The PINK1 p.I368N mutation affects protein stability and ubiquitin kinase activity

The Polish family with the PINK1 p.I368N mutation was recruited to participate in this research study from the movement disorders clinic of the Department of Neurology of the Medical University of Warsaw in Warsaw, Poland. The diagnosis of PD was made based on the UK Parkinson’s Disease Society Brain Bank clinical diagnostic criteria [27]. At the time of blood and skin specimens retrieval, the research subjects were independently clinically examined by two board certified neurologists (DK and ZKW) utilizing standardized narrative Mayo Clinic (MC) history, standardized MC narrative family history, and standardized MC neurological examination [28]. In addition the Hoehn and Yahr and Mini Mental State examination scales were obtained [29, 30]. The blood samples and 3 mm single skin specimens were retrieved in routine fashion by utilizing the standard universal precautions. Both subjects were right handed and parkinsonian features were more expressed on the dominant side. The response to the standard antiparkinsonian therapy including levodopa was excellent and there was no need to proceed with surgical interventions such as for example a deep brain stimulation.

Material for analysis of genomic DNA was extracted from blood and fibroblasts using a standard protocol. To confirm the genotype (c.T1103A/c.T1103A; p.I368N/p.I368N) and exclude other mutations in PINK1, all coding exons were directly sequenced by Sanger method and analyzed against the reference sequence NM_032409.2 (NCBI36/hg18). As far as we know this is the only family carrying this specific mutation in the Polish population. Though the analyzed siblings both carry a homozygous mutation, there was no consanguineous marriage of their parents. Gene rearrangements were analyzed by multiple ligation-dependent probe amplification (MLPA) technique using available tests SALSA P051 PD probes (MRC Holland) – probes for all PINK1 exons. Mutations in PINK1 and PARKIN were excluded in control fibroblast cells by sequencing of all coding exons.

The modeling of the full-length human PINK1 protein, NP_115785.1 (581 amino acid residues), has been described recently [26]. In brief, each individual domain was modeled as a separate unit and built into a composite full-length structure. The hybrid model is derived from consensus between the programs PRIME (Prime v3.0, Schrödinger, LLC, New York, NY) [31, 32], YASARA SSP/Homology/PSSM Method [31‐38], DISTILL (Porter) and TASSER [39‐44] and combines homology, threading, ab initio, and compositing techniques.

Mutations of amino acids were completed using the Maestro within the Schrödinger suite and parameterized using OPLS2.0 and Amber FF. Additionally, Maestro was used for placing mutated residues (or extending) automatically within an existing peptide chain. Also, MacroModel features within Maestro allow for the quick minimization of the structure for local geometry fixes to correct stereochemistry and packing of the amino acids. Further minimization was completed on the post-Schrödinger model using OPLS2005 within YASARA2, which has an AMBER set and can be used to further parameterize modifications to import into existing molecular dynamics integration engines, as the parameters for the modification are well documented for YASARA and Schrödinger [32, 37, 45‐47].

For each proposed mutation locus, we equilibrated the corresponding flexibility zone in the wild type complex and evaluated the free energy of unfolding (ΔG_wt) with the KB potential FoldX [48]. Combined with our own codes, we have a method to scan across trajectories and examine interaction zones of interest. We repeated the calculation for the mutant (p.I368N) complex to obtain the respective free energy of unfolding (ΔG_mutant). The difference between these two quantities is our estimate of the experimental change in binding free energy induced by the mutation [49]: ΔΔG = ΔG_wt- ΔG_mutant.

MDS was completed on each model for conformational sampling, using methods previously described [50‐53]. Each PINK1 system was minimized with relaxed restraints using Steepest Descent and Conjugate Gradient PR, and equilibrated in solvent with physiological salt conditions, as described in the literature [50‐53]. After equilibration was established, each system was allowed to run an additional MD production length of >250 nanoseconds. The primary purpose of MD for this study was conformational stability, refinement, and interaction calculations that may occur at the active site or dimer interface. The protocol for refinement include the following steps: (1) Minimization with explicit water molecules and ions, (2) Energy minimization of the entire system, and (3) MDS for >2 ns to relax to the force field (AMBER03). Following the refinement protocol, production simulations were completed to collect data.

