PINK1 is selectively stabilized on impaired mitochondria to activate Parkin

Abstract

Loss-of-function mutations in PINK1 and Parkin cause parkinsonism in humans and mitochondrial dysfunction in model organisms. Parkin is selectively recruited from the cytosol to damaged mitochondria to trigger their autophagy. How Parkin recognizes damaged mitochondria, however, is unknown. Here, we show that expression of PINK1 on individual mitochondria is regulated by voltage-dependent proteolysis to maintain low levels of PINK1 on healthy, polarized mitochondria, while facilitating the rapid accumulation of PINK1 on mitochondria that sustain damage. PINK1 accumulation on mitochondria is both necessary and sufficient for Parkin recruitment to mitochondria, and disease-causing mutations in PINK1 and Parkin disrupt Parkin recruitment and Parkin-induced mitophagy at distinct steps. These findings provide a biochemical explanation for the genetic epistasis between PINK1 and Parkin in Drosophila melanogaster. In addition, they support a novel model for the negative selection of damaged mitochondria, in which PINK1 signals mitochondrial dysfunction to Parkin, and Parkin promotes their elimination.

Conflict of interest statement

The authors have declared that no competing interests exist.

Figures

Figure 1. PINK1 selectively accumulates on depolarized mitochondria.

(A) HeLa cells stably expressing YFP-Parkin were treated with 10 µM CCCP in serum at time point 0, fractionated, and carbonate extracted. The carbonate-extracted pellet, which is enriched in integral mitochondrial proteins, was run on SDS gels and immunoblotted for endogenous PINK1 and the mitochondrial protein VDAC. HeLa cells stably expressing YFP-Parkin were used in our initial experiments because it was unclear whether the stability of PINK1 would be affected by the absence of Parkin, as has been reported previously . (B) M17 human neuroblastoma cells stably transduced with control shRNA or PINK1 shRNA were treated with 20 µM CCCP in serum and fractionated. The mitochondria-rich membrane fraction was run on SDS gels and immunoblotted as in (A). (C) E18 rat cortical neurons (7 d in vitro) were transfected with PINK1-V5. The next day the cells were treated with 1 µM CCCP for 6 h. Whole-cell lysates were run on SDS page gels and immunoblotted as in (A). (D) Live-cell imaging of HeLa cells transfected with PINK1-YFP (green) were treated with 10 µM CCCP in serum at time point 0. Mitochondria were labeled by pulsing with Mitotracker Red (MTR) (red) before depolarization with CCCP. (E) Mfn1/2 null MEFs transfected with PINK1-YFP (green). All mitochondria were stained with antibody against cytochrome c (Cyto C; white) and bioenergetically coupled mitochondria were stained by pulsing cells with Mitotracker Red (MTR) (red). (F) Average MTR intensity/pixel for PINK1-negative mitochondria (PINK1− Mito) and PINK1-positive mitochondria (PINK1+ Mito), respectively, was measured in eight or more cells in two independent experiments. Data from a representative experiment are shown. a.u., arbitrary units. (G) HeLa cells transfected with PINK1-YFP (green) and treated for 16 h with 2 mM paraquat. Cells were pulsed with MTR (red), fixed, and immunostained for cytochrome c (white). (H) The Pearson coefficient indexes between PINK1-YFP intensity and cytochrome c intensity, and PINK1-YFP intensity and MTR intensity were determined for eight or more cells in two independent experiments. Data from a representative experiment are shown. (I) HeLa cells were transfected with CFP-Parkin (red) and PINK1KD-YFP (green), which served as a reporter for endogenous PINK1 accumulation, and treated for 16 h with 2 mM paraquat. Cells were pulsed with MTR (white) and fixed. (J) The Pearson coefficient indexes between PINK1 KD-YFP intensity and CFP-Parkin intensity and PINK1 KD-YFP intensity and MTR intensity were determined for seven or more cells in two independent experiments. Data from a representative experiment are shown.

Figure 2. PINK1 accumulates following inhibition of voltage-sensitive cleavage.

(A) HeLa cells stably expressing YFP-Parkin were treated with DMSO for 3.5 h, 2 µM CCCP for 3.5 h, or CCCP for 3 h followed by washout of CCCP for 0.5 h in the absence of serum. 50 µM MG132 and/or 100 µM cyclohexamide were added for the last 1 h of treatment. Whole-cell lysates (WCL) run on SDS gels and immunoblotted for endogenous PINK1 and tubulin. (B) Model depicting the two-step processing of PINK1. (C) Quantitative RT-PCR was used to measure relative PINK1 mRNA expression in HeLa cells treated with DMSO or CCCP for 1 h. The graph represents the results from four independent experiments. As a positive control, relative PINK1 mRNA levels were also measured in HeLa cells following exogenous expression of PINK1. PINK1 mRNA expression levels were normalized to the housekeeping gene β-actin. WT, wild type.

Figure 3. Parkin recruitment to depolarized mitochondria requires PINK1 and its mitochondrial targeting N-terminus.

