Parkinson’s disease–associated VPS35 mutant reduces mitochondrial membrane potential and impairs PINK1/Parkin-mediated mitophagy

Human SH-SY5Y neuroblastoma cells were maintained in Dulbecco’s Modified Eagle’s Medium (Invitrogen, Waltham, MA) supplemented with 15% fetal bovine serum (Invitrogen) and 1% Penicillin-Streptomycin (Gibco, Waltham, MA) in a 37 °C incubator with 5% CO₂. Transient plasmid transfections were performed with plasmid DNAs using Lipofectamine (Thermo Fischer Scientific, Waltham, MA), according to the manufacturer’s instructions. To induce mitochondrial depolarization, the SH-SY5Y cells were treated with 10 μM or 20 μM carbonyl cyanide m-chlorophenylhydrazone (CCCP) (Sigma-Aldrich, Saint Louis, MO), 1 μM oligomycin (Sigma-Aldrich), 1 μM antimycin A (Sigma-Aldrich), or 1 μM antimycin A and 1 μM oligomycin (AO), for the indicated times, prior to cell harvesting or fixation. DMSO treatment was used as a control.

The plasmids used were pEGFP-Parkin [36] (a gift from Prof. Edward Fon (McGill University, Montreal, Quebec, Canada), Addgene plasmid #45875) and pEGFP-LC3 (a gift from Prof. Toren Finkel (University of Pittsburgh, Pittsburgh, PA), Addgene plasmid #24920) constructs [37]. The primary antibodies used for immunoblotting were mouse anti-ATPIF1 (1:1000, Abcam, Cambridge, UK, ab110277), mouse anti-β-actin (1:5000, MP Biomedicals, Irvine, CA, 8691001), mouse anti-β-tubulin (1:5000, Sigma-Aldrich T4026), mouse anti-Parkin (1:500, Santa-Cruz Biotechnology, Dallas, TX; sc-32,282), rabbit anti-PINK1 (1:1000, Cell signaling, Danvers, MA, #6946), mouse anti-TOM20 (1:500, BD Biosciences, San Jose, CA; 612278) and goat anti-VPS35 (1:1000, Abcam, ab10099). The primary antibodies used for immunofluorescence (IF) were mouse anti-TOM20 (1:200, Santa-Cruz Biotechnology sc-17764) and rabbit anti-PINK1 (1:200, Abcam ab216144). Secondary antibodies for immunoblotting were HRP-conjugated goat anti-rabbit IgG (H + L) (1:10000, Bio-Rad, Hercules, CA), HRP-conjugated goat anti-mouse IgG (H + L) (1:10000, Bio-Rad) and HRP-conjugated donkey anti-goat IgG (H + L) (1:10000, Abcam). Secondary antibodies for IF were Cy3-conjugated donkey anti-mouse IgG (H + L) (1:250, Jackson ImmunoResearch, West Grove, PA) and Alexa Fluor 488-conjugated donkey anti-rabbit IgG (H + L) (1:250, Jackson ImmunoResearch).

The D620N mutation in the VPS35 gene was obtained by Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR)-Cas9–mediated genome editing in the SH-SY5Y neuroblastoma cell line, as previously described [38]. Briefly, a 20-nt single guide RNA (sgRNA) sequence that targets exon 15 of the VPS35 gene and is predicted to cut approximately 9 base-pairs (bp) upstream of the GAT triplet encoding the aspartic acid residue on location 620 was cloned into the pSpCas9(BB)-2A-GFP (PX458) plasmid (a gift from Prof. Feng Zhang (Broad Institute, Cambridge, MA); Addgene plasmid #48138) using the BbsI restriction enzyme to form the targeting plasmid expressing Cas9-GFP. In addition, a single-stranded oligodeoxynucleotide sequence was designed to facilitate homology-directed repair of the endogenous locus and included the substitution of five nucleotides: a nucleotide substitution G > A that leads to the D620N mutation of VPS35 and four synonymous substitutions that create a novel EcoRI restriction site that also destroys the protospacer-adjacent motif sequence to avoid repetitive cutting of Cas9 by the repair template. Following validation, the PX458-sgRNA plasmid and the single-stranded oligonucleotides were transfected into the SH-SY5Y cells following the manufacturer’s protocol (Lonza, Basel, Switzerland). GFP-positive cells were single-cell sorted 48 h post-transfection using a SH800S cell sorter (Sony Biotechnology, San Jose, CA) and grown in separate cultures that were subsequently screened for the D620N mutation using the restriction enzyme EcoRI. In parallel, we mock-electroporated and sorted the same batch of cells, which were used as WT control in the following experiments. Finally, we sequenced the top three predicted off-target genomic regions within coding regions (obtained from http://​crispr.​mit.​edu) of genes POU6F1, ZNF318 and KY, but found no off-target edits (not shown). Detailed primer and template sequences are provided in Table S1.

