Antibodies and reagents
Anti-ATPB (ATP synthase subunit β) pAb/mAb (ab128743/ab5432) was purchased from Abcam (Cambridge, MA, USA), anti-ATG5 pAb (NB110-53818) from Novus Biologicals (USA), anti-Gp78 pAb (16675-1-AP) from Proteintech (USA) and anti-LC3B pAb (2775S) from Cell Signaling. Anti-β-Actin mAb (A5441), tissue culture grade DMSO (D2650), Bafilomycin A1 (Cat# C1988) and CCCP (C2759) were purchased from Sigma-Aldrich (USA). mRFP‑GFP tandem fluorescent‑tagged LC3 (tfLC3) was a gift from Tamotsu Yoshimori (Addgene plasmid # 21074; www.addgene.org/21074) . mKeima-Red-Mito-7 plasmid was a gift from Michael Davidson (Addgene plasmid #56018; www.addgene.org/56018). MitoTracker Deep Red FM, Live Cell Imaging Solution (A14291DJ), Glucose Solution (A2494001), MitoSOX Red (M36008) and MEM Non-Essential Amino Acids Solution (11140050) were purchased from ThermoFisher (USA). MitoView 633 (#70055) was purchased from Biotium (USA).
Cell lines and CRISPR/Cas9 knockout of Gp78
The HT-1080 fibrosarcoma cell line was acquired from ATCC, authenticated by Short Tandem Repeat (STR) profiling at the TCAG Genetic Analysis Facility (Hospital for Sick Kids, Toronto, ON, Canada www.tcag.ca/facilities/geneticAnalysis.html), tested regularly for mycoplasma infection by PCR (ABM, Richmond, BC, Canada) and maintained in RPMI 1604 media supplemented with 10% FBS and 1% L-Glutamine in a 37°C incubator with 5% CO2.
GeneArt-CRISPR/Cas9 Nuclease vector with OFP (Orange Fluorescence Protein) kit (A21174) was from Life Technologies (Invitrogen, USA). We used http://crispr.mit.edu to design guided RNAs and http://www.rgenome.net/cas-offinder/ (RGEN tools) to check them for off-target effects and used the following oligonucleotides for guide RNA1 (5’-CAC CGG AGG AAG AGC AGC GGC ATG G-3’, 5’-AAA CCC ATG CCG CTG CTC TTC CTC C-3’) and guide RNA2 (5’-CAC CGG CCC AGC CTC CGC ACC TAC A-3’, 5’-AAA CTG TAG GTG CGG AGG CTG GGC C-3’). From isolated genomic DNA for each clone, the DNA fragment flanking Exon1 of Gp78 was PCR amplified, cloned and sequenced. For gRNA1 clones #3 and #4 the G was deleted from the ATG start codon while for clone #7 an additional T was inserted in the start codon generating ATTG. For all three gRNA2 clones (#13, 36, 41), a T was inserted at amino acid 16 (CCTA to CCTTA) causing a frameshift mutation. The guide RNAs were in vitro annealed, cloned into the GeneArt linear vector according to the supplier’s protocol and sequence verified prior to transfection into HT-1080 cells. Sequence verified gRNA1 or gRNA2 containing GeneArt-CRISPR/Cas9 Nuclease vector with OFP were transiently transfected into HT-1080 cells, plated 24 hours previously, using Effectene transfection reagent (301425, Qiagen, USA). After 36 hours incubation, cells were harvested and genomic DNA isolated to perform GeneArt Genomic Cleavage Detection assay (A24372, Invitrogen, USA) to check cleavage efficiency. Once cleavage efficiency was confirmed, HT-1080 cells were replated for 36 hours, trypsinized, FACS sorted and OFP expressing cells were singly plated in 96 well pates by serial dilution. Single colonies were replicated in 12 well plates; one set was frozen and stored in liqN2 and the other set subjected to lysate preparation, SDS-PAGE and Gp78 western blot analysis. Arbitrarily chosen representative clones (g1-3, g1-4, g1-7; g2-13, g2-36, g2-41) from both gRNAs were expanded, tested for mycoplasma and stored as multiple freeze-downs. From isolated genomic DNA, an approximate 800bp fragment flanking Exon1 of Gp78 was PCR amplified using Q5 (Qiagen, USA) the following primer set (Forward: 5’-CTG GAG GCT ACT AGC AAA-3’, Reverse: 5’-ATG TGG CCC AGT ACC T-3’) and TA cloned. At least ten clones were sequenced from each to confirm INDEL.
