing the Swedish familial double mutations (APPKM670/671NL: SH-SY5Y APPswe). These cells exacerbate the production of APP-CTFs (C99 and C83 derived from the cleavage of APP by β and α secretases, respectively), and amyloid beta (Aβ) peptides (produced through the subsequent cleavage of C99 by γ-secretase) [59]. We examined mitochondria morphology and size by transmission electron microscopy. We first observed that APPswe-expressing cells harbor larger size mitochondria with altered cristae organization than control cells (colored arrowheads in Fig. 1a). We analyzed in depth mitochondrial alterations and classified mitochondria morphology in four categories (class I: fairly dark mitochondria, with uniform matrix filled with dense packed regular distributed cristae; class II: mitochondria with disrupted cristae and loss of matrix density; class III: empty mitochondria with disorganized cristae, or cristae on the periphery; and class IV: swollen mitochondria with disrupted membrane) (Fig. 1b). Quantification of mitochondria subclasses was then recorded revealing that while control cells exhibited 86% of mitochondria class I, 10% class II, 3% class III, and 1% class IV (Fig. 1c and suppl. Table 3, online resource), APPswe cells displayed a drastic reduction of “healthy” mitochondria class I (29%), and a concomitant enhancement of mitochondria class II (43%), III (12%), and IV (16%) (Fig. 1c and suppl. Table 3, online resource). We then showed that APPswe cells also displayed larger mitochondria (increased area and perimeter) (Fig. 1d, e and suppl. Table 3, online resource), with a significant decrease of mitochondrial number (Fig. 1f and suppl. Table 3, online resource) as compared to control cells.
Mitochondria are dynamic organelles requiring appropriate finely tuned equilibrium between fission and fusion. To corroborate structural alterations observed in APPswe cells, we measured the levels of mitochondrial fission protein (DNM1L/DRP; dynamin 1-like) and fusion proteins (MFN1; mitofusin 1, and MFN2; mitofusin 2) at the mRNA and protein levels (Suppl. Figure 1a-e, online resource). While we did not reveal a modulation of DRP1, MFN1, and MFN2 mRNAs levels between APPswe cells and their controls (Suppl. Figure 1a, online resource), the protein levels of DRP1, MFN1, and MFN2 were significantly lower in APPswe cells in both total and mitochondria-enriched protein extracts (Suppl. Figure 1b-e, online resource). These data show molecular perturbations of mitochondrial fission/fusion equilibrium supporting mitochondria morphological phenotypes taking place in APPswe cells.
Mitochondria structure alterations observed in APPswe cells are accompanied by several mitochondrial dysfunctions as demonstrated by a reduction of mitochondrial complex I NDUFB8 subunit expression (Fig. 1g, h), while the levels of mitochondrial complexes subunits II, III, IV, and V remained unchanged (Fig. 1g and suppl. Figure 1f, online resource). This was corroborated by a specific reduction of mitochondrial complex I activity (Fig. 1i), but not of complexes II, III, IV, and V (Suppl. Figure 1g-j, online resource) in APPswe cells. The Krebs cycle citrate synthase enzyme remained also unchanged between APPswe cells and their controls (Suppl. Figure 1k, online resource). In agreement with a dysfunctional respiratory chain complex I, APPswe cells showed a reduction of mitochondrial membrane potential, revealed by reduced TMRM fluorescence intensity (Fig. 1j, k).
APP-CTFs accumulation triggers mitochondrial structure alterations in cells independently of Aβ
Several studies reported that mitochondrial structure and function alterations are linked to Aβ [15, 34, 46]. We aimed hereafter at investigating the specific contribution of APP-CTFs versus Aβ to mitochondrial structure and function alterations reported in our APPswe cellular model. Thus, we pharmacologically targeted γ-secretase, the inhibition of which blocks Aβ peptide formation and enhances CTFs’ recovery (as we already reported [13, 45]). Immunological analyses with 6E10 antibody (revealing full-length APP, C99, and Aβ) or APP-Cter antibody (revealing C99, and C83, and APP intracellular domain (AICD) (Fig. 2a) confirmed that γ-secretase inhibition enhanced the accumulation of APP-CTFs (C99 and C83) and blocked Aβ peptide production in total extracts (Fig. 2a, b) as well as in mitochondria-enriched fraction (Fig. 2a, c) without affecting the level of full-length APP (Fig. 2a–c). Unexpectedly, γ-secretase inhibition did not significantly modify AICD levels in both total homogenates and mitochondrial-enriched fraction (Fig. 2a–c). We further demonstrated by immunofluorescence analyses, using the two sets of antibodies (6E10 and APP-Cter), that APP and APP-CTFs colocalize with the mitochondrial protein HSP60 in untreated and in γ-secretase inhibitor-treated APPswe cells (Fig. 2d, e). These biochemical and imaging analyses firmly demonstrate that APP-CTFs are present in mitochondria compartment, thus questioning their contribution to mitochondrial structure alterations observed in this AD cellular model.