The COX8-EGFP-mCherry sequence was obtained from the pCLBW COX8-EGFP-mCherry construct [39] (a gift from Prof. David Chan (Caltech, Pasadena, CA), Addgene plasmid #78520) through restriction enzyme digestion with ApaI and EcoRI, and was ligated into the mammalian expression vector pcDNA 3.1(+). Subsequently, the vector was transfected into WT and VPS35^D620N SH-SY5Y cells using Lipofectamine 3000 (Thermo Fischer Scientific), following the manufacturer’s protocol. Forty-eight hours after transfection, the growth medium was replaced with selection medium containing 800 ng/μl G-418 (Sigma-Aldrich). The selection medium was refreshed every other day for 10 days until only cells with the plasmid remained. Stable cell lines were cultured for three passages before performing the experiments.

SH-SY5Y cells were harvested in 2% sodium dodecyl sulfate (SDS)/phosphate-buffered saline (PBS) buffer containing a proteinase inhibitor cocktail (Roche, Basel, Switzerland) and sonicated. Crude mitochondrial fractions were isolated as previously described [40]. Briefly, SH-SY5Y cells were collected and homogenized using a Dounce homogenizer in ice-cold isolation buffer containing 320 mM sucrose and a proteinase inhibitor cocktail. The homogenized samples were differentially centrifuged at 1500 g for 15 min and 17,000 g for 30 min to obtain nuclei and crude mitochondria, respectively. The cytosolic fraction was obtained from the final supernatant. Protein concentrations were quantified using the Pierce™ BCA protein assay kit (Thermo Fischer Scientific), and samples were mixed with loading buffer containing 10% β-mercaptoethanol before being boiled at 95 °C for 5 min. Subsequently, equal amounts of total protein extracts were subjected to SDS-PAGE, transferred to nitrocellulose membranes, blocked for 1 h in skimmed milk, incubated overnight with primary antibody at 4 °C and then with the corresponding secondary antibody for 1 h at room temperature (RT). The blots were imaged on a Chemidoc™ MP Imaging System (Bio-Rad). Protein levels were quantified by densitometry using the ImageJ software (NIH, Bethesda, MD).

WT and VPS35^D620N SH-SY5Y cells that were seeded on glass coverslips in 24-well plates were fixed in 4% paraformaldehyde in PBS for 10 min at RT. Cells were then permeabilized in 0.1% Triton X-100 in PBS for 10 min and blocked with 5% donkey serum (Abcam) in PBS for 1 h. The coverslips were then incubated overnight at 4 °C with the primary antibodies diluted in blocking buffer and for 1 h at RT for secondary antibody incubation. Coverslips were finally mounted onto glass slides in 4′,6-diamidino-2-phenylindole (DAPI)-containing mounting medium (Vector Laboratories, Burlingame, CA). The slides were analyzed using either structured illumination microscopy (SIM) or confocal microscopy. SIM images were acquired with an AxioObserver Z1 compound microscope (Carl Zeiss, Oberkochen, Germany) equipped with an Apotome, 63x oil objective and an AxioCam MRm3 CCD camera (Carl Zeiss). Confocal images were acquired with a TCS SP8 high-resolution confocal laser scan microscope (Leica Microsystems, Wetzlar, Germany) and an HC PL APO CS2 63x/1.4 oil objective. For quantitative analysis, maximum intensity projections were generated from all Z-stacks, which were captured for each condition with identical exposure times or laser settings.

