Mice
Breeding pairs of B6C3-Tg(APPswe,PSEN1dE9)85Dbo/J (APP-PSEN1; MMRRC stock 34829) [24] and B6;SJL-Tg(rPEPCKSREBF2)788Reh/J (SREBF2; JAX stock 003311) [25] mice were purchased from The Jackson Laboratory. APP-PSEN1 mice express a chimeric mouse/human amyloid precursor protein (isoform 695) with the Swedish mutation (Mo/HuAPP695swe) and mutant human presenilin 1 (PSEN1dE9), both directed to neurons by the mouse PRNP (prion protein) promoter. Both mutations are associated with early-onset Alzheimer disease. In turn, SREBF2 mice express a transgenic construct containing a human SREBF2 cDNA fragment (amino acids 1-468) under the control of the rat PEPCK (phosphoenolpyruvate carboxykinase) promoter. APP-PSEN1-SREBF2 mice were generated from crossbreeding APP-PSEN1 and SREBF2 mice, which were back-crossed more than 10 generations into the B6SJL background and characterized as previously described [26]. Hippocampus from mice that overexpress SREBF2 show elevated mRNA expression levels of the LDL receptor and the cholesterol biosynthetic enzyme HMG-CoA synthase (Supplementary Fig. 1) and accordingly, they display a 1.5-fold increase in total brain cholesterol levels and a 3-fold increase in mitochondrial cholesterol content compared to WT mice values [26]. No significant changes in cholesterol values are observed when comparing mice of different ages (4-10 months of age) [26]. At the time of weaning (21 d), mice were genetically identified by PCR using DNA from ear-tips and following the genotyping protocols provided by the supplier. Because APP-PSEN1-SREBF2 mice show sex-related differences in brain cholesterol levels, we only used male mice. In some cases, mitophagy was induced in vivo by intraperitoneal (i.p.) injection of rapamycin (5 mg/kg; Santa Cruz Biotech., sc-3504). Rapamycin was reconstituted in DMSO at 25 mg/ml and diluted in phosphate-buffered saline (PBS; Sigma-Aldrich, P3813) containing 5% Tween-80 (Sigma-Aldrich, P4780) and 5% PEG-400 (polyethylene glycol 400; Sigma-Aldrich, 202398). In mice that overexpress SREBF2, the recovery of mitochondrial GSH content was accomplished by treatment with GSH ethyl ester (GSHee; 1.25 mmol/kg/day; Sigma-Aldrich, G1404), i.p. injected every 12 h for 2 weeks, as described [27].
Human brain samples
Brain samples were supplied by the Biobank of Hospital Clínic, Barcelona - IDIBAPS (Barcelona, Spain). Clinical diagnosis and neuropathological changes were informed by the Biobank. The histological samples (5-μm sections) were obtained from the hippocampus of post-mortem brain of non-demented controls and AD patients classified according to the “ABC score” proposed by Montine et al. [28], which incorporates histopathologic assessments of amyloid β deposits (A, Thal Phase for Aβ plaques), staging of neurofibrillary tangles (B, Braak and Braak NFT stage), and scoring of neuritic plaques (C, CERAD neuritic plaque score). The age, gender, post-mortem interval and ABC score of the subjects are described in Table 1.
Mitochondria isolation
Mitochondria were isolated from mouse brains as previously described in Yu et al. [29]. The method provides highly pure and structurally intact mitochondria, free from the endoplasmic reticulum, plasma membrane, and endosomal membrane contamination [27, 29]. In brief, brain was homogenized in mannitol buffer (210 mM Mannitol, 60 mM Sucrose, 10 mM KCl, 10 mM Succinate, 0.1 mM EGTA, 1 mM ADP, 0.25 mM DTT, and 10 mM HEPES, pH 7.4). For ubiquitin and phosphorylation analysis, EGTA was replaced by the cysteine protease inhibitor N-ethylmaleimide (10 mM) in the mannitol buffer. Homogenates were centrifuged at 700g for 10 min, and the supernatants were centrifuged at 10,000 g for 15 min. The resulting pellet (the mitochondria-rich fraction) was suspended in 1 ml, loaded onto 8 ml of 30% (v/v) percoll gradient and centrifuged at 95,000 g for 30 min. The mitochondrial pellet was then rinsed twice by centrifuging 15 min at 10,000 g.