We took advantage of this pharmacological approach and analyzed mitochondria structure using electron microscopy, and revealed that γ-secretase inhibitor-treated APPswe cells display spherical and fragmented mitochondria (colored arrowheads in representative images in Fig. 2f). The quantification of mitochondria subclasses showed that γ-secretase inhibitor-treated APPswe cells harbor a loss of mitochondria class I (1%) and swollen mitochondria class IV (2%) with a drastic shift of mitochondrial shape towards mitochondrial classes II and III (47%, and 50%, respectively) (Fig. 2g and suppl. Table 3, online resource). This mitochondrial morphology shift was accompanied by a significant reduction in mitochondria size (area and perimeter) (Fig. 2h, i and suppl. Table 3, online resource), and an increase of mitochondria number in γ-secretase inhibitor-treated APPswe cells (Fig. 2j and suppl. Table 3, online resource). We further strengthened these observations by analyzing mitochondrial three-dimensional (3D) structure in living cells using mitotracker dye and confocal imaging (suppl. Figure 2a, online resource). Thus, γ-secretase inhibition increases mitochondrial number (Suppl. Figure 2a, b, online resource), and reduces mitochondrial network 3D volume (Suppl. Figure 2a, c, online resource). As observed in APPswe untreated cells (Suppl. Figure 1a, online resource), mitochondrial size alteration observed upon γ-secretase inhibition is not corroborated by a modulation of DRP1 and MFN2 mRNA levels (Suppl. Figure 2d, online resource). Nevertheless, we reported a significant increase of MFN2 protein level in APPswe cells treated with γ-secretase inhibitor, while DRP1 protein level remained unchanged (Suppl. Figure 2e, online resource).
We further investigated the contribution of APP-CTFs versus Aβ to mitochondria structure alterations by analyzing APPswe cells treated with β-secretase inhibitor, blocking the production of Aβ, reducing APP-βCTF (C99) level, and enhancing the level of APP-αCTF (C83) (Suppl. Figure 3a-c, online resource). As for γ-secretase inhibitor, we also did not notice any significant change of AICD level upon β-secretase inhibition (Suppl. Figure 3a-c, online resource). Importantly, we revealed by electron microscopy analyses that β-secretase inhibition triggers a recovering of mitochondria class I morphology (76%) and a reduction of mitochondria classes II, III, and IV (13%, 7%, and 4%, respectively, suppl. Figure 3d, e, online resource, and suppl. Table 3, online resource). Thus, the comparative analyses of the impact of β- versus γ-secretase inhibitors on mitochondria structure demonstrate that both Aβ and APP-CTFs participate to mitochondrial structure alterations in APPswe-expressing cells and that γ-secretase-mediated APP-CTFs accumulation independently from Aβ exacerbate mitochondrial morphological alterations specifically characterized by cristae disorganization (mitochondria classes II and III) and changes in mitochondria size and number.
We also investigated the impact of the accumulation of endogenous APP-CTFs on mitochondrial structure upon γ-secretase inhibition in mock (pcDNA3.1)-transfected SH-SY5Y cells. We revealed low amount of endogenous APP-CTFs in both total and mitochondrial fraction of mock-transfected SH-SY5Y cells (as compared to APPswe cells) correlating with a low-level expression of full-length APP (Suppl. Figure 4a, online resource). We also noticed the accumulation of endogenous APP-CTFs in control cells upon γ-secretase inhibitor treatment both in total and mitochondria-enriched fraction (Suppl. Figure 4b, online resource). Immunofluorescence analyses support this finding and show enhanced signal detected with APP-Cter antibody colocalizing with mitochondrial protein HSP60 in γ-secretase inhibitor-treated-control cells (Suppl. Figure 4c, online resource). We then revealed that endogenous APP-CTFs accumulation led to mitochondria structure alteration (66% class II, and 30% class III) (Suppl. Figure 4d, e, online resource, and suppl. Table 3, online resource), reduced size (Suppl. Figure 4d, f, g, online resource, and suppl. Table 3, online resource), and increased number (Suppl. Figure 4h, online resource, and suppl. Table 3, online resource). Thus, γ-secretase inhibition that triggers accumulation of endogenous or overexpressed APP-CTFs leads to similar structural mitochondria alterations in cells (Fig. 2f–j, suppl. Figure 4d-h, online resource, and suppl. Table 3, online resource).
