PINK1 autophosphorylation upon membrane potential dissipation is essential for Parkin recruitment to damaged mitochondria

Abstract

Dysfunction of PINK1, a mitochondrial Ser/Thr kinase, causes familial Parkinson's disease (PD). Recent studies have revealed that PINK1 is rapidly degraded in healthy mitochondria but accumulates on the membrane potential (ΔΨm)-deficient mitochondria, where it recruits another familial PD gene product, Parkin, to ubiquitylate the damaged mitochondria. Despite extensive study, the mechanism underlying the homeostatic control of PINK1 remains unknown. Here we report that PINK1 is autophosphorylated following a decrease in ΔΨm and that most disease-relevant mutations hinder this event. Mass spectrometric and mutational analyses demonstrate that PINK1 autophosphorylation occurs at Ser228 and Ser402, residues that are structurally clustered together. Importantly, Ala mutation of these sites abolishes autophosphorylation of PINK1 and inhibits Parkin recruitment onto depolarized mitochondria, whereas Asp (phosphorylation-mimic) mutation promotes mitochondrial localization of Parkin even though autophosphorylation was still compromised. We propose that autophosphorylation of Ser228 and Ser402 in PINK1 is essential for efficient mitochondrial localization of Parkin.

Figure 1. PINK1 is phosphorylated following a decrease in mitochondrial ΔΨm.

(a) Endogenous PINK1 in the presence of CCCP (10 μM, 1 h) recruits Parkin to mitochondria more efficiently than over-produced PINK1 (CMV promoter-driven) in the absence of CCCP. The number of cells with Parkin-positive mitochondria in the Parkin-expressing cells was counted in >100 cells. Error bars represent the mean±s.d. values of three experiments. Statistical significance was calculated using analysis of variance with a Tukey–Kramer post hoc test. Example figures indicative of colocalization and no colocalization are shown on the right (scale bars, 10 μm). (b) Immunoblotting of exogenous and endogenous PINK1. The asterisk indicates the position of endogenous PINK1. Actin was used as a loading control. FL, full-length PINK1; Δ1, the N-terminal processed form. (c) HeLa cells expressing non-tagged PINK1 were treated with 10 μM CCCP for the indicated times and subjected to SDS–PAGE on a 7.5% Tris-glycine gel, a 4–12% precast gel or a 50 μM phos-tag containing gel. Red asterisks show phosphorylated PINK1. We routinely used HeLa cells and immunoblotting analysis was conducted with an anti-PINK1 antibody unless otherwise specified. (d) In vitro synthesized PINK1-3HA and the mitochondrial fraction of PINK1-3HA-expressing cells following CCCP treatment were subjected to immunoblotting. (e) The cell lysate and immunoprecipitated product of PINK1^−/− MEFs-expressing WT or the PINK1 KD mutant were subjected to SDS–PAGE on a 7.5% Tris-glycine gel ±50 μM phos-tag. Red asterisks show phosphorylated PINK1. (f) The endogenous PINK1 in HeLa cells exists as the phosphorylated form. The black asterisk indicates a cross-reacting band and red asterisks show phosphorylated PINK1. (g) Phosphatase treatment caused the high-molecular shift of both exogenous and endogenous PINK1 to disappear. The mitochondrial fraction collected from HeLa cells-expressing PINK1-HA or collected from noninfected HeLa cells was treated with CIAP at the indicated temperature. Exogenous PINK1 was detected with an anti-HA antibody. + and ++ mean 10 and 30 U per reaction of CIAP, respectively. (h) Cells expressing exogenous PINK1 were treated with CCCP (10 μM, 1 h), valinomycin (10 μM, 1 h) or rotenone (200 μM, 24 h) and subjected to SDS–PAGE on a 7.5% Tris-glycine gel ±50 μM phos-tag.

Figure 2. PINK1 phosphorylation on damaged mitochondria is inhibited by most pathogenic mutations.

(a) HeLa cells expressing PINK1-Flag with various pathogenic mutations were treated with CCCP, subjected to SDS–PAGE ± phos-tag and immunoblotted using an anti-PINK1 antibody. Red asterisks show phosphorylated PINK1. (b) PINK1^−/− MEFs co-expressing GFP-Parkin and various pathogenic PINK1 mutants were subjected to non phos-tag PAGE and immunoblotting with anti-Parkin and anti-PINK1 antibodies. Blue asterisks show autoubiquitylation of GFP-Parkin, which is taken as evidence of its mitochondrial localization. (c) The subcellular localization of GFP-Parkin in PINK1^−/− MEFs co-expressing various PINK1 mutants. Example figures indicative of colocalization and no colocalization are shown (bars, 10 μm). The number of cells with Parkin-positive mitochondria in the Parkin-expressing cells was counted in >100 cells. Error bars represent the mean±s.d. values of three experiments. Statistical significance was calculated using analysis of variance with a Tukey–Kramer post hoc test; *P<0.01; ns, not significant. (d) PINK1-GFP undergoes auto-phosphorylation following CCCP treatment. Red asterisks show the phosphorylated PINK1-GFP. (e) Various pathogenic PINK1 mutants were co-transfected with PINK1(WT)-GFP, treated with CCCP and subjected to immunoblotting using an anti-PINK1 antibody. Red asterisks show the phosphorylated bands. (f) HeLa cells expressing PINK1(G409V)-Flag with PINK1-GFP or PINK1(KD)-GFP were treated with CCCP and subjected to conventional PAGE. Red asterisks show phosphorylated PINK1(G409V)-Flag. FL, Δ1 and KD represent full-length PINK1, N-terminal processed PINK1 and KD mutant, respectively.