All image analyses were performed using the ImageJ software (NIH). Colocalization analyses of PINK1 and TOM20 were performed using ImageJ plugin Coloc 2 (https://​imagej.​net/​Coloc_​2). Regions of interest (ROIs) were created per cell in the TOM20 channel (n = ~ 80–100 cells per condition in each experiment). Pearson’s correlation coefficients were subsequently determined per ROI using the Costes method for threshold regression [41].

Mitochondrial morphology was quantified as previously described [42]. Briefly, images of single cells were pre-processed and binarized, followed by particle analysis and computation of several metrics. The number of mitochondria was determined as the number of individual particles. The aspect ratio was determined by dividing the major axis by the minor axis of each particle. A total of 80–100 cells were quantified per condition in each experiment.

For the EGFP-Parkin translocation experiment, a blinded observer scored each cell for either diffuse EGFP-Parkin or mitochondria-localized EGFP-Parkin (n = ~ 50 cells per condition in each experiment), as previously described [36].

Mitochondrial membrane potential was measured using fluorescence-activated cell sorting (FACS). WT and VPS35^D620N SH-SY5Y cells were incubated for 30 min with 100 nM tetramethylrhodamine methyl ester (TMRM) dye (Thermo Fischer Scientific) and 100 nM MitoTracker Green FM dye (Thermo Fischer Scientific) diluted in culture medium. Cells were rinsed, dissociated with 0.05% Trypsin-EDTA (Thermo Fischer Scientific) and aliquoted in multiple FACS tubes. FACS measurements were performed with a FACSCalibur flow cytometer (BD Biosciences) or a Novocyte Quanteon flow cytometer (Agilent, Santa Clara, CA ). For the timeline measurements, baseline measurements were taken, after which CCCP was added to a final concentration of 10 μM, upon which measurements were taken at each time point. For the dose-response measurements, CCCP was added to the indicated final concentrations, and 1 min later measurements were made. Slopes were determined using a simple linear regression method. For 24-h treatments, cells were treated with 10 μM CCCP or 1 μM antimycin A and 1 μM oligomycin for 24 h, then the cells were dissociated and TMRM was measured as indicated above. Three independent sorts measuring at least 10,000 cells were performed per clone per data point for all experiments. Data analysis was performed using the Kaluza Analysis software (Beckman Coulter, Brea, CA). Mitotracker Green median fluorescence intensity was used to correct for mitochondrial mass fluctuations.

For conventional transmission electron microscopy (TEM), WT and VPS35^D620N SH-SY5Y cells were treated with DMSO or 10 μM CCCP for 6 h. Then an equal volume as the culture media of double-strength fixative (4% paraformaldehyde, 5% glutaraldehyde in 0.1 M sodium cacodylate buffer, pH 7.4) was then added to the cells and incubated for 20 min at RT, followed by further fixing the cells with the same volume of single-strength fixative (2% paraformaldehyde, 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer, pH 7.4) for 2 h at RT. After five washes with 0.1 M sodium cacodylate buffer (pH 7.4), the cells were scraped and embedded as previously described [43]. Subsequently, 70-nm ultrathin sections were cut using a Leica EM UC7 ultra microtome (Leica Microsystems) and stained with uranyl acetate and lead citrate as previously described [43]. The cell sections were analyzed using an 80 kV transmission electron microscope CM100bio TEM (FEI, Eindhoven, The Netherlands).

The analysis of the different mitochondrial profiles per cell type was performed by random screening of sections derived from at least three different grids per sample. The mitochondrial profiles were categorized as follows: classical mitochondria with well-defined cristae (category I), dark mitochondria with well-defined cristae often swelling (category II), mitochondria with undefined cristae (category III), dark mitochondria with undefined contours and cristae (category IV), and large mitochondria with very light content and few remnant cristae (category V). The number of each mitochondrial type per condition was determined by counting 665, 579 and 521 mitochondria profiles from the DMSO-treated WT and VPS35^D620N cell (clones 1 and 2) sections, respectively, as well as 727, 1028 and 914 mitochondria profiles from the CCCP-treated WT and VPS35^D620N cell (clones 1 and 2) sections, respectively.

Cell viability upon treatment with CCCP was determined using a 3-(4,5-demethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) reduction assay (Abcam). The SH-SY5Y cells were plated in 96-well plates 12 h prior to incubation with CCCP for 24 h. MTT assay compounds were added following the manufacturer’s protocol and absorbance was measured using a Synergy HT optical plate reader (Biotek, Winooski, VT).