Mitochondrial function and apoptosis are differently impacted by Aβ and APP-CTFs
We investigated the consequences of mitochondrial APP-CTFs accumulation on mitochondrial function. We first showed that both γ-secretase and β-secretase inhibition enhanced the expression level of NDUFB8 complex I subunit (Fig. 3a, b and suppl. Figure 3f, online resource), and accordingly the activity of respiratory complex I (Fig. 3c and suppl. Figure 3g, online resource), without altering the expression levels (Suppl. Figure 5a, online resource) or activities (Suppl. Figure 5b-e, online resource) of mitochondrial complexes II, III, IV, and V. Citrate synthase enzyme remained also unchanged upon γ-secretase inhibition (Suppl. Figure 5f, online resource). Similarly, we did not notice any change in mitochondrial complexes II, III, IV, and V expression and activities upon β-secretase inhibition (data not shown). We then reported that reduced mitochondrial potential in untreated APPswe cells (compared to their controls) was unaffected by γ-secretase inhibitor (Fig. 3d and representative histograms in suppl. Figure 5g, online resource), but was restored by β-secretase inhibitor (suppl. Figure 3h, online resource). However, we observed enhanced mitochondrial ROS accumulation in APPswe versus control cells that was further exacerbated upon γ-secretase inhibition (Fig. 3e and suppl. Figure 5h, online resource), and unchanged upon β-secretase inhibition (data not shown).
We then quantified caspase-3 activation known as an executioner caspase in mitochondrial-dependent apoptosis and used fluorometric and immunoreactivity approaches [18]. We first showed higher sensitivity to a sub-apoptotic dose of staurosporine in APPswe cells as compared to control cells as illustrated by enhanced caspase-3 activity and cleaved caspase-3 (c-Casp 3: active form) (Fig. 3f–h). Although γ-secretase inhibitor did not impact caspase-3 activity in basal conditions (Fig. 3f), it triggers a significant reduction of STS-induced caspase-3 activity (Fig. 3f), and cleavage (Fig. 3i, j) in APPswe cells. We also confirmed the reduction of caspase-3 activity with two different γ-secretase inhibitors DFK, and DAPT in STS-treated APPswe cells (Suppl. Figure 5i, online resource). Similarly, we also observed a reduction of STS-induced caspase-3 activity upon treatment with β-secretase inhibitor (Suppl. Figure 3i, online resource).
We ascertain the specificity of γ-secretase inhibitor towards APP substrate using SH-SY5Y control cells and mouse embryonic fibroblasts (MEF) isolated from control mice (MEF-WT) or from mice genetically invalidated for both APP and its additional family member APLP2 gene (MEF-APPKO) (Suppl. Figure 6, online resource). We showed that γ-secretase inhibitor: (1) did not significantly change mitochondrial potential in SH-SY5Y control cells (Suppl. Figure 6a, online resource), and in MEF-WT and MEF-APPKO cells (Suppl. Figure 6b, online resource); (2) did not change STS-induced caspase-3 activity in MEF control and APPKO fibroblasts (Suppl. Figure 6e, online resource). Importantly, we reported increased mitochondrial ROS in SH-SY5Y control cells (Suppl. Figure 6c, online resource) and in MEF-WT cells but not in MEF-APPKO cells (Suppl. Figure 6d, online resource) treated with γ-secretase inhibitor. Thus, γ-secretase inhibitor-induced mitochondrial deleterious phenotype appears to be APP-dependent.
Overall, these results revealed the contribution of both Aβ and APP-CTFs to mitochondrial dysfunctions. The comparative analyses of the impact of γ-secretase versus β-secretase inhibitors in APPswe cells emphasized the role of Aβ in mitochondrial respiratory chain complex I dysfunction, mitochondrial membrane potential loss, and caspase-3 activation and that of APP-CTFs accumulation in mitochondrial ROS elevation.
