Activation of endogenous PRKN by structural derepression is linked to increased turnover of the E3 ubiquitin ligase

Abstract

Loss-of-function mutations in the PINK1 and PRKN genes are the most common cause of early-onset Parkinson disease (PD). The encoded enzymatic pair selectively identifies, labels, and targets damaged mitochondria for degradation via the macroautophagy/autophagy-lysosome system (mitophagy). This pathway is cytoprotective and efforts to activate mitophagy are pursued as therapeutic avenues to combat PD and other neurodegenerative disorders. When mitochondria are damaged, the ubiquitin kinase PINK1 accumulates and recruits PRKN from the cytosol to activate the E3 ubiquitin ligase from its auto-inhibited conformation. We have previously designed several mutations that effectively derepress the structure of PRKN and activate its enzymatic functions in vitro. However, it remained unclear how these PRKN-activating mutations would perform endogenously in cultured neurons or in vivo in the brain. Here, we gene-edited neural progenitor cells and induced pluripotent stem cells to express PRKN-activating mutations in dopaminergic cultures. All tested PRKN-activating mutations indeed enhanced the enzymatic activity of PRKN in the absence of exogenous stress, but their hyperactivity was linked to their own PINK1-dependent degradation. Strikingly, in vivo in a mouse model expressing an equivalent activating mutation, we find the same relationship between PRKN enzymatic activity and protein stability. We conclude that PRKN degradation is the consequence of its structural derepression and enzymatic activation, thus resulting only in a temporary gain of activity. Our findings imply that pharmacological activation of endogenous PRKN will lead to increased turnover and suggest that additional considerations might be necessary to achieve sustained E3 ubiquitin ligase activity for disease treatment.Abbreviations: BSA: bovine serum album, CCCP: carbonyl cyanide 3-chlorophenylhydrazone; ECL: electrochemiluminescence; EGF: epidermal growth factor; ELISA: enzyme-linked immunosorbent assay; FGF: fibroblast growth factor; iPSC: induced pluripotent stem cell; KI: knock-in; KO: knockout; MAP2: microtubule associated protein 2; MFN2: mitofusin 2; MSD: Meso Scale Discovery; mt-Keima: mitochondrial targeted Keima; NPC: neural progenitor cell; PD: Parkinson disease; PDH: pyruvate dehydrogenase; p-S65-PRKN: Serine 65 phosphorylated PRKN; p-S65-Ub: Serine 65 phosphorylated ubiquitin; REP: repressor element of PRKN; TH: tyrosine hydroxylase; TX: Triton X-100, Ub: ubiquitin; UBL: ubiquitin-like; WT: wild-type.

Keywords: Autophagy; PINK1; Parkin; Parkinson’s disease; mitophagy.

Figure 1.

PRKN structure and mechanisms of auto-inhibition and release. (A) Schematic of the human PRKN protein with color-coded domains in which domain interactions are shown as arrows that repress (top) or activate (bottom) PRKN. Briefly, the ubiquitin-like (UBL) and the repressor element (REP) of PRKN block the RING1 domain from E2 enzyme binding, indicated as A and B, respectively. The RING0 domain represses the RING2 domain, inhibiting the E3 Ub ligase activity of PRKN, indicated as C. PRKN-activating mutations are shown on top with green arrows indicating release of the respective auto-inhibitory interactions. The PINK1 phosphorylation site (S65), the p-S65-Ub binding site (H302), and the active site (C431) of PRKN are indicated at the bottom. The sequential activation of PRKN is numbered as follows: (1) p-S65-Ub binds to PRKN at H302 and promotes its translocation to mitochondria; (2) PRKN itself is phosphorylated by PINK1 at S65; (3) an Ub-charged E2 enzyme can bind to PRKN; and (4) transfer Ub onto the active site C431. (B) Shown is a graphical summary of the gene-edited models used throughout the study and PRKN measurements performed. Inactive PRKN crystal structure (PDB: 5C1Z) [26] and active PRKN AlphaFold model (ModelArchive: ma-1fhux, https://www.modelarchive.org/doi/10.5452/ma-1fhux) [32] are shown with the same color legend as in (A).