In SH-SY5Y cells, a mitochondria-rich fraction was obtained by digitonin permeabilization as described in Díaz et al. [30]. Briefly, we trypsinized and suspended 2.4x106 cells in cold PBS supplemented with proteases inhibitors. We added 2 mg/ml digitonin previously boiled and incubated the samples on ice for 10 min. After the incubation period, the digitonin was immediately diluted with PBS and samples were centrifuged at 21,130 g for 5 min at 4ºC. The pellet was washed with PBS, centrifuged again, and proteins were extracted in RIPA buffer (Santa Cruz Biotech., sc-24948).
Cell cultures and treatments
Embryonic cortical-hippocampal neurons from mice were isolated at day 16-17 of pregnancy following a standard protocol. Dissociated cells were grown in NeurobasalTM medium (Thermo Fisher Sci., 21103-049) supplemented with 2.5% (v:v) B27 supplement (Thermo Fisher Sci., 17504-001), 0.5 mM L-glutamine (Sigma-Aldrich, G7513) and 5 µg/ml plasmocinTM (InvivoGen, ant-mpt), and plated onto poly-D-lysine (Sigma-Aldrich, P6407) and laminin (Sigma-Aldrich, L2020)-coated plates at a density of 2 x 105 cells/cm2. Half of the culture medium was changed every 3 or 4 days. More than a 95% of neuronal purity was confirmed by immunochemistry using antibodies targeted to neuron and glial markers. Experiments were performed at 7 to 10 days in vitro (DIV). Cytotoxicity of the different compounds was assessed by lactate dehydrogenase (LDH) assay (Invitrogen™ CyQUANT™ LDH Cytotoxicity Assay, Thermo Fisher Sci., 16280972).
Human neuroblastoma-derived SH-SY5Y cells (ECACC Cat# 94030304, RRID:CVCL_0019) were grown in GibcoTM DMEM/F12, with GlutaMAXTM (Thermo Fisher Sci., 11559726) supplemented with 10% FBS (Thermo Fisher Sci., 11550356) and 5 µg/ml plasmocinTM (InvivoGen, ant-mpt).
Cholesterol enrichment of SH-SY5Y cells was achieved by using a cholesterol:methyl-β-cyclodextrin complex (CHO:MCD; Sigma-Aldrich, C4951). Cells were incubated with the complex (50 μg/ml of cholesterol) for 1 h, then washed with PBS, and incubated with new medium for 4 h. Cholesterol content was assessed by labeling the cells with 0.25 mg/ml filipin III (Sigma-Aldrich, F4767) for 20 min.
Preparation of Aβ peptides
Human Aβ42 hydrochloride salt (Bachem, H-6466) was dissolved to 1 mM in hexafluoroisopropanol (HFIP; Sigma-Aldrich, 10,522-8) and aliquoted into microcentrifuge tubes, then the HFIP was evaporated and the peptides were stored at -20ºC until use. For oligomeric assembly, concentrated peptides were resuspended by sonication in DMSO at 5 mM concentration and then diluted to 100 µM in phenol red-free medium and incubated at 4ºC for 24 h. Oligomeric forms of Aβ were confirmed by western blot, as previously described [26].
Mt-mKeima mitophagy analysis
The coding sequence of the fluorescent mitochondrial-targeted monomeric Keima protein (mt-mKeima) [31] was cloned into the pWPI lentiviral vector (RRID:Addgene_12254; the plasmid was a gift from Didier Trono). The vector was packaged in lentiviral particles using psPAX2 (RRID:Addgene_12260; the plasmid was a gift from Didier Trono) and pMD2.G (RRID:Addgene_12259; the plasmid was a gift from Didier Trono). SH-SY5Y cells were transduced with a titration of the lentivirus containing the vector. The number of proviruses integrated per genome was assessed by determining the number of WPRE copies per diploid genome by digital polymerase chain reaction (dPCR). The number of diploid genomes was measured by multiplex amplification of two single copy nuclear genes: human TATA-box binding protein 1 (TBP1) and human mitochondrial transcription elongation factor (TEFM) as previously described [32] (primers sequences are detailed in Table 3). SH-SY5Y cells with 3 integrations/diploid genome were selected and collected by FACS (fluorescence-activated cell sorting). The cell sorter was programmed to collect only cell populations displaying low fluorescence intensity at Ex 488 nm and no signal at Ex 561 nm. Selected cells were grown and exposed to the indicated treatments. Dual-excitation ratio imaging of cells was carried out with two sequential excitation lasers (458 and 561 nm) in a Zeiss Lsm780 inverter laser scanning confocal microscope equipped with an argon laser, and using a 40x/1.2-0.28 water C-APO objective and a confocal pinhole set at 1 Airy unit. Images were taken on xyz axis with 6 slices per image. To obtain the mitophagy index, we followed the procedure described in Sun N. et al.[33] Briefly, we selected all pixels with a ratio of red/green pixels higher than 1.5 in a scatter plot (using the FIJI Image J software [34]) and the mitophagy index was calculated by dividing this number of pixels by the total of pixels after subtracting the background.