To further explore the contribution of APP-derived CTFs to mitochondrial dysfunctions, we used the neuroblastoma stable and inducible cell line expressing C99 fragment only [45]. First, as we previously described [29], we observed that C99 undergoes a γ-secretase-mediated cleavage giving rise to a high C83 expression (Fig. 4a, b). We also showed enhanced C99 and C83 fragments in mitochondrial-enriched fraction prepared from C99-expressing cells as compared to mock-transfected cells (control) (Fig. 4a, b). As observed in APPswe cells (Fig. 2a–c), the treatment with γ-secretase inhibitor enhanced C83 level in both C99 and control cells (Fig. 4a, b). We further demonstrated that both endogenous and overexpressed APP-CTFs colocalize with mitochondrial HSP60 protein as illustrated in C99 cells untreated and in control and C99 cells treated with γ-secretase inhibitor (Fig. 4c). We next examined whether C99 expression triggers mitochondrial dysfunction. Indeed, like APPswe cells, C99-expressing cells harbor a reduction of the expression of NDUFB8 (Fig. 4d, e) and of the activity of respiratory chain complex I (Fig. 4f), without altering the mitochondrial complexes II, III, IV, and V subunit expression and their activities (Suppl. Figure 7a–e, online resource), and citrate synthase activity (Suppl. Figure 7f, online resource). C99 overexpression triggers an increase of mitochondrial ROS production (Fig. 4g), and intriguingly enhanced mitochondrial membrane potential (Fig. 4h), both remained elevated upon γ-secretase inhibitor treatment. We also revealed that C99-expressing cells harbor a slight but not-significant increase in staurosporine-induced apoptosis that remained not modulated by γ-secretase inhibitor (Fig. 4i). All over, these data demonstrate the colocalization of overexpressed C99 with mitochondria triggering mitochondrial dysfunctions, and that APP-CTFs accumulation per se did not lead to apoptotic cell death.
APP-CTFs accumulation leads to basal mitophagy failure in cells independently of Aβ
Mitochondrial fragmentation and enhanced ROS production are considered major paradigms for selective elimination of superfluous or dysfunctional mitochondria by a specific autophagy process known as mitophagy [64]. We thus questioned the putative contribution of APP-CTFs accumulation to mitophagy. We examined the induction of basal autophagy through the conversion of soluble microtubule-associated protein 1A/1B-light chain 3 (LC3-I) to lipid-bound LC3-II and quantified their expression levels and that of the autophagy substrate SQSTM1/p62 (named p62 hereafter). We specifically quantified LC3 conversion and p62 level in mitochondrial fraction. Our data show a slight but not significant increase of LC3-I but a drastic enhancement of LC3-II level and LC3-II/LC3-I ratio, indicating a basal autophagy induction in APPswe cells (Fig. 5a, b). We further measured mitophagy using LC3-GFP probe and fluorescence microscopy, and confirmed the conversion of LC3-I to LC3-II through aggregation and colocalization of LC3-GFP with HSP60 mitochondrial protein in APPswe cells (Fig. 5e). Paradoxically, the level of p62 remained unchanged (Fig. 5a, b). The increase in LC3-I conversion, LC3-II accumulation, and unchanged p62 levels supports enhanced basal autophagy induction and a blockade in downstream degradation. We then analyzed mitochondrial mitophagy priming occurring through PINK1/Parkin pathway [87]. We observed significantly enhanced levels of Parkin and PINK1 in mitochondrial fraction isolated form APPswe as compared to control cells (Fig. 5a, c). Increased Parkin recruitment to mitochondria was further confirmed by immunofluorescence through its colocalization with mitochondrial HSP60 protein in APPswe cells (Fig. 5f). Accordingly, we also showed a slight increase of mitochondrial localization of phospho-poly-ubiquitin (p-S65-Ub: p-Ub), a PINK1 substrate recruited by Parkin to mitochondria [87] in APPswe cells (Fig. 5f). As control, we analyzed protonophore carbonyl cyanide m-chlorophenyl hydrazone (CCCP)-induced mitophagy and showed enhanced Parkin and p-Ub localization within mitochondria in both control and APPswe-treated cells (Fig. 5f).