HT-1080 and Gp78 CRISPR/Cas9 knockout clones were grown only up to six passages. Cells were passed every 48 hours at a density of 300,000 cells per 10 cm petri dish, rinsed every 24 hours with 10 ml of PBS and supplied with 10 ml of fresh complete medium. HT-1080 cells and the g2-41 Gp78 CRISPR/Cas9 knockout clones were stably transfected with mRFP‑GFP tandem fluorescent‑tagged LC3 (tfLC3) plasmid using Effectene (Cat. #301425, Qiagen, USA) following the manufacturer’s protocol. After 24 hours incubation, transfected cells were selected against G418 (400ug/ml) for about 14 days. The resistant cell population was pooled and maintained in 50 µg/ml G418.
siRNA knockdown, plasmid transfection and western blotting
siControl and siATG5 (Cat# D-001810-01-05, Cat# L-004374-00-0005) were purchased from Dharmacon and transiently transfected wherever indicated to wild-type HT-1080 cells or g1-4 or g2-41 Gp78 CRISPR/Cas9 knockout clones using Lipofectamine 2000 (Cat# 11668019, Invitrogen, USA) following the manufacturer’s protocol. All siRNA transfection experiments were for 48 hours and treatments were performed 24 hours post siRNA transfection. Alternatively, cells were transiently transfected with mammalian protein expressing plasmids using Effectene (Qiagen, Germany) following the manufacturer’s protocol. Where indicated, cells were treated with 10 µM of CCCP or a corresponding volume of DMSO as control 24 hours prior to fixation or harvesting cells. Western blotting was performed as previously described using Horseradish Peroxidase (HRP)-conjugated secondary antibody followed by addition of ECL (GE Healthcare Bio-Sciences Corp., USA) to reveal chemiluminescence . Densitometry quantification was done using ImageJ (https://imagej.nih.gov/ij/docs/faqs.html#cite) software. Full, uncropped Western blots can be viewed in Supplemental Figs. 4, 5 and 6.
Fluorescent labeling of mitochondria
For immunofluorescent labeling, cells were: 1) fixed with 3.0% PFA for 15 minutes at room temperature and washed with PBS-CM (phosphate buffer solution supplemented with 1 mM CaCl2 and 10 mM MgCl2); 2) permeabilized with 0.2% Triton X-100 for 5 minutes and washed with PBS-CM; 3) blocked with 1% BSA for 1 hour at room temperature; 4) labeled with anti-ATPB for one hour followed by washing with PBS-CM; 5) incubated with secondary antibodies for 1 hour followed by washing with PBS-CM; and 6) mounted in ProLong Diamond (ThermoFisher) and cured for 24 hours at room temperature before imaging. Confocal image stacks were obtained on a III-Zeiss spinning disk confocal microscope with either Zeiss Plan-Apochromat 63X/1.2NA or 100X/1.4NA oil objectives using SlideBook 6.0 image acquisition and analysis software (Intelligent Imaging Innovation Inc). Anti-ATPB label was thresholded from 3D images to measure mitochondrial volume with SlideBook 6.0 image analysis software.
To assess the impact of Gp78-dependent basal mitophagy on mitochondrial health and mitochondrial ROS, wildtype HT-1080 and the g2-41 Gp78 CRISPR/Cas9 knockout clone were plated into an Ibidi chamber for 24 hours. Cells were then transiently transfected with siRNA targeting ATG5 for 48 hours and labelled with either mitochondrial health sensor dye MitoView 633 or mitochondrial ROS dye MitoSOX, at concentrations of 50 nM and 2.5 µM, respectively, for 30 minutes, washed 3X with warm PBS and incubated in Molecular Probes Live Cell Imaging Solution. Alternatively, cells were treated with CCCP for 4 hours and then labeled with MitoView 633 for 30 minutes. Live-cell imaging was performed at 37ºC with a Leica TCS SP8 confocal microscope with a 100×/1.40 Oil HC PL APO CS2 objective (Leica, Wetzlar, Germany) equipped with a white light laser, HyD detectors, environmental chamber and Leica Application Suite X (LAS X) software. Images were analyzed using ImageJ software to identify integrated densities of mitochondrial objects as well as the total area of the mitochondrial label, per cell. Integrated density per mitochondrial object reports specifically on dye intensity within individual mitochondria or mitochondrial segments and is therefore not impacted by varied mitochondrial content per cell.