Data of Western blot densitometry measurements, mitochondrial membrane potential measurements, TEM and mitochondrial morphology were analyzed using a linear model by one-way or two-way analysis of variance (ANOVA) followed by Tukey’s post-hoc test. Count data, such as the EGFP-LC3 and mitochondrial particle quantification data were modelled using a generalized linear model followed by one-way or two-way ANOVA and Tukey’s post-hoc test. PINK1-TOM20 colocalization data were analyzed using the Kruskal-Wallis test followed by pairwise Mann-Whitney U-test with Benjamini-Hochberg multiple testing correction. Proportional data of COX8-EGFP-mCherry mitophagy and EGFP-Parkin localization were analyzed using beta regression analysis. Data are means ± SEM from at least three independent experiments, unless otherwise specified. P < 0.05 was considered as statistically significant. Statistical analyses were performed in the statistical computing environment R (version 1.3.959; The R Foundation for Statistical Computing; Vienna, Austria).

To date, most studies have investigated the p.D620N variant in VPS35 (VPS35^D620N) by stably overexpressing VPS35^D620N in in vitro and in vivo models. However, the enhanced VPS35 levels in these models may affect the retromer functioning, as higher or lower levels of VPS35 correlate with alterations in mitochondrial fragmentation [14, 15]. This motivated us to use CRISPR-Cas9–mediated genome editing to introduce the p.D620N variant in VPS35 into the human neuroblastoma SH-SY5Y cells widely used in PD research [38, 44] (Fig. 1a). Restriction fragment length polymorphism analysis using the EcoRI enzyme on a 604-bp genomic DNA region surrounding the variant revealed two putative positive clones (Fig. 1b). Sanger sequencing validated the presence of the p.D620N variant, resulting in a GAT-to-AAT codon change, in only one of two VPS35 alleles, mimicking the heterozygous carrier status seen in patients (Fig. 1c). Additionally, immunoblotting showed that the introduction of the p.D620N variant did not affect the expression level of VPS35 compared to the WT cells (Fig. 1d). These cell lines were used for the experiments of this study.

To investigate whether the p.D620N variant in VPS35 affects the PINK1/Parkin-mediated mitophagy, we used the protonophore CCCP to induce mitochondrial stress by dissipating the mitochondrial membrane potential (Δψ_m) and thereby activate PINK1/Parkin-mediated mitophagy [21, 24, 25]. We used immunoblotting to investigate PINK1 accumulation over time in the WT and VPS35^D620N cells upon 10 μM CCCP treatment. As expected, the total PINK1 level increased slightly after 3 h of CCCP treatment, and PINK1 accumulation was pronounced after 24 h of CCCP treatment in the whole cell extracts and crude mitochondrial fractions of WT cells (Fig. 2c, d; Fig. S1a). This coincided with Parkin accumulation in the crude mitochondrial fraction (Fig. 2c, d) and a decrease in the whole lysate (Fig. S1b, c; Fig. 2c), likely due to the autoubiquitination and increased proteasomal turnover of mitochondria-bound Parkin [45]. However, the total PINK1 level was substantially lower in the whole extracts and crude mitochondrial fractions of CCCP-treated VPS35^D620N cells at both time points compared to the WT cells (Fig. S1a, Fig. 2a–d). Likewise, the total Parkin levels remained similar to those in the untreated condition (Fig. S1a, Fig. 2c). Of note, the VPS35 levels did not change upon CCCP treatment (Fig. S1a; Fig. 2c, d), and VPS35 was present in the crude mitochondrial fraction (Fig. 2c), consistent with previous reports [14, 15].

Previous studies have shown a dose-dependent effect of CCCP and thus we questioned whether a higher dose of CCCP would be able to stabilize PINK1 on mitochondria in the VPS35^D620N cells. Indeed, 20 μM CCCP led to higher PINK1 levels compared to 10 μM CCCP in WT cells after 24 h of treatment and marked PINK1 accumulation in the VPS35^D620N cells (Fig. 2a, b). However, the total PINK1 level in the VPS35^D620N cells remained significantly lower than those in the WT cells (Fig. 2a, b). Consistent with the increased PINK1 level upon 20 μM CCCP treatment, the proteasomal degradation of Parkin in the WT cells also further increased with 20 μM CCCP [45], which was not observed in the VPS35^D620N cells (Fig. S1b, c). Of note, 20 μM CCCP demonstrated increased cytotoxicity compared to 10 μM CCCP (Fig. S1d). These data suggest that 10 μM CCCP induces milder damage to mitochondria than 20 μM CCCP and exposes a not-yet-characterized deficit in the VPS35^D620N clones.

