Quantitative proteomic analysis of Parkin substrates in Drosophila neurons
- 3.8k Downloads
Parkin (PARK2) is an E3 ubiquitin ligase that is commonly mutated in Familial Parkinson’s Disease (PD). In cell culture models, Parkin is recruited to acutely depolarised mitochondria by PINK1. PINK1 activates Parkin activity leading to ubiquitination of multiple proteins, which in turn promotes clearance of mitochondria by mitophagy. Many substrates have been identified using cell culture models in combination with depolarising drugs or proteasome inhibitors, but not in more physiological settings.
Here we utilized the recently introduced BioUb strategy to isolate ubiquitinated proteins in flies. Following Parkin Wild-Type (WT) and Parkin Ligase dead (LD) expression we analysed by mass spectrometry and stringent bioinformatics analysis those proteins differentially ubiquitinated to provide the first survey of steady state Parkin substrates using an in vivo model. We further used an in vivo ubiquitination assay to validate one of those substrates in SH-SY5Y cells.
We identified 35 proteins that are more prominently ubiquitinated following Parkin over-expression. These include several mitochondrial proteins and a number of endosomal trafficking regulators such as v-ATPase sub-units, Syx5/STX5, ALiX/PDCD6IP and Vps4. We also identified the retromer component, Vps35, another PD-associated gene that has recently been shown to interact genetically with parkin. Importantly, we validated Parkin-dependent ubiquitination of VPS35 in human neuroblastoma cells.
Collectively our results provide new leads to the possible physiological functions of Parkin activity that are not overtly biased by acute mitochondrial depolarisation.
KeywordsParkin (PARK2) Parkin substrates Ubiquitination VPS35 Neurodegeneration In vivo Drosophila melanogaster Parkinson’s Disease (PD) Alzheimer’s Disease (AD) Label Free Quantification (LFQ)
Amyloid precursor protein
Endosomal sorting complexes required for transport
Label Free Quantification
Outer Mitochondrial Membrane
PTEN-induced kinase 1
RING between RING
Retinal pigment epithelial cells
- UBL domain
Parkinson’s Disease (PD) is the most common form of Parkinsonism and the second most common neurodegenerative disease after Alzheimer’s. PD patients display resting tremor, rigidity and postural disability, usually accompanied by other non-motor symptoms. Phenotypically, PD is mainly characterized by loss of dopaminergic neurons in the substantia nigra and the presence of Lewy bodies [1, 2]. PD has been classically considered a sporadic disease linked to aging with an unknown aetiology. However, in about 10% of the cases, mutations in specific genes cause Familial forms of PD. These genes show Mendelian inheritance and can be classified as either autosomal dominant (SNCA, LRRK2, VPS35) or autosomal recessive (PARK2, PARK7, PINK1, ATP13A2, FBXO7, PLA2G6, DNAJC6, SYNJ1). Moreover, recent exome sequencing and cohort genome-wide association studies (GWAS) have identified several other risk factor genes associated with sporadic PD and other Parkinsonian syndromes [3, 4, 5, 6]. The set of genes implicated in PD encode for proteins involved in mitochondrial homeostasis, autophagy, endo-lysosomal trafficking, Ca+2 homeostasis and dopamine homeostasis [7, 8, 9, 10]. However, the exact pathophysiological mechanisms leading to the disease are not yet clear. At the moment there is no effective biomarker for the diagnosis of PD, which can only be determined by postmortem brain analysis .
Amongst PD related genes, mutations in Parkin (PARK2) and PTEN-induced kinase 1 (PINK1) cause early-onset Familial PD [12, 13]. Parkin is a RING between RING (RBR) E3 ubiquitin ligase, which conveys the transfer of ubiquitin onto selected substrate proteins [14, 15]. Parkin null Drosophila melanogaster display Parkinsonian-like phenotypes including reduced life span, climbing and flying disability, sterility, mitochondrial defects and dopaminergic neurodegeneration . Genetic studies in Drosophila established that pink1 acts upstream of parkin to maintain mitochondrial integrity [17, 18]. Upon mitochondrial depolarization PINK1 accumulates at the Outer Mitochondrial Membrane (OMM), where it phosphorylates both ubiquitin and the Ubiquitin-like (UBL) domain of Parkin to recruit and activate latent Parkin ubiquitin ligase activity [19, 20, 21, 22, 23, 24, 25]. Activated Parkin ubiquitinates several OMM proteins and promotes both proteasome-dependent degradation of specific proteins and mitophagy, a specialised type of autophagy where the whole mitochondrion is engulfed into autophagosomes [26, 27, 28].