Mitophagic flux assays
To study mitophagic flux in HT-1080 or Gp78 CRISPR/Cas9 knockout (g1-4, g2-41) cells, early passage cells (420,000 cells per well) were plated in six well plates for 20 hours, then washed with 1X PBS and treated with DMSO or CCCP in regular medium or medium lacking serum for 4 hours. For each treatment, cells were challenged with 100 nM of BafA1 (Sigma) for 0, 30, 60 or 120 minutes prior to the end of the 4-hour incubation period. Incubation was stopped by washing cells with ice-cold 1X PBS; cells were then harvested on ice lysed with M2-Lysis buffer  supplemented with phosphatase and protease inhibitors tablets (Roche), and lysates ran on 15% SDS-PAGE at constant voltage (75V for fifteen minutes followed by 90V for 2 hours). Separated proteins were electrotransferred onto 0.2 µ pore size PVDF membrane (BioRad), fixed with 0.1% glutaraldehyde in PBST (0.2%) for 30 minutes, blocked with 5% milk in PBST and immunoprobed for LC3B-I and II and β-Actin. LC3B-II and β-actin bands were densitometrical quantified using ImageJ software, normalized and statistically analyzed.
To monitor mitophagic flux with mito-Keima, HT-1080 and Gp78 CRISPR/Cas9 knockout g2-41 cells (8,000 cells per well) were plated in an ibidi chamber for 24 hours and then transfected with mito-Keima plasmid  using Effectene transfection reagent (301425, Qiagen, USA). After 24 hours, cells were washed with PBS and treated with DMSO or CCCP in regular medium. Following a 24-hour incubation, the cells were washed 3X with PBS and then incubated in Live Cell Imaging Solution supplemented with 10% FBS, L-glutamine, D-glucose, and MEM Non-Essential Amino Acids Solution prior to imaging on a Leica TCS SP8 confocal microscope equipped with a 100x/1.40 Oil HC PL APO CS2 objective (Leica, Wetzlar, Germany), white light laser and HyD detectors (Leica, Wetzlar, Germany). mito-Keima in a neutral pH environment was detected by excitation at 470nm and in an acidic environment by excitation at 561nm. The HyD detector was open from 592nm to 740nm and equipped with a time gate limiting detection from 0.3ns to 6.5ns after laser activation. To quantify the presence of mitolysosomes, the fluorescent images were loaded into FIJI with the mito-QC Counter macro installed . Settings (Radius for smoothing = 2.5, Ratio threshold = 1.6, Red channel threshold = 3.2) for ratio analysis were determined to report most accurately on mitolysosome expression across all data sets. For some images, large numbers of spots were detected that did not correspond to observed mitolysosomes. For consistency, the two images presenting the largest number of mitolysosomes per group were removed from the analysis.
To monitor autophagic flux with tfLC3, stably transfected HT-1080 and Gp78 CRISPR/Cas9 knockout g2-41 cells were plated overnight and then treated with either DMSO or CCCP for 4 hours and with or without 100 nM BafA1 for the final 2 hours of the incubation period. Mitochondria were labelled with MitoTracker Deep Red FM half an hour prior to the end of the total incubation period. After incubation, cells were gently washed 3X with warm PBS and then incubated in warm Live Cell Imaging Solution just prior to image acquisition. Live-cell imaging was performed using Leica TCS SP8 confocal microscope with a 100×/1.40 Oil HC PL APO CS2 objective (Leica, Wetzlar, Germany) equipped with a white light laser, HyD detectors, and Leica Application Suite X (LAS X) software. Image acquisition was performed in a temperature-controlled system set to 37°C.
For time lapse imaging, HT-1080 cells expressing tfLC3 were plated in ibidi chambers in Molecular Probes Live Cell Imaging Solution supplemented with 10% FBS, L-glutamine, D-glucose, and MEM Non-Essential Amino Acids Solution and labeled with MitoView 633 prior to imaging at 37ºC with the 100X/NA 1.45 PL APO objective (Zeiss) of a 3i Yokogawa X1 spinning disk confocal. Image stacks of 7 images with a 500 nm Z spacing were acquired every minute for 40 minutes with a QuantEM 512SC Photometrics camera. Average intensity of MitoView 633 positive pixels overlapping each GFP-mRFP-positive tfLC3 puncta was assessed relative to average intensity of all MitoView 633-positive pixels in either the adjacent segmented mitochondria or in the cell.