Next, we used IF to quantify the translocation of cytosolic EGFP-Parkin to mitochondria upon treatment with CCCP, since endogenous Parkin was not detectable in our cells. WT and VPS35^D620N cells were transiently transfected with EGFP-Parkin and subsequently treated with 10 μM CCCP for 6 h and stained for OMM protein TOM20. As expected, mitochondrial depolarization due to CCCP caused translocation of cytosolic EGFP-Parkin to mitochondria in WT cells, as shown by the colocalization between EGFP-Parkin and TOM20 (Fig. 2e). Additionally, less Parkin translocation was seen in the VPS35^D620N cells (± 23% and ± 25%) compared to the WT cells (±63%) (Fig. 2f). We further examined colocalization between endogenous PINK1 and TOM20 using IF in WT and VPS35^D620N cells upon CCCP treatment (Fig. 3a). The CCCP-treated VPS35^D620N cells showed less colocalization between PINK1 and TOM20 compared to the WT cells, with a dose-dependent effect of CCCP (Pearson correlation coefficient: 10 μM CCCP, WT median 0.12 vs clone 1 median 0.04 and clone 2 0.05; 20 μM CCCP, 0.52 vs 0.36 and 0.35) (Fig. 3b). Together, these data suggest that the CCCP-induced PINK1 accumulation is hampered in VPS35^D620N cells, leading to impaired Parkin recruitment onto the mitochondria.

To prove that the hampered PINK1 and Parkin recruitment onto mitochondria upon CCCP treatment does lead to compromised PINK1/Parkin-mediated mitophagy in VPS35^D620N cells, we used previously published dual color fluorescence-quenching EGFP-mCherry mitophagy reporter [39], and stably expressed it in the WT and VPS35^D620N cells. Under normal conditions, mitochondria emitted both a red and a green fluorescence signals, resulting in a yellow color (Fig. 4a). Mitochondria damaged by CCCP treatment were transported to lysosomes for degradation, and the EGFP fluorescent signal was quenched within this acidic organelle, leaving mainly a red fluorescent signal (Fig. 4a, b). At steady state, both WT and VPS35^D620N cells primarily showed a yellow reticulated mitochondrial network, with only a few red-only puncta, probably reflecting mitochondria within lysosomes, i.e. mitolysosomes, and there was no significant difference between the cell lines (Fig. 4a, b). In contrast, after 24 h of 10 μM CCCP treatment, WT cells showed a substantial increase in mitochondria with a red-only signal, indicating activation of mitophagy [39], while VPS35^D620N cells did not display a shift from yellow to red-only mitochondria (Fig. 4a, b). Interestingly, punctate rearrangement of the mitochondrial network was observed in the VPS35^D620N cells after CCCP treatment, in which the mitochondrial clumps seemed larger compared to the WT cells (Fig. 4a, right bottom panel compared to left bottom panel). This suggests that VPS35^D620N cells do react to CCCP but experience impairment in PINK1/Parkin-mediated mitophagy.

To confirm this finding, we investigated mitophagy using a different approach by transiently transfecting WT and VPS35^D620N cells with EGFP-LC3, a protein marker for autophagosomes [46]. Mitophagy was induced by 10 μM CCCP treatment for 6 h, and we subsequently used IF to examine the colocalization between LC3 puncta and TOM20 (Fig. 4c, arrowheads). CCCP treatment in the WT cells led to approximately twice the amount of LC3- and TOM20-positive mitophagosomes compared to the VPS35^D620N cells (Fig. 4d). Moreover, multiple VPS35^D620N cells did not form TOM20-positive autophagosomes, a phenomenon rarely seen in the WT cells (Fig. 4d). Altogether, these results confirm that the CCCP-induced mitophagy is impaired in VPS35^D620N cells.

