Mitochondrial hypermetabolism precedes impaired autophagy and synaptic disorganization in App knock-in Alzheimer mouse models

Abstract

Accumulation of amyloid β-peptide (Aβ) is a driver of Alzheimer's disease (AD). Amyloid precursor protein (App) knock-in mouse models recapitulate AD-associated Aβ pathology, allowing elucidation of downstream effects of Aβ accumulation and their temporal appearance upon disease progression. Here we have investigated the sequential onset of AD-like pathologies in App^NL-F and App^NL-G-F knock-in mice by time-course transcriptome analysis of hippocampus, a region severely affected in AD. Strikingly, energy metabolism emerged as one of the most significantly altered pathways already at an early stage of pathology. Functional experiments in isolated mitochondria from hippocampus of both App^NL-F and App^NL-G-F mice confirmed an upregulation of oxidative phosphorylation driven by the activity of mitochondrial complexes I, IV and V, associated with higher susceptibility to oxidative damage and Ca²⁺-overload. Upon increasing pathologies, the brain shifts to a state of hypometabolism with reduced abundancy of mitochondria in presynaptic terminals. These late-stage mice also displayed enlarged presynaptic areas associated with abnormal accumulation of synaptic vesicles and autophagosomes, the latter ultimately leading to local autophagy impairment in the synapses. In summary, we report that Aβ-induced pathways in App knock-in mouse models recapitulate key pathologies observed in AD brain, and our data herein adds a comprehensive understanding of the pathologies including dysregulated metabolism and synapses and their timewise appearance to find new therapeutic approaches for AD.

Fig. 1. Transcriptome profiling identified Aβ-induced alterations in genes and pathways in hippocampus of App knock-in mice.

A Hippocampi were dissected from two-, six- and 12-month-old App^NL-G-F mice and six-, 12- and 18-month-old App^NL-F mice and age matched WT controls (n = 3/genotype). RNA was extracted from dissected hippocampi and cDNA libraries were synthesized. RNA sequencing was performed by a Hiseq 3000 sequencer. Validation studies were performed by RT-qPCR using the same RNA samples used for RNA sequencing. Western blotting with hippocampal brain homogenate and Olink proteomics were conducted for validations at protein level. Mitochondria and crude synaptosomal fraction were isolated for mechanistic studies of mitochondrial and autophagic functions. Mitochondrial dysfunction and autophagic alterations were revealed by electron microscopy. B t-SNE plot representing difference of App^NL-F, App^NL-G-F and WT transcriptomes. Each symbol represents one mouse individual. Each color represents one mouse genotype, red: App^NL-F, blue: App^NL-G-F, gray: WT, and each shape of symbol represents mouse age, circle: two-month-old, triangle: six-month-old, inverted triangle: 12-month-old, square: 18-month-old. C Venn diagram of significantly DEGs in App^NL-F and App^NL-G-F vs WT mice (FDR < 0.1). D Volcano plots of the gene expression profiles in different time points of App^NL-F or App^NL-G-F vs WT mice. Red and blue dots indicate significantly up- and downregulated genes (FDR < 0.1), respectively, in App knock-in mice. Grey dots indicate non-significantly altered genes. Genes validated by RT-qPCR are highlighted. E Heatmap of selected pathways related to AD, glucose metabolism, neuroinflammation and autophagy in App^NL-F vs WT mice (columns 1 to 3), App^NL-G-F vs WT mice (columns 4 to 6) and App^NL-F vs App^NL-G-F mice (columns 7 and 8). The p-value of each enriched pathway was converted to a Z score. Significantly up- and downregulated pathways have the absolute value of Z scores ≥ 1.96.

Fig. 2. OxPHOS gene expression and activity were upregulated in early symptomatic App knock-in mice.