tfLC3 spot detection analysis (SPECHT)
To identify tfLC3 labeled autophagic vacuoles (autophagosomes), we applied the SPECHT object detection method, that is consistent across channels and robust to intensity variations . SPECHT evolved from the ERGO software for density detection in single molecule localization microscopy  and accepts as input a confocal image and produces, for each channel, a binary mask of detected fluorescent marker concentrations (spots or puncta). SPECHT leverages the Laplacian-of-Gaussian (LoG) object detection method, but ensures detection is adaptive to the image intensity distribution by computing an automatic threshold to postprocess LoG detected objects. The user can express a preference for recall or precision, which SPECHT then translates into channel/image specific threshold values. This preference is referred to in this manuscript as ‘z-value’. A higher value increases precision, at the cost of recall. A lower value can result in higher recall, at cost of precision. To ensure no artificial objects are introduced, the isotropic Gaussian std. dev. was set to round(precision/2) = 3 pixels (pixel size = 56.6 nm). Objects with area smaller than 25 pixels were removed to avoid false counting of artifacts below the precision limit of the acquisition. Distances between objects were measured using Euclidean distance (pixels) between the closest edges of nearest objects. Puncta within 5 pixels (ceil(precision)) of mitochondria (i.e. the resolution limit of ~ 250 nm) cannot be distinguished from overlapping puncta and were thus counted as overlapping. The area of objects is represented by pixel mask counting. When a red mRFP and green GFP object (puncta) shared a non-zero intersection, the union of the red-green overlapping puncta were considered to correspond to early, neutral pH autophagosomes. Colocalized GFP-mRFP tfLC3 puncta upon BafA1 treatment encompass acidic autophagolysosomes and an increase in GFP-mRFP tfLC3 puncta following BafA1 treatment is a measure of autophagic flux. We also quantified the number of BafA1-induced GFP-mRFP overlapping tfLC3 puncta within 5 pixels of mitochondria.
To ensure single cell analysis, ROIs encompassing complete, individual cells within the field of view were manually segmented (Fig. 6). While this was feasible for the single time point analysis, for the time lapse series (Fig. 8), cells were moving and we therefore added an automated preprocessing stage to obtain cell segmentation masks. Input to SPECHT was a sequence of 2D images (1 Z-slice), 3 channels per timepoint. We recover the outline by applying a median filter (window sizes 3x3, 5x5, 9x9) after filtering out the 90% intensity distribution quantile, binarizing the resulting image, and detecting disjointed objects separated by black (filtered) background using the connected components algorithm, such that the sole complete cell will be the largest object. To accommodate the highly fluctuating intensity distribution of live cell imaging over time, we enable SPECHT’s autotuning mode configured to recover all possible objects (recall/precision ratio = 3.75). To prevent inclusion of false positives we: 1) compute the effect size (Cohen’s d) of its intensity distribution with respect to that of the cell and discard any object with a negative effect size; 2) we use the heuristic that the local maxima contained within each detected object, should be a statistical outlier with respect to the overall intensity distribution (Q3 + 1.5 IQR, respectively 3rd quartile and interquartile range) and discard objects that have a maximum intensity that is not at the extremum of the intensity distribution; 3) objects of area smaller or equal to 4 pixels are discarded, as they cannot be shown to be observable under the precision of the system (2 pixels). To ensure we do not compromise objects at the cell edge we widen the cell mask by a dilation operation 4 times (2 x system precision of 2 pixels). A closing operation ensures no holes are left in the cell mask should one channel have no or weak labelling in part of the resulting mask. To ensure our segmentation is valid, without the user having to screen each image, we test that the cell mask is consistent across channels. In addition, we disregard processing of any image where the cell mask touches the border of the image. The combination of high recall followed by high precision filtering, ensures a balanced, robust automatic pipeline for object detection designed for the live cell imaging data. For each GFP-mRFP overlapping spot (C12), we compute the mean mitochondria intensity it overlaps relative to the mean intensity of the associated mitochondria segment as well as the mean intensity of all mitochondria segments in each given 2D image (C). We then compute a box plot of ratios 1 and 2 for all C12 objects, for each cell (Fig. 8), where each cell is represented by 7 2D images (per Z-slice), over 40 timepoints. Output is saved in csv files for statistical analysis and postprocessing. The processing code is being prepared for open-source release (Affero GPLv3), and available upon reasonable request.
One-way ANOVA with Dunnett’s multiple comparison test was used for both the fixed and live-cell ROS experiments. One-way ANOVA with Tukey’s multiple comparison test was used for the live cell tfLC3 flux, the mitoKeima experiments, the spinning disk mitochondrial volume experiments and the Western blot flux experiments. A 2 tailed t-test was applied for the siATG5 blots. Statistical analyses were performed using GraphPad Prism 6.0 software.