Next, we questioned if PINK1/Parkin-mediated mitophagy in VPS35^D620N cells would be impaired by treatment with two agents that, like CCCP, also lead to substantial mitochondrial depolarization: subcomplex III inhibitor antimycin A and F₁F₀ ATPase inhibitor oligomycin [47]. Antimycin A causes a collapse of the proton gradient across the inner mitochondrial membrane by blocking the mitochondrial electron transport chain, whereas oligomycin inhibits the flow of protons through F₁F₀ ATPase inhibition, leading to a complete Δψ_m collapse. Importantly, antimycin A is also a potent generator of oxidative stress, which is known to induce PINK1/Parkin-mediated mitophagy as well [48‐50]. As shown by immunoblotting, antimycin A (1 μM, 24 h) alone was not sufficient to stabilize PINK1 levels in WT cells, while treatment with oligomycin (1 μM, 24 h) did (Fig. 5a, b). As seen with CCCP, the oligomycin-treated VPS35^D620N cells showed less accumulation of PINK1 and higher levels of Parkin compared to WT (Fig. 5a, b). Notably, co-incubation with AO caused high PINK1 accumulation and loss of Parkin in both WT and VPS35^D620N cells, and in a similar manner (Fig. 5a, b). To corroborate this finding, mitochondrial PINK1 accumulation was determined by IF after 24 h of 1 μM AO treatment (Fig. 5c). In agreement with our immunoblotting data, PINK1 colocalized with TOM20 in almost all WT and VPS35^D620N cells, and no differences in the level of colocalization were observed between the different cell lines (Fig. 5d). Finally, we monitored AO-induced mitophagy using the dual color mitophagy reporter stably expressed in WT and VPS35^D620N cells and observed no difference (Fig. 5e, f). Together, these findings show that PINK1/Parkin recruitment and mitophagy can occur in VPS35^D620N cells in response to specific kinds of mitochondrial damage. However, the type and/or severity of insult to the mitochondrial membrane potential determines whether or not PINK1/Parkin-mediated mitophagy is initiated in VPS35^D620N cells.

To investigate whether AO treatment caused a different type of mitochondrial damage from that by CCCP treatment, we examined the rearrangement of the mitochondrial network upon exposure to these treatments. To do so, we analyzed the TOM20 distribution using IF to study the morphological characteristics of mitochondria including the number, aspect ratio and length of mitochondria in AO- and CCCP-treated cells (Fig. S2a–d). Both treatments caused mitochondrial fragmentation (Fig. S2a), as evidenced by a substantial increase in mitochondrial particles (Fig. S2b) and decreases in aspect ratio (Fig. S2c) and mitochondrial length (Fig. S2d). However, AO treatment led to more fragmentation than CCCP treatment, as the number of mitochondrial particles was significantly higher (Fig. S2a, b). In addition, consistent with our previous results, no differences were observed between WT and VPS35^D620N cells upon AO treatment. Interestingly, upon CCCP treatment, the mitochondrial particles appeared less rounded and longer, as reflected by the increase in aspect ratio and length, respectively, in the VPS35^D620N cells compared to WT cells (Fig. S2b–d). These data suggest that mitochondria in cells respond differently to AO and CCCP treatment and that AO causes more severe mitochondrial damage/fragmentation than CCCP in all cell lines. Additionally, the mitochondria in VPS35^D620N cells are affected by the treatments, i.e. they display mitochondrial fragmentation, albeit to a lesser extent than in WT cells.