Activated mitophagy process is reflected by enhanced degradation of mitochondrial outer and inner membrane and matrix proteins. SDS-PAGE analyses performed on mitochondria-enriched fraction showed increased levels of several mitochondrial proteins (TOMM20, TIMM23, HSP60, and HSP10) in APPswe cells (Fig. 5a, d), suggesting a compromised targeting of dysfunctional mitochondria to lysosomal compartment and/or its defective degradation. Indeed, we observed reduced colocalization of mitochondria with lysosomes in APPswe versus control cells as illustrated through the analyses of the colocalization of mitochondrial Mit-RFP probe with lysosomal LAMP1-GFP probe (Fig. 5g). Furthermore, we used Cox8-EGFP-mCherry mitophagy reporter, [71] (Fig. 5h left) and reported a not statistically different number of cells harboring red mitochondria puncta in APPswe and control cells, thus attesting for defective mitochondrial engulfment in lysosomal acidic compartment (Fig. 5h, i). We validated Cox8-EGFP-mCherry mitophagy reporter by demonstrating enhanced mitophagy (cells harboring red puctae) in control and APPswe cells treated with the iron chelator DFP (deferiprone) (a potent mitophagy inducer in SH-SY5Y cells [1] (Fig. 5h, i), and with a combination of oligomycin A and antimycin A in APPswe cells [1] (Suppl. Figure 8a, online resource).
Importantly, pharmacological blockade of γ-secretase in APPswe cells definitely demonstrates that mitophagy failure could be accounted by APP-CTFs accumulation (Fig. 6a–e). Hence, APPswe-treated cells showed a huge increase of LC3-II level and LC3-II/LC3-I ratio, unchanged levels of p62 in mitochondria-enriched fraction (Fig. 6a, b), and enhanced total p62 fluorescence (Suppl. Figure 8b, online resource), attesting for basal autophagy induction and degradation defect. APPswe cells treated with γ-secretase inhibitor also showed defective mitophagy priming with a slight but non-statistically significant increase of Parkin and PINK1 levels in mitochondria (Fig. 6a, c). Unchanged Parkin and p-Ub targeting to mitochondria were further showed by immunofluorescence analyses (Suppl. Figure 8b, online resource). Accordingly, a significant increase of mitochondrial proteins (TOMM20, TIMM23, HSP60, and HSP10) (Fig. 6a, d) in mitochondria-enriched fraction suggested a compromised targeting of dysfunctional mitochondria to lysosomes and their degradation in γ-secretase inhibitor-treated APPswe cells. Indeed, we revealed unchanged colocalization of Mit-RFP probe with LAMP1-GFP probe (Suppl. Figure 8c, online resource), and confirmed unchanged number of mitochondria in fusion with lysosomes using Cox8-EGFP-mCherry probe (Fig. 6e).
Importantly, γ-secretase inhibitor-mediated accumulation of endogenous APP-CTFs was also associated with mitophagy failure in SH-SY5Y control cells characterized by enhanced LC3-II/LC3-I ratio unchanged levels of mitochondrial p62 (Suppl. Figure 9a, b, online resource), enhanced total p62 staining (Suppl. Figure 9d, online resource), unchanged Parkin, and p-Ub (Suppl. Figure 9a, c, and d, online resource), and a significant increase of TIMM23, and HSP10 expressions (Suppl. Figure 9a, c, online resource). We confirmed this mitophagy failure phenotype by showing unchanged colocalization of Mit-RFP probe with LAMP1-GFP probe upon γ-secretase inhibitor treatment (Suppl. Figure 9e, online resource).
Accordingly, we validated mitophagy defect in C99 cellular model and showed a slight but no-statistically significant increase of LC3-II level and LC3-II/LC3-I ratio (Fig. 6f, g) associated with a significant increase of p62 (Fig. 6f, g), unchanged PINK1 and Parkin (Fig. 6f, h), and enhanced TOMM20, and HSP10 expression levels (Fig. 6f–i). Both control and C99-expressing cells treated with γ-secretase inhibitor also showed a mitophagy failure phenotype supported by a significant increase of LC3-II level, and of LC3-II/LC3-I ratio (Fig. 6f, g) associated with unchanged or enhanced p62 (Fig. 6f, g), unchanged PINK1 and Parkin (Fig. 6f, h), enhanced TOMM20, TIMM23, and increase or unchanged HSP10 (Fig. 6f, i). We further showed unchanged levels of Parkin in control and C99-expressing cells treated with γ-secretase inhibitor by immunofluorescence (Suppl. Figure 8d, online resource). Using Cox8-EGFP-mCherry probe, we then demonstrated basal mitophagy defect upon C99 expression or accumulation (Fig. 6j), and revealed as observed in APPswe cells enhanced DFP-induced mitochondria targeting to lysosomal compartment in both control and C99 cells (Fig. 6j).