AMBER03 force fields were used with the current release of NAnoscale Molecular Dynamics 2 engine [54, 55]. The PINK1 systems simulated, including hydrogens, consist of 1.74 × 10⁴ atoms prior to solvation with TIP3P water and ions. In all cases, we neutralized with counter-ions, and then created a solvent with 150 mM Na + Cl- to recreate physiological strength. TIP3P water molecules were added around the protein at a depth of 12–15 Å from the edge of the molecule depending upon the side [56]. Our protocol has been previously described in the literature [50, 51, 53, 57]. Solvated protein simulations consist of a box with 2.23–3.35 × 10⁵ atoms including nucleic acids, proteins, counter-ions, solvent ions, and solvent waters. Simulations were carried out using the particle mesh Ewald technique with repeating boundary conditions with a 9 Å nonbonded cut-off, using SHAKE with a 2-fs timestep. Pre-equilibration was started with three stages of minimization with 10,000 steps of SD, PRCG, relaxing restraints, then followed by 1000 ps of heating under MD, with the atomic positions of nucleic and protein fixed. Then, two cycles of minimization (5000 steps each) and heating (1000–5000 ps) were carried out with soft restraints of 10 and 5 kcal/(mol · Ų) applied to all nucleic and protein backbone atoms. Next, 5000-steps of minimization were performed with solute restraints reduced to 1 kcal/(mol · Ų). Following that, 2000–4000 ps of MDS were completed using relaxing restraints (1 kcal/(mol · Ų)) until all atoms are unrestrained, while the system was slowly heated from 1 to 310 K using velocity rescaling upon reaching the desired 310 K during this equilibration phase. Additionally, NPT equilibration based MD was used with velocity rescaling for 1–2 ns. Finally, production runs of MD were carried out with constant pressure boundary conditions (relaxation time of 1.0 ps). A constant temperature of 310 K was maintained using the Berendsen weak-coupling algorithm with a time constant of 1.0 ps. SHAKE constraints were applied to all hydrogens to eliminate X-H vibrations, which yielded a longer simulation time step (2 fs). Our methods for equilibration and production run protocols are in the literature [50, 51, 53, 57]. Equilibration was determined from a global flattening of RMSD over time after an interval of 1–3 ns or similarly with total energy. Translational and rotational center-of-mass motions were initially removed. Periodically, simulations were interrupted to have the center-of-mass removed again by a subtraction of velocities to account for the “flying ice-cube” effect [58]. Following the simulation, the individual frames were superposed back to the origin, to remove rotation and translation effects. Archived snapshots were recorded every 2 ps, yielding >2000 snapshots per simulation.

Using mixture of Piper protein-protein docking for generating 40,000 conformers of random seed orientation over the PINK1 and ubiquitin proteins, we applied restriction parameters as follows: only kinase domains from PINK1 (N-lobe and C-lobe) and Ser65 face of Ub were weighted with increased attraction parameters +0.25, whilst we also did multiple LC-MOD/MC generated side chain re-orientations during the final combination with an simulated annealing protocol to allow for settling of interactions before applying a PRCG minimization protocol. We examined the docking interfaces with brief bursts of simulated annealing and MDS for further refinement for taking the top seeded structures and elimination of the poor scorers. Top surviving poses were scored and best from each series was retained [50, 52, 53, 59, 60]. The best pairs generated were refined and studied using MDS calculations.