A–D OCR of mitochondria in coupled state, isolated from (A, B) six-month-old WT and App^NL-F mice and (C, D) two-month-old WT and App^NL-G-F mice was measured using the Seahorse apparatus, and calculations of state II, state III induced by ADP (4 mM), state IIIu induced by FCCP (4 μM), and state IVo induced by oligomycin (3.2 μM, Oligo) were performed (n = 5–6; effect size for App^NL-F: state II = 26.58 ± 8.22, state III = 38.49 ± 8.22, state IIIu = 32.79 ± 8.22, state IVo = 11.28 ± 8.22; effect size for App^NL-G-F: state II = 49.44 ± 18.87, state III = 58.91 ± 18.87, state IIIu = 48.17 ± 18.87, state IVo = 39.1 ± 18.87). E Chord plot of significantly altered genes (FDR < 0.1) included in GO terms related to mitochondrial function. The color of the circle edge boxes indicate up- (red) or down- (blue) regulation. F Organization of OxPHOS genes sorted by mitochondrial complex in the mitochondrial cristae. Significantly upregulated genes in two-month-old App^NL-G-F mice are shown in dark red (FDR < 0.1). Percentage of upregulated OxPHOS genes grouped by mitochondrial complex (Cx I, III–V) in two-month-old App^NL-G-F mice based on their FDR value. G–J Electron flow of uncoupled mitochondria isolated from (G, H) six-month-old WT and App^NL-F mice and (I, J) two-month-old WT and App^NL-G-F mice was evaluated using the Seahorse apparatus. Mitochondrial complex inhibitors and substrates, 2 μM rotenone (Rot), 10 mM succinate (Succ), 4 μM antimycin A (Ant A), and 1 mM ascorbate (Asc)/100 mM TMPD, were sequentially injected to analyze the mitochondrial complex I-IV activities (n = 5–8; effect size for App^NL-F: Cx I = 17.83 ± 1.19, Cx II = 27.36 ± 1.19, Cx III = 21.58 ± 1.19, Cx IV = 37.64 ± 1.19; effect size for App^NL-G-F: Cx I = 24.93 ± 18.87, Cx IV = 52.3 ± 18.87). K Levels of H₂O₂ in isolated mitochondria from six-month-old WT and App^NL-F mice were fluorometrically quantified using amplex red reagent (n = 6; effect size = 0.358 ± 0.231). L–O Calcium uptake of hippocampal mitochondria was evaluated with the fluorescent probe Calcium-green. Five pulses of 10 μM CaCl₂ were added to evaluate the mito-Ca²⁺ retention capacity (n = 4–5; effect size for App^NL-F: 0.278 ± 0.1091; effect size for App^NL-G-F: 0.258 ± 0.1253). Statistical significance was analyzed using non-parametric Mann–Whitney test. *p < 0.05, **p < 0.01. OxPHOS oxidative phosphorylation, Cx mitochondrial complex, OCR oxygen consumption rate, IMM inner mitochondrial membrane, IMS inner mitochondrial space, OMM outer mitochondrial membrane.

Fig. 3. Decay in mitochondrial complexes activities and Ca²⁺ handling capacity characterize symptomatic App knock-in mice.

A–D OCR of mitochondria in coupled state, isolated from (A, B) 15-month-old WT and App^NL-F mice and (C, D) 12-month-old WT and App^NL-G-F mice was measured using the Seahorse apparatus (n = 5–6). E–H Electron flow of uncoupled mitochondria isolated from (E, F) 15-month-old WT and App^NL-F mice and (G, H) 12-month-old WT and App^NL-G-F mice was evaluated using the Seahorse apparatus (n = 5–6; effect size for App^NL-F: Cx I = 12.53 ± 12.01, Cx II = 25.86 ± 12.01, Cx III = 24 ± 12.01, Cx IV = 44.9 ± 12.01; effect size for App^NL-G-F: Cx I = 22.07 ± 16.71, Cx II = 40.49 ± 16.71, Cx III = 37.89 ± 16.71). I Levels of H₂O₂ in isolated mitochondria from 15-month-old WT and App^NL-F mice were fluorometrically quantified using amplex red reagent (n = 6). J–L Calcium uptake of hippocampal mitochondria was evaluated with the fluorescent probe Calcium-green. Five pulses of 10 μM CaCl₂ were added to evaluate the mito-Ca²⁺ retention capacity (n = 5–6; effect size: in App^NL--F = 0.224 ± 0.099; in App^NL-G-F = 0.312 ± 0.1195). M Significantly downregulated GO biological process terms associated with synaptic and transport functions in App^NL-G-F mice. Statistical significance was analyzed using non-parametric Mann–Whitney test. *p < 0.05, **p < 0.01.