Interestingly, we also observed unchanged basal autophagy (LC3-I, LC3-II, and p62 levels), mitophagy priming (PINK1, and Parkin), and mitochondrial proteins levels (TOMM20, TIMM23, HSP60, and HSP10) in β-secretase inhibitor-treated APPswe cells, further demonstrating the deleterious effect of accumulated APP-CTFs independently from Aβ on mitophagy failure phenotype (Suppl. Figure 3j, online resource). Overall, these data consistently demonstrate that the accumulation of both endogenous and overexpressed APP-CTFs impaired basal mitophagy.
Mitochondrial structure alterations in AD mice models are linked to APP-CTFs accumulation but independently of Aβ
We already reported early and progressive APP-CTFs accumulation in the subiculum of 3xTgAD mice model [43], and showed that in vivo treatment of young 3xTgAD mice with the γ-secretase inhibitor enhances APP-CTFs accumulation and triggers massive increases in endosome autophagic lysosomes (EAL) accumulation [45]. We took advantage of this model to examine the impact of APP-CTFs accumulation on mitochondrial structure and mitophagy in vivo. Young mice (aged 5 months) were treated daily for 1 month with γ-secretase inhibitor and analyzed for APP-CTFs’ expression/accumulation in mitochondria-enriched fraction. We first show the presence of APP-CTFs in mitochondria of 3xTgAD but not WT mice hippocampi (Fig. 7a). γ-secretase inhibitor unravels APP-CTFs accumulation in WT mice and drastically enhanced it in 3xTgAD mice (Fig. 7a). Mitochondria morphology and size were then analyzed by electron microscopy in the subiculum area of WT and 3xTgAD mice treated with vehicle or with γ-secretase inhibitor (Fig. 7b-e). We constrained our classification in vivo to two mitochondria classes, where class I corresponds to mitochondria with uniform matrix filled with dense packed regular distributed cristae, and class II corresponds to mitochondria showing morphological abnormalities, ranging from focal loss of cristae with empty spaces to severe loss of cristae and matrix (corresponding to mitochondria classes II and III in cells). We did not observe swollen mitochondria class IV in vivo. We first show that vehicle-treated WT and 3xTgAD mice harbor 91% and 79% of class I mitochondria, respectively (Fig. 7c and suppl. Table 3, online resource). Thus, 3xTgAD mice did not show noticeable early alteration of mitochondria cristae shape. However, we observed that mitochondria of 3xTgAD mice harbor a significant reduction of perimeter and area as compared to wild-type mice (Fig. 7d, e and suppl. Table 3, online resource). Importantly, the treatment with γ-secretase inhibitor dramatically and significantly lowered the percentage of class I mitochondria to 19%, and 52%, and enhanced the percentage of class II mitochondria that reached 81% and 48% in 3xTgAD and WT mice respectively (Fig. 7b, c and suppl. Table 3, online resource). γ-secretase inhibitor treatment also enhanced mitochondria perimeter and area in 3xTgAD mice without affecting that of WT mice (Fig. 7d, e and suppl. Table 3, online resource). To confirm that C99 accumulation per se also impacts mitochondria morphology and size in vivo; we compared young (aged 2–3 months) and old (aged 12 months) adeno-associated-virus (AAV)-C99 injected mice previously described [45]. Immunofluorescence analyses using APP-Cter antibody showed intraneuronal expression of C99 in the cortex and the subiculum (Sub) and dentate gyrus (DG) regions of the hippocampus of AAV-C99 injected mice (Fig. 8a). C99 expression was detectable in mitochondrial fractions isolated from both young and old AAV-C99 mice brains (Fig. 8b) and only faintly detectable in AAV-Free-injected mice (Fig. 8b). This observation was corroborated by demonstrating the colocalization of C99 (using APP-Cter antibody) with mitochondrial TIMM23 protein (Fig. 8c). We analyzed mitochondrial morphology in the cell body of cortical neurons and revealed a large number of mitochondria class I in young and old control mice (91% and 96% respectively). In contrast, we noticed a reduction of mitochondria class I in both young and old AAV-C99 injected mice (47% and 52%, respectively) with concomitant increases in class II mitochondria (53% and 48%, respectively) (Fig. 8d, e and suppl. Table 3, online resource). In AAV-C99-injected mice, C99 is also expressed in the hippocampus (Fig. 8a and suppl. Figure 10a, online resource), and triggers reduced class I mitochondria in the cell body of hippocampal neurons (Suppl. Figure 10b, c, online resource). Interestingly, we observed unchanged mitochondria size (area and perimeter) in AAV-C99 versus AAV-Free young mice (Suppl. Figure 10b, d, and e, online resource), but significantly increased size in AAV-C99 old mice as compared to their age-matched control mice (Fig. 8f, g).