(A) Primary MEFs from PINK1+/+ or PINK1−/− mice cotransfected with YFP-Parkin (green) and the indicated construct (vector, PINK1-V5, PINK1 kinase-deficient [KD]-V5, or PINK1 156–581 [ΔN]-V5) in a 1∶4 ratio were treated with DMSO or 20 µM CCCP in serum for 3 h. Mitochondria were immunostained for Tom20 (red). The images in the column on the right, which are merged images of the middle and left-hand columns, are expansions of the boxed regions in the middle column. (B) Colocalization between YFP-Parkin and mitochondria in (A) was scored for ≥100 cells/condition in three or more independent experiments. (C) Transformed MEFs from independently generated PINK1+/+ and PINK1−/− mice were transfected and treated as in (A) and were scored as in (B). (D) M17 human neuroblastoma cells stably transduced with control shRNA or PINK1 shRNA were treated with 10 µM CCCP in serum for 3 h and imaged as in (A). (E) Colocalization between YFP-Parkin and mitochondria in (D) was scored as described in (B). (F) Control shRNA and PINK1 shRNA M17 cells transfected and treated as in (D) were fractionated into mitochondria-rich membrane fraction (Memb) and supernatant (Sup). Fractions were run on SDS gels and immunoblotted with anti-Parkin and anti-VDAC antibodies. Loading was adjusted for approximately equal concentrations of YFP-Parkin in the postnuclear supernatants (PNS) between the two cell types. Scale bars in (A and D) represent 10 µm. Error bars in (B, C, and E) indicate standard deviation.

Figure 4. PINK1 is required for Parkin-induced autophagy of depolarized mitochondria.

(A) Primary MEFs from PINK1+/+ or PINK1−/− mice cotransfected with YFP-Parkin were treated with DMSO or 20 µM CCCP in serum for 24 h. Mitochondria were stained with an anti-Tom20 antibody. (B) Percentage of cells with no detectable mitochondria in (A) was scored for >150 cells/condition in three or more independent experiments. (C) M17 human neuroblastoma cells stably transduced with control shRNA or PINK1 shRNA were treated with 10 µM CCCP for 24 h and stained as in (A). Images in the bottom rows of (A and C) are expansions of the images indicated by the boxes in the middle rows. Scale bars in (A and C) represent 10 µm. (D) Percentage of cells with no mitochondria was scored for (C) as described in (B). (E) M17 cells stably transduced with control shRNA or PINK1 shRNA were treated with DMSO or 10 µM CCCP for 24 h and stained with Mitotracker Green (MTG). MTG, which stains mitochondrial lipid in a membrane potential independent manner, is a sensitive measure of mitochondrial mass. The graph represents change in Mitotracker Green intensity between DMSO- and CCCP-treated samples in three independent experiments. (F) M17 cells stably transduced with control shRNA or PINK1 shRNA were pulsed with Mitotracker Green in the presence of CCCP. Loss of MTG intensity was measured at 0 h, 16 h, and 24 h with a plate reader. The graph shows data from three biological replicates and is representative of three independent experiments. The error bars in (B, D, and E) indicate the standard deviation.

Figure 5. Kinetics of Parkin recruitment are modulated by PINK1 expression.

(A) HeLa cells transfected with mCherry-Parkin (red) alone or mCherry-Parkin (red) and PINK1-YFP in a 1∶1 ratio were imaged live following the addition of 10 µM CCCP in serum at time point 0 min. (B) HeLa cells transfected with mCherry-Parkin and the indicated construct in a 1∶1 ratio were treated as in (A) and imaged live (one frame/minute) following the addition of CCCP. Time to the beginning of Parkin translocation was defined as the first appearance of puncta in two or more quadrants of the cell for two or more consecutive images for six or more cells in a minimum of three independent experiments. N.S., nonsignificant. (C) Live confocal image of HeLa cells transfected with YFP-Parkin (green) or YFP-Parkin (green) and PINK1-myc (in a 1∶4 ratio). Cells were loaded with TMRE (red) to stain polarized mitochondria. Cells were not treated with CCCP. Scale bar in last image represents 10 µm. Images in the middle and right-hand panels are expansions of the boxed regions in the panels on the left. (D) Cells treated as described in (C) were scored for colocalization between YFP-Parkin and TMRE. ≥50 cells/experiment were scored in three or more independent experiments.

Figure 6. Stable expression of PINK1 on the outer mitochondrial membrane is sufficient for Parkin recruitment.