Fig. 4. Aberrant synaptic morphology is associated with loss of mitochondria, increase in synaptic vesicles and autophagosome accumulation.

A, F Electron microscopy images of hippocampal CA1 from (A) 22-to 24-month-old WT and App^NL-F mice (n = 3 including one WT male, with an average of 50 cells and 70 synapses analyzed per genotype; effect size in C = 62.77 ± 19.05; in D = 7.40 ± 2.84; in E = 5.99 ± 1.77), and (B) 10- to 12-month-old WT and App^NL-G-F mice (n = 4, with an average of 50 cells and 70 synapses analyzed per genotype; effect size in G = 0.16 ± 0.06, H = 0.74 ± 0.157, I = 88.58 ± 43.02; in J = 26.82 ± 4.29; in K = 15.21 ± 2.25). Black arrowhead: AV, blue arrowhead: AM/AL, red arrowhead: Mitochondria, black arrow: postsynaptic density, and red arrow: MVB. Scale bar: 2 µm. Quantification of mitochondrial profiles in the synapses (B, G), ER aspect ratio (H), synaptic area (C, I), number of synaptic vesicles (D, J) and post-synaptic thickness (E, K) in WT and App knock-in mice. Statistical significance was analyzed using non-parametric Mann–Whitney test. *p < 0.05, **p < 0.01, ****p < 0.0001. AV autophagic vacuole, AM/AL amphisome/autolysosome, Mit Mitochondria, MVBs multivesicular bodies, PSD postsynaptic density, SVs synaptic vesicles, Syn synapse.

Fig. 5. Impaired synaptosomal autophagy in aged App^NL-G-F mice is observed especially around Aβ plaques.

A Chord plot of significantly (FDR < 0.1) DEGs related to autophagy. The color of the circle edge boxes indicate up- (red) or down- (blue) regulation. B The relative mRNA expression of selected genes was normalized to Tubb3 (n = 3; size effect for Trim30a 6-month-old App^NL-G-F = 4.004 ± 1.676, for Rubcnl 6-month-old App^NL-G-F = 0.9521 ± 0.2321, for Lamp2 6-month-old App^NL-G-F = 0.4017 ± 0.1530, for Rab7b 6-month-old App^NL-G-F = 2.354 ± 0.6304, 12-month-old App^NL-G-F = 4.910 ± 0.2842, 12-month-old vs 2-month-old App^NL-G-F = 4.678 ± 0.2835). Statistical significance was analyzed using Kruskal-Wallis tests followed by Dunn’s multiple comparison test. *p < 0.05. * vs age matched WT mice, ^# vs 2-month-old genotype-matched mice. C Co-immunofluorescence staining of Aβ, LC3 and synaptophysin with 12 month-old WT and App^NL-G-F mice (n = 4). Red: Aβ, Green: LC3, Yellow: Synaptophysin, Blue: nucleus. Scale bar: 20 µm. D, E Signal intensity of Aβ (Red), LC3 (Green), Synaptophysin (Yellow) and nucleus (Blue) under the white line of WT and App^NL-G-F mouse brain staining gated in (C). F–I Phospho-p62 (S403), total p62, LC3-I and LC3-II protein levels in hippocampal crude synaptosomal fraction (P2) or soluble fraction (S2) from 12-month-old App^NL-G-F hippocampus were visualized by Western blotting (n = 4; size effect for p-p62 in P2 = 0.8713 ± 0.2675 and p-p62 in S2 = 1.387 ± 0.4519, for p62 in P2 = 0.3021 ± 0.08530, for LC3-II in P2 = 1.350 ± 0.2286). Protein levels were normalized to β3-tubulin. Statistical significance was analyzed using unpaired t-test. *p < 0.05, **p < 0.01.