Primary human dermal fibroblasts were grown in Dulbecco’s modified Eagle medium (DMEM, Invitrogen) supplemented with 10% fetal bovine serum (FBS, BioWest), 1% PenStrep and 1% non-essential amino acids (both Invitrogen). Control fibroblasts were received from Cell Applications Inc., mutations in PINK1 and PARKIN genes were excluded by sequencing and PINK1-dependent responses in those cells have been characterized before [18, 25, 26]. HeLa cells (ATCC) were cultured in DMEM supplemented with 10% FBS. Stable HeLa cell clones expressing EGFP-Parkin or 3xFLAG-Parkin C431S have been described before [61]. All cells were maintained at 37 °C, 5% CO₂ in a humidified atmosphere.

Carbonyl cyanide m-chlorophenyl hydrazone (CCCP), epoxomicin and cycloheximide were purchased from Sigma-Aldrich, valinomycin from Axxora. siRNA transfections were performed with 20nM control (all stars negative control) or PINK1-specific siRNA (5′ GACGCTGTTCCTCGTTATGAA-3′) using HiPerfect (all from Qiagen) or Lipofectamine2000 (Invitrogen). siRNA resistant PINK1-V5 constructs have been described before [5, 24]. The mutations p.F104A, p.L347P and p.I368N were introduced by site-directed mutagenesis and verified by Sanger sequencing. The standard transfection protocol for DNA was as follows: for one well of a 12-well plate 1 μg of DNA and 2.5 μl of Lipofectamine2000 (Invitrogen) were each mixed with 100 μl of Opti-MEM media (Invitrogen), incubated for 5 min at room temperature, mixed together and incubated 20 min before adding to the cells. For low-level expression DNA and lipofectamine amounts were reduced by 50–75% (total DNA per 12 well: 0.5 or 0.25 μg).

Total mRNA from fibroblasts was prepared using the RNeasy mini kit (Qiagen) according to the manufacturer’s protocol and then reverse-transcribed into cDNA using a High Fidelity cDNA kit (Roche Applied Science). cDNA was amplified by real-time PCR on a LightCycler 480 system (Roche) using Taqman probes (BioRad PINK1 Id qHsaCEP0053094 and RPL27 Id qHsaCEP0051648). Levels were determined using the second derivative method.

The following primary antibodies were used for Western blot (WB) and/or immunofluorescence (IF): rabbit anti-CDC37 (WB, 1:1,000, 4793, Cell Signaling), mouse anti-HSP90 (WB, 1:2,000, 610419, BD), mouse anti-FLAG (WB, 1:150,000, F3165, Sigma Aldrich), mouse anti-GAPDH (WB, 1:150,000, H86504M, Meridian Life science), mouse anti-Mitofusin 1 (WB, 1:5,000, ab57602, Abcam), rabbit anti-p38 MAPK (WB, 1:1,000, #9212, Cell Signaling), rabbit anti-PINK1 (WB, 1:2,000, BC100-494, Novus Biologicals), rabbit anti-PINK1 (WB, 1:2,000, IF, 1:100, #6946, Cell Signaling), mouse anti-Parkin (WB, 1:3,000, #4211, Prk8, Cell Signaling), mouse anti-TOM20 (IF, 1:250, sc-17764, Santa Cruz Biotechnology), rabbit anti-TOM20 (WB, 1:20,000, 11802-1-AP, ProteinTech Group), rabbit anti-TOM40 (WB, 1:1,000, 18409-1-AP, ProteinTech Group), mouse anti-Ub (WB, 1:1,000, MAB1510, ubi-1, Millipore), rabbit anti-p-Ser65-Ub (WB, 1:5,000-1:15,000, IF, 1:250, in house [18]), mouse anti-V5 (WB, 1:5,000, R960-25, Invitrogen), rabbit anti-V5 (WB, 1:5,000, ab9115, Abcam), mouse anti-vinculin (WB, 1:100,000-500,000, V9131, Sigma). Streptavidin HRP (WB, 1:150,0000, Pierce).