(A) Schematic diagram depicting the construction of PINK1-YFP (green), PINK1 (111–581)-YFP (green), and OPA3-PINK1 (111–581)-YFP (green). (B) Confocal images depicting the localization of PINK1-YFP, PINK1 (111–581)-YFP, and OPA3-PINK1 (111–581)-YFP in HeLa cells. Mitochondria are stained with the potentiometric dye TMRE (red). (C) HeLa cells were transfected with PINK1-YFP, PINK1 (111–581)-YFP, or Opa3-PINK1 (111–581)-YFP and treated with DMSO or 2 µM CCCP in serum-free medium for 3 h. Whole-cell lysates (WCL) were run on SDS gels and immunoblotted for PINK1, GFP, and tubulin. (D) Confocal images of HeLa cells cotransfected with mCherry-Parkin (red) and PINK1-YFP (green), PINK1 (111–581)-YFP (green), or OPA3-PINK1 (111–581)-YFP (green). Cells were not treated with CCCP. (E) HeLa cells in (D) were scored for mCherry-Parkin forming puncta characteristic of mitochondria in ≥150 cells in three or more independent experiments. Cells were not treated with CCCP. (F) HeLa cells were transfected with FRB-PINK1 (111–581)-YFP, which is in the cytosol, TOM20(1–33)-FKBP, which is on mitochondria, and mCherry-Parkin. In the presence of the rapamycin analog, AP21967, the FRB and FKBP domains of the respective fusion proteins (PINK1 (111–581) and TOM20's outer mitochondrial membrane anchor) heterodimerize, if they have access to the same compartment (e.g., the cytosol). Cells treated with vehicle or 250 nM AP21967 for 8 h were scored for mCherry-Parkin in puncta characteristic of mitochondria in ≥150 cells in three or more independent experiments. (G) Confocal images of HeLa cells transfected with PINK1-YFP (green), PINK1 (111–581)-YFP (green), or OPA3-PINK1 (111–581)-YFP (green) with or without ECFP-Parkin and cultured for 96 h in the absence of CCCP. Cells were immunostained for Tom20 (red). To aid in visualizing cells that lack mitochondria, some individual cells have been outlined. (H) Cells treated as in (G) were scored for the absence of detectable mitochondria in ≥150 cells in three or more independent experiments. Scale bars in all images represent 10 µm.

Figure 7. PINK1 accumulation following depolarization with CCCP may be required for Parkin recruitment.

(A) HeLa cells stably expressing YFP-Parkin were treated with 2 µM CCCP 1 h alone or CCCP 1 h + 2 µM CHX (30-min pretreatment and 1-h treatment) in the absence of serum. Whole-cell lysates were run on SDS gels and immunoblotted for endogenous PINK1 and the loading control GAPDH. (B) Cells treated as in (A) were fractionated. The mitochondria-enriched membrane fraction (Memb) was run on SDS gels and immunoblotted for endogenous PINK1 and VDAC. (C) HeLa cells were transfected with YFP-Parkin (green) and treated with 10 µM CCCP 1 h alone, CCCP + 10 µM of actinomycin (30-min pretreatment and 1-h treatment), or CCCP 1 h + 100 µM CHX (30-min pretreatment and 1-h treatment) in the presence of serum and immunostained for Tom20 (red). (D) Colocalization between YFP-Parkin and mitochondria in (C) was scored for ≥150 cells/condition in three or more independent experiments. (E) HeLa cells stably expressing YFP-Parkin were treated as in (A) and fractionated. The mitochondria-rich fraction was run on an SDS gel and immunostained for Parkin. Scale bars in all images represent 10 µm.

Figure 8. Disease-causing PINK1 mutants fail to reconstitute Parkin recruitment to depolarized mitochondria.

(A) Primary MEFs from PINK1−/− mice cotransfected with YFP-Parkin (green) and indicated V5-tagged PINK1 constructs in a 1∶4 ratio were treated with DMSO or 20 µM CCCP in serum for 3 h. Mitochondria were stained with an anti-Tom20 antibody (red). Scale bar in images represents 10 µm. Images in the middle and bottom rows are expansions of the images indicated by the boxes in the top row. (B) Colocalization between YFP-Parkin and mitochondria in (A) was scored for >150 cells/condition in three or more independent experiments. Error bars indicate standard deviation. (C) HeLa cells stably expressing YFP-Parkin were transfected with the indicated V5-tagged constructs, treated with DMSO or 2 µM CCCP for 3 h in serum-free medium, and fractionated. The mitochondria-rich membrane fraction was run on an SDS gel and immunoblotted for PINK1, the V5 tag, and the mitochondrial protein VDAC.

Figure 9. Disease-causing mutations in Parkin disrupt Parkin recruitment to mitochondria and/or Parkin-induced mitophagy.

(A) HeLa cells were transfected with YFP-Parkin (white and green) containing indicated mutations and treated with CCCP for 1 h. Mitochondria labeled with an anti-Tom20 antibody (red). Images in the bottom row are expansions of the regions indicated by the boxes in the middle row. WT, wild type. (B) Colocalization between YFP-Parkin and mitochondria in (A) scored for ≥150 cells/condition in three or more independent experiments. (C) HeLa cells transfected and treated as in (A) and fractionated into postnuclear supernatant (PNS), mitochondria-rich heavy membrane fraction (HMF), and supernatant (Sup). Fractions run on SDS gels and immunoblotted for Parkin and VDAC. (D and E) HeLa cells transfected as in (A) and treated with CCCP or DMSO for 24 h. (D) Number of HeLa cells with no mitochondria scored for ≥150 cells/condition in three or more independent experiments. (E) Images of WT, R42P, and R275W Parkin (green) stained as in (A). An asterisk (*) indicates engineered mutation; all others have been linked to Parkinson disease. Scale bars in all images represent 10 µm. Error bars in (B and D) indicate standard deviation.