The similar data obtained in γ-secretase inhibitor-treated 3xTgAD mice (Fig. 7a–e) and AAV-C99 mice (Fig. 8a–g and suppl. Figure 10b-e, online resource) fully demonstrate that APP-CTFs accumulate in mitochondria and induce early and age-dependent mitochondrial structure alterations.
Basal mitophagy failure in AD mice models is associated with APP-CTFs accumulation
The analysis of mitophagy marker expressions showed enhanced mitochondrial LC3-I and LC3-II and unchanged p62 levels in 3xTgAD mice than in WT mice reflecting both basal autophagy induction (conversion of LC3-I- to LC3II) and impairment of autophagy degradation (LC3-I, LC3-II accumulation and unchanged p62 level) (Fig. 7f, g). Since at 5-month age, 3xTgAD mice display high APP-CTFs levels, while Aβ remained barely detectable [11], we may assume that mitophagy failure is rather associated with APP-CTFs accumulation. Corroborating this statement, we observed increase of LC3-I level and of LC3-II/LC3-I ratio, unchanged p62 level in 3xTg-AD versus WT mice treated with γ-secretase inhibitor (Fig. 7f, g). Accordingly, old but not young AAV-C99-injected mice showed basal autophagy induction and defective autophagy degradation (LC3-I, LC3-II accumulation and unchanged p62 level) (Fig. 8h, i). In agreement with defective degradation of dysfunctional mitochondria, we noticed unchanged level of the mitophagy priming protein PINK1 in mitochondria fraction of vehicle- or with γ-secretase inhibitor-treated 3xTgAD mice (Fig. 7f, h), and enhanced HSP10 protein level in 3xTg-AD mice versus WT mice (Fig. 7f, h), and observed a significant enhancement of TIMM23, HSP10, and MFN2 protein levels in 3xTg-AD versus WT mice treated with γ-secretase inhibitor (Fig. 7f, h). These data fully demonstrate that APP-CTFs accumulation triggers mitophagy failure in vivo.
Mitochondria structure alterations and mitophagy are differently impacted by late-stage Aβ and APP-CTFs accumulation in AD mice models
To delineate the respective contribution of APP-CTFs and Aβ to mitochondrial structure and mitophagy defects, we compared 3xTg-AD and 2xTg-AD mice. Thus, while both AD models accumulate APP-CTFs similarly, only 3xTg-AD animals produce Aβ (at late age) that remains always barely detectable in 2xTg-AD mice [11]. In fact, we confirmed similar APP-CTFs accumulation in mitochondria-enriched fractions from both 2xTg-AD and 3xTg-AD and noticeable absence of Aβ peptide in 2xTgAD mice (Fig. 9a). We also confirmed the presence of large amyloid plaques (AP) surrounded by dystrophic neurites (DN) in the subiculum of 3xTg-AD mice but not in 2xTg-AD mice [Fig. 9b, image 3xTg-AD (”)]. Interestingly, electron microscopy analyses revealed that while old WT mice display equal proportions of classes I and II mitochondria, both old 2xTgAD and 3xTgAD mice display a higher population of class II mitochondria (81% and 82%, respectively) (Fig. 9c and suppl. Table 3, online resource) and significant increases in mitochondria perimeter (Fig. 9d and suppl. Table 3, online resource) and area (Fig. 9e and suppl. Table 3, online resource). These data demonstrate that mitochondria structure alteration in old mice is triggered by the accumulation of APP-CTFs independently of Aβ and pTau (accumulating in old 3xTg-AD but not in 2xTg-AD mice [11]) (Fig. 9b–e).