Cells were collected and lysed in RIPA buffer (50 mM Tris pH 8.0, 150 mM NaCl, 1% NP-40, 0.5% deoxycholate, 0.1% SDS) plus Complete protease and PhosSTOP phosphatase inhibitors (Roche Applied Science) on ice for 30 min. Proteins were collected as supernatants after centrifugation at 20,000 g for 15 min at 4 °C. SDS–PAGE was performed using 8–16% or 16% Tris Glycine gels (Invitrogen). Proteins were transferred onto PVDF membranes and detected using standard immunoblotting procedures. To detect phosphorylated PINK1, 8% Tris Glycine gels containing 50 μM Phos-tag acrylamide (Wako chemicals) and 100 μM ZnCl₂ were used [18]. After electrophoresis, Phos-tag acrylamide gels were washed using transfer buffer with 0.01% SDS and 1 mM EDTA for 20 min and then with transfer buffer containing 0.01% SDS without EDTA for 10 min to remove excess Zn²⁺ before transfer.

Cells were harvested in fractionation buffer containing 20 mM HEPES-KOH (pH 7.4), 1 mM EDTA, 250 mM sucrose and cocktail of protease and phosphatase inhibitors (Roche Applied Science). After homogenization with a 22 G syringe (20 strokes), nuclei and cell debris were removed by centrifugation at 1500 g for 5 min. The post-nuclear supernatant (PNS) was transferred into a new tube and centrifuged at 12,000 g for 10 min. The mitochondria-enriched pellet (‘mitochondrial fraction’) was dissolved in RIPA buffer. Cytoplasmic fractions were isolated as supernatant after ultracentrifugation at 100,000 g for 30 min. For solubility studies, the PNS was dissolved in RIPA buffer and pelleted at 14,000 g for 15 min. The supernatant was used as the “soluble” fraction, the pellet was further washed and solubilized in 8 M urea and used as the “insoluble” fraction.

Transfected HeLa 3xFLAG-Parkin C431S cells [61] were harvested in preheated (95 °C) SDS lysis buffer (50 mM Tris pH 7.6, 150 mM NaCl, 1% SDS). Lysates were homogenized by 10 strokes through a 20G needle. To verify the band shift by oxyester formation, aliquots of lysates were treated with or without NaOH (final concentration: 100 mM) for 1 h at 37 °C before loading onto SDS gels.

Fibroblasts were seeded onto PDL (P0899, Sigma-Aldrich) coated glass coverslips treated with valinomycin, CCCP or left untreated. Cells were then fixed with 4% (w/v) paraformaldehyde, permeabilized with 1% Triton-X-100, blocked with 10% goat serum and incubated with primary antibodies followed by secondary antibodies (anti-rabbit Alexa Fluor-488, anti-mouse IgG Alexa Fluor-568, 1:1000, Invitrogen). Nuclei were stained with Hoechst 33342 (1:5000, Invitrogen). For staining of PINK1 tyramide signal amplification (T20922, Invitrogen) was used. Coverslips were mounted onto microscope slides using fluorescence mounting medium (Dako). High resolution confocal images were taken with an AxioObserver microscope equipped with an ApoTome Imaging System (Zeiss).

To quantify p-Ser65-Ub levels in valinomycin treated fibroblasts and EGFP-PARKIN translocation in CCCP treated HeLa cells, we used the BD Pathway 855 HCI platform (BD Biosciences) as described earlier [18, 26, 61]. To collect signal only from cells transfected with siRNA and various PINK1-V5 constructs, co-expressed mCherry fluorescent intensity was used as a restriction parameter for the analysis of PARKIN translocation.

HeLa cells were transiently transfected with V5-6xHis-tagged PINK1 WT, p.I368N or p.L347P for 48 h and were then treated with 15 μM CCCP for 2 h or left untreated. Cells were lysed in co-IP buffer (50 mM HEPES pH 7.5, 150 mM NaCl, 10 mM KCl, 1.5 mM MgCl₂, 1 mM EDTA, 0.5 mM EGTA, 10% glycerol and 0.2% NP-40) supplemented with protease and phosphatase inhibitors (Roche Applied Science) and briefly sonicated. For PINK1-V5 immunoprecipitation (IP), total cell lysate was incubated overnight at 4 °C with mouse anti-V5-agarose beads (A7345, Sigma-Aldrich). Formed immuno-complexes were spun down, washed 3x with co-IP buffer and eluted from beads with 50 μl 2x SDS sample buffer and boiled at 95 °C. 10 μg of input cell lysates and 10 μl of immunoprecipitates were analyzed by Western blot.