To further explore why VPS35^D620N cells were affected by CCCP-induced damage but did not activate PINK1/Parkin mitophagy, we investigated the Δψ_m collapse upon CCCP treatment. The Δψ_m collapse triggers mitochondrial fragmentation, PINK1 accumulation on mitochondria and induction of mitophagy [21, 29]. Δψ_m was measured with the cell-permeant fluorescent dye TMRM in WT and VPS35^D620N cells over time upon treatment with 10 μM CCCP. Although CCCP treatment rapidly decreased Δψ_m in both WT and VPS35^D620N cells after 1 min, and the Δψ_m gradually decreased further during the next 19 min (Fig. 6a), the collapse in Δψ_m was significantly lower in VPS35^D620N cells compared to the WT cells at all measured time points. Additionally, the VPS35^D620N cells exhibited a lower Δψ_m at resting condition (±25% less) compared to the WT cells (Fig. 6b). To test whether the diminished Δψ_m reduction after 10 μM CCCP treatment in the VPS35^D620N cells is dose-dependent, Δψ_m was measured 1 min after treatment with 1, 5, 10, 20 and 50 μM CCCP. A greater Δψ_m reduction was observed with higher concentrations of CCCP in WT cells (slope − 0.011). In the VPS35^D620N cells, however, Δψ_m reduction remained diminished compared to the WT cells irrespective of the applied dose of CCCP (slope − 0.002 and − 0.006 for clones 1 and 2, respectively) (Fig. 6c). Finally, Δψ_m was measured in WT and VPS35^D620N cells after 24 h of treatment with 10 μM CCCP or 1 μM AO to compare the effect of these treatments. In line with the previous results, Δψ_m was greatly reduced after 24 h of CCCP treatment in the WT cells but diminished in the VPS35^D620N cells. In contrast, Δψ_m remained higher after 24 h of AO treatment compared to CCCP and no difference was observed between the WT and VPS35^D620N cells. Together, these data reveal that the mitochondrial membrane potential in the mitochondria of VPS35^D620N cells is already altered at steady state and could explain the altered mitochondrial susceptibility to CCCP but not to AO. We hypothesized that the difference in Δψ_m reduction could be partly due to a difference in levels of ATPIF1, an endogenous inhibitor of the F₁F₀ ATPase that affects PINK1/Parkin-mediated mitophagy [51]. ATPIF1 has been shown to maintain Δψ_m by blocking the reversal of the F₁F₀ ATPase to inhibit the outflow of protons and prevent ATP consumption [52]. Subsequently, we investigated the ATPIF1 protein levels in WT and VPS35^D620N cells, and found higher ATPIF1 levels in the VPS35^D620N cells compared to the WT cells (Fig. 6e, f), suggesting a role for ATPIF1 in the mitophagy deficits observed in VPS35^D620N cells.

Given that our IF data on mitochondrial distribution were inconsistent with previous reports about cells (over)expressing VPS35^D620N [14, 15], likely due to the resolution limitations, we used TEM to study WT and VPS35^D620N cells at steady state and under CCCP-treated conditions (10 μM CCCP for 6 h). Here, we observed that the VPS35^D620N cells already had smaller, fragmented mitochondria compared to the WT cells at steady state (Fig. 7a), something that we had not observed with IF since the resolution of this technique is not sufficient to distinguish longer mitochondrial tubules from multiple fragmented mitochondria in close proximity. CCCP treatment led to mitochondrial fragmentation in WT cells that resembled the mitochondrial phenotype of VPS35^D620N cells at steady state (Fig. 7a, b). Notably, no further mitochondrial fragmentation was detected in the CCCP-treated VPS35^D620N cells compared to the mitochondrial fragmentation seen in the VPS35^D620N cells at steady state (Fig. 7b).

Furthermore, five morphologically distinct categories of mitochondria were observed in the various samples (Fig. 7c): (I) classical healthy mitochondria with well-defined cristae, (II) swollen mitochondria with defined cristae and dark in content, (III) mitochondria with unclear, partially visible cristae, (IV) mitochondria with very dark content and no visible cristae, and (V) aberrant mitochondria with remnants of cristae and light in content. At steady state, most mitochondria (~ 84%) in the WT cells were category I, and the remainder were categories II (~ 9%) and III (~ 5%). In contrast, in the VPS35^D620N cells at steady state, a large proportion (~ 45%) of the mitochondria were in category II, and we observed significantly fewer healthy category I mitochondria compared to the WT cells (Fig. 7d). Upon CCCP treatment, we observed a shift from category I (~ 60%) to category II (~ 26%) mitochondria in the WT cells, as well as an increase in category IV mitochondria (from ~ 1% to ~ 9%). This suggests that the category II mitochondria were damaged. Intriguingly, CCCP treatment did not cause a compositional shift in the mitochondrial population in the VPS35^D620N cells. These data suggest that VPS35^D620N cells at steady state already contain a population of damaged and fragmented mitochondria and, in agreement with our other results, confirm that this population of mitochondria does not respond further to CCCP treatment.