Figure 2.

Differentiated ReNcell VM neurons carrying PRKN-activating mutations have significantly lower PRKN protein levels and reduced p-S65-Ub induction upon mitochondrial damage. Midbrain-derived, isogenic differentiated ReNcell VM neurons expressing either of the four different PRKN-activating mutations or cells with PRKN KO were seeded alongside controls, differentiated to neurons, and treated with 20 µM CCCP for the indicated times. Representative western blots (A) and densitometric quantification (B) show similar PINK1 levels but lower total PRKN, p-S65-PRKN and p-S65-Ub levels in the clones with PRKN-activating mutations compared to controls. Shown is the result of experiments with three isogenic clones per mutation (two-way ANOVA with Tukey’s post hoc test). Asterisks denote a statistically significant difference compared to WT at the same time point. Representative western blots (C) and densitometric quantification (D) of differentiated ReNcell VM neurons from control and from PRKN KO cells show that p-S65-Ub induction is lower in PRKN KO cells compared to controls, while PINK1 levels are similar. Statistical analysis was performed by two-way ANOVA with Sidak’s post-hoc test. (E) neurons expressing either of the PRKN-activating mutations, show a more pronounced decline of PRKN protein upon 20 µM CCCP treatment compared to controls. PRKN western blot levels from untreated cells of each genotype were set to one at 0 h. Asterisks denote a statistically significant difference compared to the 0 h time point for of the same genotype (two-way ANOVA followed by Tukey’s post-hoc test). (F) PRKN mRNA levels of neurons with PRKN-activating mutations are unchanged compared to controls. Three independent experiments with the same isogenic cell clone were performed and analyzed with unpaired, two-sided student’s t-test. Data is shown as mean -/+ SEM for all graphs with *p < 0.05, **p < 0.005, ***p < 0.0005.

Figure 3.

iPS-derived dopaminergic cultures with PRKN-activating mutations confirm strongly reduced PRKN levels. Representative immunohistochemistry images (A) and image quantification (B) of two-week-old dopaminergic cultures from the different cell lines stained for the dopamine neuron marker tyrosine hydroxylase (TH, red), the microtubule associated protein (MAP2, green) and counterstained with a nuclear dye (Hoechst, blue). Data is shown as mean -/+ SEM from three technical replicates from one experiment. Scale bar: 50 µm. (C) Relative expression of the PRKN mRNA in two weeks old mutant neurons compared to PRKN WT neurons. Data is shown as mean -/+ SEM from two technical replicates from at least one experiment. Data from mutant neurons was analyzed with one-way ANOVA with Dunnett’s multiple comparison test, compared to WT PRKN neurons. Data are shown as *p < 0.05, **p < 0.01, and ***p < 0.001. (D) Representative western blot of two-week-old dopaminergic cultures that were incubated with DMSO or 10 µM CCCP for 24 h. Western blots were probed with antibodies against PRKN (PRK8), TH, and the mitochondrial markers MFN2 and PDH. ACTB was used as loading control.

Figure 4.