Immunoprecipitated WT and p.I368N PINK1-V5 from CCCP treated and untreated HeLa cells were washed twice with co-IP buffer and once with in vitro phosphorylation buffer (20 mM HEPES pH 7.4, 50 mM NaCl, 5 mM MgCl₂, 0.1 mM EGTA, 0.01% Triton X-100). After complete removal of the buffer 100 μl of the in vitro reaction mixture was added. The reaction contained 1 μg of N-terminally biotinylated Ub (Boston Biochem), 1 mM TCEP (Gold Bio) and 2 mM ATP trisodium salt (Sigma-Aldrich) in phosphorylation buffer. Reactions were carried out at 37 °C under constant agitation for 24 h, stopped by addition of 6x SDS sample buffer, heated to 56 °C for 15 min and analyzed by immunoblotting using p-Ser65-Ub antibody. Detection of biotinylated Ub with streptavidin-HRP (Pierce) served as loading control.

Densitometric analysis of Western blots was performed with the Image Studio software (LiCor). The half-life of PINK1 proteins was calculated by non-linear one phase decay curve fitting using GraphPad Prism version 6.

All quantitative results are expressed as mean ± SEM from at least three independent experiments. Statistical comparisons were performed using one-way and two-way ANOVA, respectively, with Tukey’s post-hoc test (*, p < 0.05; **, p < 0.005; ***, p < 0.0005). All analyses were performed using GraphPad Prism version 6.

Here we analyzed two patients that are homozygous carriers of a recently identified PINK1 c.T1103A (p.I368N) mutation [62]. Detailed clinical data of the affected members are provided in Additional file 1: Table S1. Both patients have early-onset and slowly progressive PD that started at relatively young age (28 and 33 years, respectively), consistent with the typical clinical presentation of homozygous PINK1 loss-of-function [63]. Some clinical features (e.g. laterality of parkinsonism, resting tremor, bradykinesia and good response to levodopa) are indistinguishable from sporadic late-onset PD. Yet, REM sleep behavior disorder (RBD), which is a clinical biomarker of synucleinopathies, was not detected in these patients [64]. Both patients show mild autonomic failure (e.g. constipation), but no dementia, and no depression or anxiety, which are occasionally recognized as a non-motor symptom of PINK1-associated PD [65].

The residue Ile368 is located at the +1 position of the catalytic loop of the consensus motif HRDLKxxN in Ser/Thr kinases (Fig. 1a). Using homology modeling, we have recently generated a structural model of full-length PINK1 at an all-atom resolution [26]. To examine how the p.I368N mutation would affect protein structure we performed molecular modeling and free, unbiased molecular dynamics simulations (MDS) (Additional file 2: Movie S1 and Additional file 3: Movie S2 for PINK1 WT and p.I368N mutant, respectively). The calculated Root Mean Square Fluctuations (RMSF) provide the movement of each individual residue of PINK1 (581 amino acids) as a time-averaged sum over the course of the simulation. Though global conformational changes were not observed, overall RMSF of PINK1 p.I368N were slightly lower than WT, yet with few higher peaks around residues of the ATP-binding pocket (Fig. 1b). Overlay of the WT and mutant kinase domains together with Root Mean Square Deviations (RMSD), as a measure of structural change, revealed an immediate but minor difference before and after MDS (Fig. 1c, d). However, specifically the orientation of helices surrounding the ATP-binding pocket was altered in the mutant compared to PINK1 WT (Fig. 1e) and resulted in local changes of RMSD within the ATP-binding site (Fig. 1f). In addition, surface analyses revealed a distorted ATP pocket shape for p.I368N (Fig. 1g). Together with FoldX measurements of Gibb’s free energy that indicated decreased local stability (Additional file 4: Table S2) and an increase in the radius of gyration (Fig. 1h) this indicated a negative effect on the catalytic activity of PINK1.