Recombinant GST-PFO probe
The perfringolysin O (PFO) produced by Clostridium perfringens is a bacterial toxin that specifically binds cholesterol and forms pores. To evaluate the cholesterol content in human hippocampal tissues, we used a recombinant PFO fusioned with Glutathione-S-Transferase (GST-PFO), as described [35]. A pGEX4T plasmid vector containing the coding DNA sequence for PFO-GST, kindly provided by Dr. Kwiatkowska [35], was transformed into BL21(DE3) strain of E. coli (Thermo Fisher Sci., EC0114). Bacteria were grown at 37°C in LB medium containing 100 μg/ml ampicillin to OD = 0.6, when 0.5 mM IPTG was added. After 20 h at 18ºC, bacteria were centrifuged at 4,000 g for 10 min at 4ºC, washed with PBS and incubated 10 min at 4ºC in lysis buffer (50 mM Tris pH 8, 100 mM NaCl, 1 mM EGTA, 1 mM EDTA, 5 mM β-mercaptoethanol, 1 mM PMSF, 0.35 mg/ml lysozyme and proteases inhibitor cocktail). The suspension was sonicated on ice for 15 min at 0.3 cycles and amplitude 33 % after adding 1% Triton X-100. The lysate was clarified at 18,000 g for 30 min at 4ºC and loaded onto a GSTrapTM 4B column (GE Healthcare, 28-4017-45). Purification was made according to the commercial protocol, adding 5 mM DTT and 1% Triton X-100 to the wash buffer. GST-PFO was eluted with 10 mM GST, 5 mM DTT, 50 mM Tris pH8 and the elute was filtered using Amicon Ultra-15 10k centrifugal filters (Merk, UFC901008). The protein was aliquoted and stored at -80ºC with 20% sucrose. Purity of recombinant GST-PFO was assessed by western blot in comparison with the unbound proteins of the lysate.
Protein-lipid overlay assay
To analyze the ability of the recombinant GST-PFO to recognize cholesterol, we performed a protein-lipid overlay assay, as described [35]. 1 µl of the CHO:MCD complex solution containing 100–800 pmols of cholesterol was spotted onto a 0.2 μm nitrocellulose membrane (GE Healthcare, 1060015). Once completely dry, the membrane was pressed with a hot block at 60ºC for 5 s to fix the lipid. Then, it was blocked with 5 % nonfat milk for 1h and incubated with 2 μg/ml recombinant GST-PFO in TBS with 0.1 % Tween-20 (TBS-T) for 1h at RT. Protein-lipid blots were probed with anti-GST (1:1.000; Santa Cruz Biotech., sc-374171) overnight at 4ºC, washed three times and probed with mouse HRP-linked IgG antibody (1:20,000; Sigma-Aldrich, GENA931) in TBS-T with 1% nonfat milk for 1h at RT. Bound antibodies were revealed using Clarity Max ECL western blotting substrate (Bio-Rad, 1705062).
Western blot analysis
Lysates were prepared in RIPA buffer (Santa Cruz Biotech., sc-24948). Samples (20 to 70 μg of protein/lane) were resolved by SDS-PAGE and transferred to nitrocellulose membranes. Protein transfer was checked by Ponceau S solution (Sigma-Aldrich, P7170). Blots were probed with the primary antibodies listed in Table 2 overnight at 4ºC. The membranes were then incubated with anti-rabbit (1:10,000), anti-mouse (1:20,000) or anti-rat (1:40,000) IgG horseradish peroxidase-coupled secondary antibodies (Sigma-Aldrich, A8275, GENA931, GENA935, respectively) and the immunoreactive bands were detected using Clarity ECL western blotting substrate (Bio-Rad, 1705061) or Clarity Max ECL western blotting substrate (Bio-Rad, 1705062) for low protein concentrations. To analyze phosphorylated proteins, isolated mitochondria were subjected to Phos-tag SDS-PAGE. Resolving gels were made with 7% acrylamide, 100 nM MnCl2 and 50 µM Phos-tagTM AAL-107 (Wako Chemicals, 304-93521), according to manufacturer instructions. After electrophoresis, Phos-tag gels were soaked 2 times in transfer buffer (25 mM Tris-HCl pH 8.3, 192 mM glycine, 0.05% SDS, and 20 % methanol) containing 1mM EDTA for 10 min, with gentle agitation for the elimination of the manganese ions from the gel. The resolved proteins were transferred to PVDF membranes previously activated 30 s in 100% methanol and equilibrated 15 min in transfer buffer. Blots were probed with the primary antibodies detailed in Table 2. Uncropped scans of representative western blots from Fig. 1, 3, 4 and 5 are shown in Supplementary Fig. 13 to 16.