PRKN-activating mutations change PRKN activity. (A-D) Isogenic differentiated ReNcell VM cultures expressing either of the four different PRKN-activating mutations were seeded alongside controls, differentiated, and treated with 20 µM CCCP for the indicated times. Three different clones were analyzed per mutation and used as biological replicates for p-S65-Ub measurements by MSD ELISA or western blot quantification as in Figure 2. (A) Differentiated cells with PRKN-activating mutation showed lower p-S65-Ub levels compared to PRKN WT controls upon 20 µM CCCP treatment. Asterisks denote a statistically significant difference compared to WT at the same time point (two-way ANOVA and Tukey’s post hoc test). (B) At baseline, cells with PRKN-activating mutation show a tendency toward higher p-S65-Ub levels. Some mutations show a significant difference compared to WT PRKN controls (unpaired, two-sided t-test). (C) Upon normalizing p-S65-Ub levels per PRKN protein, the ratio of p-S65-Ub per PRKN increased for the mutations. Some mutations show significantly increased p-S65-Ub to PRKN ratio compared to WT PRKN controls, which is set to 1 (unpaired, two-sided t-test). (D) Upon normalizing p-S65-PRKN per PRKN protein at 4 h 20 µM CCCP treatment, neurons carrying PRKN-activating mutations show a higher p-S65-PRKN to PRKN ratio compared to WT controls (one-way ANOVA with Dunnett’s post hos test). (E) Schematic showing a modified E2 enzyme that binds to PRKN forming a PRKN-Ub-E2 enzyme complex that measures PRKN transthiolation. (F-G) Differentiated ReNcell VM clones were treated with either DMSO or 20 µM CCCP for 2 h. Lysates were adjusted and similar PRKN amounts were incubated with or without an E2~ABP to measure PRKN transthiolation activity. (F) Representative western blot analysis with PRKN (5C3) antibody. Short (bottom) and long (top) exposures are shown. A proportion of the unmodified, inactive PRKN (filled arrowhead), is shifted into a probe-bound configuration (open arrowhead), which represents transthiolation active PRKN. (G) Quantification of labeled to unmodified PRKN. Statistically significant comparisons (two-way ANOVA followed by Sidak’s post-hoc test) are indicated. (H-I) Cells were differentiated from clones stably expressing mt-Keima and ratiometrically analyzed using flow cytometry. (H) In the absence of CCCP, the ratio of acidic to neutral mt-Keima was similar across different genotypes. (I) Upon treatment with 20 µM CCCP the ratio of acidic/neutral mt-Keima increased in all lines but not as much as in control cells. Shown is the Fold induction of the mt-Keima ratio as mean -/+ SEM of at least three experiments per genotype. Statistical analysis was performed with one-way ANOVA with Dunnett’s post-hoc test. Data is shown as mean -/+ SEM for all graphs with *p < 0.05, **p < 0.005, ***p < 0.0005.

Figure 5.

Activated PRKN protein levels are not substantially stabilized by proteasome or autophagy inhibition. (A) Isogenic differentiated ReNcell VM neurons carrying PRKN-activating mutations were seeded alongside controls, differentiated to neurons, and treated with 20 µM CCCP for the indicated times. Cells were analyzed for LC3B and SQSTM1/p62 as a readout for autophagy. Representative western blots show no overt changes in LC3 or SQSTM1/p62 between controls cells and cells with PRKN-activating mutations at baseline and upon CCCP treatment. (B, C) ReNcell VM from controls or isogenic lines with PRKN-activating mutations were differentiated to neurons and treated with either with epoxomicin to inhibit proteasomal degradation (B), or with bafilomycin A₁ (BafA1) to inhibit autophagy (C). Representative western blots show consistently lower PRKN levels in cells with PRKN-activating mutations, but no stabilization with either treatment. Treatment was successful as seen in increased levels of SQSTM1/p62 and CDKN1A/p21 upon bafilomycin A₁ and epoxomicin treatment, respectively. (D) Differentiated ReNcell VM neurons from control cells (WT) or cells with PRKN^Y143E or PRKN^W403A PRKN were used for sequential extraction. The majority of the PRKN signal was present in the 1% Triton X-00 soluble (TX) fraction, whereas the 2% SDS soluble fraction (S) only showed some PRKN signal. GAPDH was used as a loading control, and the soluble VCL protein to show the successful fractionation. (E,F) lysates from untreated differentiated ReNcell VM neurons from control cells (WT) or cells with PRKN^Y143E or PRKN^W403A were used for western blot with different PRKN antibodies (E) or for dot blots (F) with two different amounts of lysates. PRKN signal reduction was observed with several antibodies and with molecular mass independent method.