To study the enzymatic activity, we had recently docked the substrate Ub into the kinase domain of PINK1 WT [26]. We could show that auto-phosphorylation of PINK1 at Ser228 and Ser402 increased interactions with Ub that not only stabilized the kinase/substrate complex but also positively affected atomic distances for a more efficient phosphate transfer. Here, we docked Ub to a p.I368N homodimer to investigate binding and alignment of the substrate within the active site of the kinase (Fig. 2a). Indeed, closer inspection of the mutant enzymatic complex, revealed a slight mispositioning of the Ub substrate (Fig. 2b). However, distances measured between key atoms of PINK1, ATP (terminal phosphate), and Ub (oxygen of Ser65) were only slightly altered, and still within the range for effective catalysis. In summary, structural analyses suggested a kinase defect for the PINK1 p.I368N mutation would likely be caused by impaired ATP stabilization in the active site and potentially altered binding of the substrate Ub.

To study the effects of endogenous mutant PINK1 in cells, we analyzed primary skin fibroblasts from controls and both patients carrying the homozygous p.I368N mutation. We quantified PINK1 mRNA levels before and after mitochondrial depolarization with the K⁺ionophore valinomycin by quantitative reverse transcriptase PCR (qRT-PCR), but found no significant difference between WT and mutant PINK1 fibroblasts (Fig. 3a). Next, we examined PINK1 protein levels under basal conditions and upon mitochondrial damage (Fig. 3b). In the absence of stress, PINK1 protein was hardly detectable by Western blot (WB) under endogenous conditions. While full-length PINK1 WT robustly accumulated in controls upon treatment with the mitochondrial uncoupler CCCP, stabilization of p.I368N mutant PINK1 protein was significantly impaired (Fig. 3b, c). To confirm these findings, fibroblasts were subjected to immunofluorescence (IF) with antibodies against PINK1 and the mitochondrial marker protein TOM20. Consistently, PINK1 signal was considerably weaker in p.I368N mutant cells compared to WT controls (Additional file 5: Figure S1A). Despite significantly reduced protein levels, localization of the mutant PINK1 p.I368N to damaged mitochondria seemed unaltered. This was further corroborated by subcellular fractionations, where both WT and mutant full-length PINK1 were detected in the mitochondrial fraction upon depolarization (Additional file 5: Figure S1B). Taken together, p.I368N mutant cells showed no changes in mRNAs levels, but significantly reduced levels of the full-length PINK1 protein on mitochondria upon depolarization.

To determine the enzymatic activity of the PINK1 p.I368N mutation under endogenous conditions we used recently developed antibodies specific to the PINK1-phosphorylated form of Ub (p-Ser65-Ub) [18, 19, 26]. While Ub was strongly and time-dependently phosphorylated upon CCCP treatment in control cells, no p-Ser65-Ub signal was detected by WB in cells from patient’s carrying the PINK1 p.I368N mutation (Fig. 3b). The same results although with an overall faster stress response were obtained upon depolarization of mitochondria with valinomycin (Additional file 5: Figure S1C). To further confirm this by an independent method, fibroblasts were treated with valinomycin and analyzed by IF (Fig. 3d). In line, PINK1 mutant cells showed no induction of specific p-Ser65-Ub signal above background, which was quantified by high content imaging (HCI) (Fig. 3e and Additional file 5: Figure S1D). Consistent with the lack of p-Ser65-Ub signal in patients’ cells, PARKIN activation/modification and Mitofusin1 (MFN1) ubiquitylation were not observed (see Fig. 3b). Overall, these results show that the PINK1 p.I368N mutation reduces protein stabilization upon stress resulting in failure to activate PARKIN and mitochondrial quality control.