Mitophagy analyses revealed in old 3xTgAD mice reduced LC3-I level, unchanged LC3-II level, and enhanced LC3-II/LC3-I ratio (Fig. 9f, g). In accordance, with active mitophagy, old 3xTgAD mice showed reduced p62 (Fig. 9f, g), enhanced PINK1 (Fig. 9f, h), and reduced levels of TIMM23 and MFN2 (Fig. 9f, h). We also noticed reduction trends of HSP10 and DRP1 proteins, but that did not reach statistical significance (Fig. 9f, h). Unlike 3xTgAD mice, 2xTgAD mice harbor a mitigate mitophagy phenotype showing reduced LC3-I, unchanged LC3-II levels and LC3-II/LC3-I ratio (Fig. 9f, g), unchanged p62 (Fig. 9f, g), and enhanced PINK1 and a slight reduction of TIMM23 but not of HSP10, MFN2, or DRP1 (Fig. 9f–h). Altogether, these results pinpointed a late-stage contribution of Aβ and pTau to mitophagy induction in old 3xTgAD mice (Fig. 9f–h), while early APP-CTFs accumulation accounts for impaired mitophagy in young 3xTgAD (Fig. 7f–h) and old 2xTgAD mice (Fig. 9f–h). The specific contribution of Aβ versus pTau to mitophagy activation in old 3xTg-AD mice needs a dedicated study.
Correlation between basal mitophagy failure and APP-CTFs accumulation in human AD brains
We lastly questioned whether APP-CTFs accumulate in human SAD brains and investigated the potential correlation between APP-CTFs levels in mitochondria and mitophagy markers. The presence and accumulation levels of APP-CTFs were explored in mitochondria-enriched fraction of a large human SAD brains cohort (patients’ information in suppl. Table 1, online resource). We observed unchanged level of full-length APP (Fig. 10a, b), and enhanced C83 and C99 expressions (Fig. 10a, c–e) in AD brains using the two sets of antibodies APP-Cter and 82E1 (directed to the first and free aa residue of Aβ and C99) (Fig. 10a, d, e). In this cohort, we also revealed a positive and significant correlation between C99 level and total Aβ load (Fig. 10f). Using immunohistochemistry and antigen retrieval approaches [16], we revealed in AD brain slices (patients information in suppl. Table 1, online resource) intracellular APP-CTFs and Aβ membranous-like signal (Fig. 10g, inserts b, d), in parallel to extracellular APP-CTFs’ aggregates and focal amyloid plaques, respectively, detected with APP-Cter and 82E1 antibodies (Fig. 10g, inserts a, c). We investigated in the same brain cohort mitophagy markers (LC3-I, LC3-II, P62, Parkin, and PINK1) expression in mitochondria-enriched fractions (Fig. 10h–n). Interestingly, we showed a significant increase of LC3-II/LC3-I and enhanced p62 level in AD brains as compared to controls (Fig. 10h, i). Supporting mitophagy priming failure, we also revealed a consistent reduction of PINK1 and Parkin proteins (Fig. 10h, j). Importantly, expression of all these markers significantly correlated with mitochondrial C99 level (Fig. 10k–n). Since our AD cohort shows elevated expression levels of Aβ (Fig. 10a, f) and pTau (Suppl. Figure 11e, online resource), we also analyzed the potential correlation of mitophagy markers expression with Aβ and pTau levels. Intriguingly, we noticed a significant negative correlation of total Aβ load with Parkin (Suppl. Figure 11c, online resource), but not with LC3-II/LC3-I, p62 and PINK1 (Suppl. Figure 11a, b, and d, online resource). We also observed a positive correlation of pTau with LC3-II/LC3-I ratio (Suppl. Figure 11f, online resource), but not with the other markers (Suppl. Figure 11 g-i, online resource). These data are in agreement with previous studies, showing that Parkin trigger intracellular Aβ42 clearance [12] and the colocalization of pTau with LC3 in FAD brains [65]. Overall, these data firmly emphasize a strong correlation between APP-CTFs accumulation and defective mitophagy phenotype in human AD brains, thus supporting our findings in various cellular and mice AD models.