Figure 6.

PRKN protein levels are robustly stabilized upon PINK1 KO (and upon inactivation of PRKN). ReNcell VM with PRKN^Y143E or PRKN^W403A mutation as well as WT PRKN controls cells were gene-edited to knock out PINK1. Parental and PINK1 KO lines were seeded side-by-side, differentiated to neurons and treated with CCCP to show the presence or absence of PINK1 and p-S65-Ub signal. (A) Representative western blots showed increase of PRKN levels upon PINK1 KO in WT PRKN and to a greater extent to in cells with PRKN^Y143E or PRKN^W403A. (B) PRKN protein levels were quantified and expressed as ratio to parental lines. Shown in the mean -/+ SEM of three independent PINK1 KO clones. Asterisks on top of bars indicate significantly higher stabilization of PRKN upon PINK1 KO compared to WT cells (one-way ANOVA with Tukey’s post-hoc test [*p < 0.05, **p < 0.005, ***p < 0.0005]). (C) Scheme to indicate PRKN^H302A gene-editing in the ReNcell VM model. Depicted are from top to bottom: the genomic organization of the target site, the amino acid sequence, the genomic DNA sequence, the binding site of the guideRNA with PAM sequence, and the resulting DNA sequence with example Sanger sequencing histograms. (D) Introduction of PRKN^H302A mutation into WT PRKN or PRKN^W403A differentiated ReNcell VM neurons leads to PRKN stabilization. (E) PRKN protein levels were quantified and expressed as ratio to parental lines. Shown in the mean -/+ SEM of three technical repeat experiments. (F) Scheme to indicate PRKN^C431S gene-editing in the ReNcell VM model. Depicted are from top to bottom: the genomic organization of the target site, the amino acid sequence, the genomic DNA sequence, the binding site of the guideRNA with PAM sequence, and the resulting DNA sequence with example Sanger sequencing histograms. (G) Representative image of PRKN levels in differentiated ReNcell VM neurons with and without PRKN^C431S mutation.

Figure 7.

Prkn W402A mice recapitulate cell culture findings with decreased PRKN^W402A and p-S65-Ub and strong stabilization upon KO of Pink1. (A) Scheme to indicate how gene-edited PRKN^W402A KI mice were generated. Protein lysates from 6-month-old WT or mice with heterozygous (WT/KI) or homozygous (KI/KI) PRKN W402A knock-in were analyzed. (B) Representative western blots showed decreased and absent PRKN signal in brains from hetero- and homozygous PRKN W402A mice with PRK8 antibody and reduced signal with 5C3 antibody. (C) Densitometric quantification of PRKN western blot levels showed statistically reduced levels with both PRK8 and 5C3 antibodies. (D) p-S65-Ub levels as measured by MSD ELISA are unaltered between the genotypes. (E) p-S65-Ub MSD signal normalized to PRKN protein levels show a statistically significant increase for PRKN^W402A from WT/KI and KI/KI mice compared to WT PRKN. (B-D) shown is the mean -/+ SEM from brains of WT (n = 29), WT/KI (n = 14) and KI/KI (n = 20) animals. Statistical analysis was performed with one-way ANOVA followed by Tukey’s post-hoc test (*p < 0.05, **p < 0.005, ***p < 0.0005). (F) Representative western blots showed a robust PRKN protein level increase upon knockout of Pink1 (pink ^−/−). (G) Densitometric quantification of PRKN western blot levels showed statistically increased levels upon pink1 KO for both WT and PRKN^W402A with a similar Fold change, as indicated. Shown is the mean -/+ SEM from brains of WT (n = 31), WT with pink1 KO (n = 22), PRKN W402A KI/KI (n = 22) and PRKN W402A KI/KI with pink1 KO (n = 8). Statistical testing was performed with one-way ANOVA and Tukey’s post-hoc test (**p < 0.005, ***p < 0.0005).