In order to reveal whether differences in PINK1 protein levels are caused by aberrant processing or protein instability, we co-treated cells with a mitochondrial uncoupler and a proteasome inhibitor to stabilize the cleaved forms of PINK1. WB analysis showed that although full-length PINK1 protein (~63 kDa) was significantly reduced in p.I368N mutant cells upon mitochondrial depolarization, the PARL-cleaved form (~52 kDa) was stabilized upon epoxomicin treatment to the same extent as PINK1 WT in the control cells (Fig. 4a, b). Full-length PINK1 and in particularly the PARL-cleaved form has been reported before as an insoluble protein [7, 66, 67]. Thus, we separately analyzed soluble and insoluble fractions of control and p.I368N mutant fibroblasts that had been treated with a combination of valinomycin and epoxomicin. Full-length WT PINK1 was mostly found in the soluble fraction, while its cleaved form strongly accumulated in the insoluble fraction upon proteasome inhibition (Fig. 4c). While the PARL-cleaved form of p.I368N accumulated in the insoluble fraction similar to PINK1 WT, full-length mutant PINK1 remained barely detectable in either fraction. Yet, we could not identify any distinct, additional cleavage product for p.I368N that would indicate aberrant processing of mutant PINK1. Accordingly, several computational prediction tools failed to report de novo protease cleavage sites for PINK1 p.I368N (data not shown).

To examine whether full-length PINK1 p.I368N is indeed an inherently unstable protein, we compared WT and mutant levels upon overexpression in HeLa cells. In contrast to endogenous PINK1 in fibroblasts, expression levels of full-length and cleaved p.I368N mutant were surprisingly similar to PINK1 WT upon standard transfection (1 μg DNA per well of a 12-well plate), regardless of mitochondrial depolarization (Additional file 6: Figure S2A). However, transfection of only half of the DNA amounts (0.5 μg versus 1 μg), resulted in decreased levels of PINK1 p.I368N compared to WT protein comparable to the endogenous condition in patients’ fibroblasts. In line with previous studies that had raised caution, our data show that high-level PINK1 overexpression can override important regulatory mechanisms and may mask defects of PINK1 mutations [26].

Although the exact reasons are unclear, epoxomicin treatment negatively affected stabilization of full-length PINK1 on damaged mitochondria in line with the literature [68]. To prevent any confounding effect of proteasome inhibition, we made use of an engineered mutation that specifically blocks degradation of cleaved PINK1. It is known that PINK1, once imported through the IMM, is cleaved by PARL in its transmembrane region between Ala103 and Phe104, resulting in its degradation by the N-end rule pathway [7]. The F104A mutation is able to block degradation of the PARL- cleaved form of PINK1 independently of the proteasome activity. Upon knockdown of endogenous PINK1 in HeLa cells, we expressed PINK1 siRNA-resistant variants with single or double p.I386N and p.F104A mutations at low levels (Fig. 4d). Under these conditions, full-length PINK1 p.I368N was less stable than PINK1 WT, but similar levels were reached for the p.F104A mutant. Cleaved PINK1 was barely detectable for WT or p.I368N, but strongly accumulated in the single and double mutants harboring the p.F104A mutation. Strikingly, also full-length PINK1 p.I368N was significantly stabilized upon CCCP treatment by the presence of the p.F104A mutation (Fig. 4e). Altogether these results show that while generation, processing and stability of PINK1 p.I368N are not impaired under basal conditions, specifically the mutant full-length form does not properly accumulate on the OMM upon mitochondrial stress.