Figure 3. S402A mutation partly inhibits the autophosphorylation of PINK1.

(a) The S402A mutation altered the phosphorylation pattern of PINK1. PINK1 with the indicated mutations were transfected into HeLa cells and then subjected to SDS–PAGE on a 7.5% Tris-glycine gel ±50 μM phos-tag. Asterisks indicate phosphorylated PINK1 (see details of red and black asterisks in the text). (b) Ser/Thr residues in the activation loop of PINK1 from various organisms are shown in red font. Ser402 (boxed) is evolutionarily conserved.

Figure 4. S228 is another autophosphorylation site for PINK1.

(a) Mass spectrometric analysis of the in vivo autophosphorylation site of PINK1. PINK1-GST purified from cells +/−CCCP treatment was subjected to LC-MS/MS analysis with a phosphorylated peptide equivalent to amino acids 206–230 detected only from CCCP-treated cells. (b) The MS/MS data suggested that phosphorylation occurs at Ser228, Ser229 or Ser230. (c) Multiple sequence alignment of PINK1 residues neighboring Ser228 from various organisms. Ser228 (boxed) has been evolutionarily conserved across most species. (d) Structural model (Protein Model Database ID: PM0076345) of PINK1 revealed that the possible phosphorylation sites including S228 and S402 are spatially close to one another. Ser residues are shown in yellow and hydroxyl groups are highlighted in red. (e) The S228A mutation changed the phosphorylation pattern of PINK1. The first band (shown by a red asterisk) in phos-tag PAGE is not observed in cell expressing the S228A mutation. (f) Autophosphorylation-derived signals of PINK1 in cells expressing the S228A/S402A double mutation are completely abolished. Asterisks indicate the phosphorylated form.

Figure 5. PINK1 autophosphorylation of S228/S402 is essential for mitochondrial localization of Parkin.

(a) HeLa cells were co-transfected with GFP-Parkin and CMV or CMV(d1) promoter-driven PINK1. The number of cells with Parkin-positive mitochondria was counted in >100 cells at 24 h post transfection. Error bars represent the mean±s.d. values of three experiments. Statistical significance was calculated using analysis of variance with a Tukey–Kramer post hoc test. (b) Immunoblotting with an anti-PINK1 antibody to measure the quantity of PINK1 (marked by asterisks) expressed by the CMV or CMV(d1) promoter, and autoubiquitylation activity of GFP-Parkin when the indicated PINK1 plasmids were co-transfected. The slower-migrating bands were derived from ubiquitylation (Ub). (c) PINK1^−/− MEFs co-expressing GFP-Parkin and CMV(d1) promoter-driven PINK1 harboring the S228A/S402A or S228D/S402D mutation were subjected to immunoblotting with an anti-Parkin antibody. Ub shows autoubiquitylation of GFP-Parkin, which is taken as an indicator of its mitochondrial localization. (d) Both the S228A/S402A and S228D/S402D mutations hindered autophosphorylation of PINK1 in phos-tag PAGE. The asterisk indicates the phosphorylated form. FL and Δ1 mean full-length and N-terminal processed PINK1, respectively. (e) Subcellular localization of indicated PINK1 mutants in HeLa cells following CCCP treatment. Immunocytochemistry confirmed that both of the S228A/S402A and S228D/S402D PINK1 mutants localized on mitochondria following CCCP treatment. Bars, 10 μm. (f) Subcellular localization of GFP-Parkin in PINK1^−/− MEFs co-expressing CMV(d1) promoter-driven PINK1 harboring the S228A/S402A or S228D/S402D mutation. Immunocytochemistry showing that the S228A/S402A PINK1 mutant disturbed the mitochondrial localization of Parkin, whereas the S228D/S402D mutant recruited Parkin to the mitochondria equivalent to WT PINK1. Bars, 10 μm. (g) The number of cells with Parkin-positive mitochondria was counted in >100 cells following CCCP treatment. Statistical analysis was performed as in (a).

Figure 6. A model for Parkin recruitment to the damaged mitochondria.

The integrity of the mitochondria is sensed and transduced to Parkin by two sequential processes, that is the escape from membrane potential-dependent degradation and autophosphorylation of PINK1 at Ser228/Ser402. Both steps are essential for optimized Parkin recruitment and activation.