Immunofluorescence and laser confocal imaging
Paraffin-embedded blocks from mouse brains were prepared by sequential dehydration in graded ethanol and infiltration in paraffin before embedding. Blocks were serially sectioned between –1.2 mm and –2.4 mm from Bregma at a thickness of 5 µm. Dewaxed and rehydrated hippocampal sections from mouse and human brains were first boiled in Tris-EDTA buffer, pH 9.3. Then, sections were blocked in Antibody Diluent with Background Reducing Components (Dako, S3022) for 20 min at RT and incubated overnight at 4°C with the antibodies listed in Table 2. After washing with PBS, the immunoreaction was visualized using the following secondary antibodies: anti-goat Cy3 (1:300, Jackson ImmunoResearch Labs Cat# 705-166-147, RRID:AB_2340413), anti-rabbit Cy3 (1:300, Jackson ImmunoResearch Labs Cat# 711-166-152, RRID:AB_2313568), anti-rabbit Alexa Fluor 488 (1:300, Molecular Probes Cat# A-21206, RRID:AB_2535792), anti-mouse Alexa Fluor 488 (1:300, Molecular Probes Cat# A-21202, RRID:AB_141607) or anti-mouse Alexa Fluor 555 (1:300, Thermo Fisher Scientific Cat# A-31570, RRID:AB_2536180). Nuclei were stained with DRAQ5 fluorescent probe solution (5 μM; Biostatus, DR50200) or bisBenzimide Hoechst 33258 (1μg/ml; Sigma Aldrich; B2883). Hippocampal-cortical neurons and SH-SY5Y cells were fixed with 4% paraformaldehyde and permeabilized with 0.1% saponin (Sigma-Aldrich, 47036) in blocking buffer (1% BSA and 0.5% glycine in PBS) for 15 min before adding the primary antibodies. In some cases, samples were incubated with 20 μg/ml GST-PFO in antibody diluent (Antibody Diluent with Background Reducing Components) for 3 h (tissue sections) or 45 min (cells), and washed three times with PBS before proceeding with the immunostaining. Fluorescence microphotographs were acquired with a Nikon Eclipse E-100 microscope using the objective 40x/0.75 plan fluor DLL. Confocal images were collected with a Leica TCS SPE laser scanning confocal microscope using the objectives 20x/0.7 HC PL APO and 63x/1.4-0.60 oil HCX PL APO, and a confocal pinhole set at 1 Airy unit. For whole hippocampus imaging, individual images were acquired and stitched using a 20x/0.17 MI-oil plan fluor objective in an Andor Dragonfly spinning disk confocal microscope equipped with a Zyla 4.2 PLUS sCMOS camera and coupled with Fusion software (Andor).