To better characterize the instability of PINK1 p.I368N over time, we performed experiments in the presence of a protein synthesis inhibitor. PINK1 siRNA knockdown HeLa cells were transfected with only 25% of the recommended DNA amounts (0.25 μg) to ensure low expression of PINK1-V5 close to endogenous levels. Cells were left untreated or treated with cycloheximide (CHX) and the decay of full-length PINK1 protein was monitored in the absence or presence of CCCP, respectively (Fig. 5a and b). Under basal conditions, without mitochondrial damage, the half-life of PINK1-V5 WT was approximately 29 min, while PINK1 p.I368N was stable for only about 7 min (Fig. 5a). Mitochondrial damage dramatically increased the half-life of PINK1-V5 WT (>4 h) and slightly increased also the stability of the p.I368N mutant to 23 min (Fig. 5b). The calculated half-life for PINK1 WT was in good agreement with a previous report [69]. Together this confirms that PINK1 p.I368N is normally expressed, similar to PINK1 WT, but is destabilized on the protein level, in particular during mitochondrial stress.

To elucidate the reasons for the destabilization of the p.I368N mutation, we investigated key protein interactions of PINK1 by co-immunoprecipitation (Fig. 5c and d). We focused on the HSP90/CDC37 chaperone complex that supports folding of several kinases in the cytosol including PINK1 [70]. The PD-associated PINK1 mutation p.L347P had been reported before as unstable due to its of lower binding affinity for HSP90/CDC37 [66, 71] and thus was included as a control. While PINK1-V5 WT co-immunoprecipitated substantial amounts of the two chaperones with and without CCCP treatment, both proteins were significantly reduced in immunoprecipitates of PINK1-V5 p.I368N and p.L347P, respectively. In addition, we examined the interaction of PINK1 with the integral OMM protein TOM40 that forms the channel of the protein import complex [72]. It is known that full-length PINK1 accumulates into a higher molecular weight protein complex with the import machinery [10, 73], which regulates its insertion into the OMM, stability and enzymatic function towards PARKIN activation. In line with this, PINK1-V5 WT was found to interact with TOM40 specifically upon CCCP treatment, while the PINK1 mutations p.I368N or p.L347P failed, despite comparable amounts of immunoprecipitated PINK1-V5. Together, these data suggest that the stability of the PINK1 p.I368N mutation is altered through reduced interactions with stabilizing chaperone and the mitochondrial protein import complex.

To investigate kinase activity of the PINK1 mutation independent of protein instability, HeLa cells were transfected with WT or p.I368N mutant using standard methods (1 μg DNA). At equal expression levels, only PINK1 WT was auto-phosphorylated upon mitochondrial depolarization as shown by separation and detection on phos-tag gel (Fig. 6a). Accordingly, p-Ser65-Ub levels were only increased in HeLa cells transfected with PINK1 WT, but not in cells expressing the p.I368N mutant or the empty vector, also when endogenous PINK1 WT was silenced (Additional file 6: Figure S2B). To directly confirm lack of kinase activity of PINK1 p.I368N, we performed in vitro kinase assays with recombinant Ub. PINK1-V5 WT and p.I368N were immunoprecipitated from overexpressing cells that had been treated with or without CCCP and samples were incubated with biotinylated mono-Ub as a substrate. Phosphorylated Ub was detected exclusively in presence of PINK1 WT obtained from CCCP-treated cells (Fig. 6b).

To confirm a biological downstream effect, we analyzed translocation of GFP-PARKIN to damaged mitochondria in stable HeLa cells by HCI [61]. As expected, siRNA-mediated knockdown of endogenous PINK1 fully abrogated PARKIN translocation to damaged mitochondria (Fig. 6c), while exogenous expression of PINK1 WT restored it but the PINK1 p.I368N mutant failed. To biochemically verify failure to activate PARKIN, PINK1 WT or p.I368N mutant were transiently transfected into HeLa cells stably expressing an active-site mutant of PARKIN. PARKIN C431S traps Ub in an oxyester bond (8 kDa band shift) and facilitates the analysis of the E3 ligase’s Ub-charging as a read-out for its activation [22]. Consistently, significant PARKIN C431S ~ Ub levels were only observed upon re-transfection of cells with PINK1 WT but not with the p.I368N mutant or an empty vector (Fig. 6d, e). Altogether, these data confirm the lack of p.I368N kinase activity even at protein levels comparable to PINK1 WT, and thus its inability to activate PARKIN and the protective mitochondrial quality control.