Mitochondrial DNA (mtDNA) quantification
The prefrontal cortex, hippocampus and cerebellum from mouse brains were added directly into 800 µl of the solubilization reagent DireCtQuant 100ST (Frontex Biomed, #DCQ100ST) and homogenized in tubes with glass beads at 350 rpm for 45 s. After homogenization, each sample was incubated at 90 °C for 3 min, cooled to RT and centrifuged at 10,000 g for 10 min. The supernatant was diluted 1:3 in DireCtQuant 100ST and stored at -20 °C. Prior quantification, the samples were diluted to 1:16 and 1:16,000 for genomic and mitochondrial measurements, respectively. Copies of mtDNA were quantified by dPCR. Before dPCR, we set up a restriction enzyme digestion as described in Podlesniy and Trullas [36]. Briefly, we added 8.8 μl of the diluted sample into a new tube with 0.5 μl of both SaqA1 and BsuRI (for genomic DNA) or 1 μl FastDigest SspI (for mtDNA) restriction enzymes, 10 μl of 2X QX200 ddPCR EvaGreen Supermix (Bio-Rad, 186-4033) and the corresponding primer mix (Table 3) to a final PCR volume of 20 µl. To quantify the number of genome copies we used two sets of primers targeting two different single-copy nuclear genes that produce amplicons of different sizes (Bax-72 and Gsk3β-86). The reaction mixture was incubated for 15 min at 37 ºC. After restriction digestion, the mixture was partitioned and emulsified in 65 μl of droplet generation oil for EvaGreen (Bio-Rad, 186-4005) in a QX200 Droplet Generator. The emulsion was transferred to a 96-wells plate and dPCR was performed in a C1000 Touch Thermal cycler (Bio-Rad) with the following cycling conditions: 95 ºC for 5 min, 95 ºC for 30 s and 57 ºC (for mtDNA) or 60 ºC (for genomic DNA) for 1 min for 40 cycles, 4 ºC for 5 min, 90 ºC for 5 min and 11 ºC for storage before dPCR analysis. The quantification of positive and negative droplets was assessed using a QX200 Droplet Reader. Results were analyzed with QuantaSoftTM Analysis Pro software. Based on the ratio of positive to total droplets, the software estimates the concentration of target molecules per reaction, which was used to calculate the absolute number of copies of the target molecules per volume of sample analyzed. To calculate the numbers of gene copies per diploid genome, the number of copies of mtDNA was divided by the number of copies of the single-copy genes Bax and Gsk3β present in the sample.
Selfie RT-qPCR
We determined the mRNA expression levels of Tfam, Ldlr and Hmgcs1 by Selfie RT-qPCR, a method that determines the absolute amount of an RNA transcript produced by its coding DNA by measuring, in the same sample containing unpurified RNA and DNA, the amount of DNA amplified by qPCR before and after reverse transcription [36]. Briefly, the hippocampus isolated from mice brains was homogenized, and the resulting samples were diluted as described in the above section. We performed a pre-annealing step in duplicate, mixing the lysed sample with 0.5 µM reverse primer (complementary to Tfam RNA transcript) at 70 °C for 5 min. Next, we added RiboLock RNase Inhibitor (ThermoFisher Sci., EO0381), 10 mM dNTPs and glycerol or Maxima H Minus Reverse Transcriptase (ThermoFisher Sci., EP0751), and the mixture was retro-transcribed for 30 min at 60 °C and 5 min at 85 °C. Finally, 0.5 μM forward primer was added and amplified by conventional RT-qPCR (5 min at 95 °C, followed by 40 cycles of 15 s at 95 °C, 35 s at 60 °C and 25 s at 72 °C), using SsoAdvanced™ Universal SYBR® Green Supermix (Bio-Rad, 1725272). The number of transcripts per encoding gene present in the sample was calculated by subtracting the amount obtained before reverse transcription from that obtained after reverse transcription yields and divided by the amount obtained before reverse transcription.
Endo-lysosome and autophagosome fractionation
Autophagosomes and endo-lysosomes were isolated by centrifugation in a discontinuous nycodenz gradient, as described previously [37]. The corresponding membrane fraction resulting from 5 brains was suspended in 1ml of 50% nycodenz and loaded onto 3.5 ml of 26% nycodenz placed at the bottom of the tubes. Density gradient separation of autophagic vacuoles was achieved after layering above the sample with 1.5 ml of each 24%, 20% and 10% nycondenz, and ultracentrifugation at 26,000 rpm for 4 h in a SW40Ti rotor (Beckman Coulter). The bands were collected from the gradient, diluted five times with 0.3 M sucrose and sedimented at 37,000 g for 10 min. Autophagosomes were recovered in the 20-15% interface and endo-lysosomes in the 26-24% interface. Purity and efficiency was evaluated as described previously [18]. Autophagosomal and lysosomal fractions were both individually labeled with anti-LC3 antibody plus FITC-labeled secondary antibody and LysoTracker Red. The number of particles for each fluorophore was quantified and divided by the total count from its corresponding phase contrast images. We only used fractions with a percentage of positive stained particles per microscopic field over 70%.
Statistics
Results are expressed as means ± standard deviation (SD) of the number of experiments. Statistical comparisons were performed using unpaired 2-tailed Student’s t-test or one-way analysis of variance followed by Tukey’s test for multiple comparisons. Pearson's correlation coefficient (PCC) was determined as a statistic for quantifying colocalization using the FIJI ImageJ plugin Coloc 2 with a ROI selection of interest in some cases [34]. A P < 0.05 value was considered statistically significant.