PINK1 and Parkin are widely considered neuroprotective and different studies have shown that PINK1/Parkin over-expression can protect against cell death in a number of contexts in vitro and in vivo . Therefore it has been proposed that drugs promoting PINK1/Parkin - dependent mitophagy could serve as effective treatments for PD. However, recent evidence demonstrates that excessive Parkin over-expression results in sensitization to cell death using in vitro [30, 31, 32] and in vivo models .
It is essential to identify physiologically relevant Parkin substrates to understand the pathways leading to PD in order to develop a treatment. A considerable number of proteins have been reported to be Parkin substrates but most of the work has relied on cultured cells, mainly of epithelial origin, usually upon treatment with mitochondrial depolarising agents [27, 34, 35, 36, 37, 38]. Here we extend this approach by performing a high throughput mass spectrometry proteomic study of Parkin substrates in vivo. We have utilised a fly model expressing constitutively biotinylated ubiquitin [39, 40, 41, 42, 43] to purify proteins ubiquitinated by Parkin in Drosophila neurons. Our study identifies both established and novel Parkin substrates.
Drosophila park gene was amplified from a Drosophila cDNA library (DGC realease 1.0, Berkeley Drosophila Genome Project) and FLAG-tag cloned at its 5’-end using the FLAG-parkin-Fw (GCCCTCGAGATGGATTACAAGGATGATGACGATAAGATGAGTTTTATTTTTAAATTTATTGCCAC) and parkin-Rv (GCCTCTAGATTAGCCGAACCAGTGGGCTCC) primers. This construct was then inserted into a pUASattb vector between the XhoI and XbaI sites. Ligase-dead FLAG-Parkin (ParkinLD) was generated by mutating the C449 to S using the QuikChange Site-Directed Mutagenesis Kit (Stratagene) according to manufacturer’s instructions. The primers used for mutagenesis were C449S-Fw (GGAGCGAGATGGCGGTAGCATGCACATGGTCTGCACACG) and C449S-Rv (CGTGTGCAGACCATGTGCATGCTACCGCCATCTCGCTCC). Untagged human Parkin and Parkin (C431S) were amplified from pcDNA3.1(+)-HA-Parkin and pcDNA3.1(+)-HA-Parkin(C431S) respectively with primers forward (GCCGAAGCTTAACCATGATAGTGTTTGTCAGG) and reverse (AGTCTAGACTACACGTCGAACCAGTGGTCCTGGG). PCR products were inserted into pcDNA3.1(+) between HindIII and XbaI.
The following antibodies were used against Drosophila proteins: goat anti-biotin-horseradish peroxidase (HRP) conjugated antibody (Cell Signalling); chicken polyclonal anti-BirA antibody (Sigma); rabbit polyclonal anti-Parkin antibody ; mouse monoclonal anti-Syx1A antibody (DSHB); rabbit polyclonal anti-RdhB ; rabbit polyclonal anti-ArgK ; rabbit polyclonal anti-Vps4 ; rabbit polyclonal anti-Fax antibody (a gift from Eric Liebl); rabbit polyclonal anti-Ubiquitin antibody (Sigma). The following antibodies were used against Human proteins: goat polyclonal anti-VPS35 antibody (Abcam); mouse monoclonal anti-Cleaved Parp-85 fragment (Cell Signaling); mouse monoclonal anti-Parkin (Santa Cruz); rabbit polyclonal anti-PINK1 (Novus Biologicals); rabbit polyclonal anti-Miro1 (Sigma); rabbit polyclonal anti-Tim44 (Sigma); rabbit polyclonal anti-Tom20 (Sigma); mouse monoclonal (Abcam) and rabbit polyclonal (Sigma) anti-Actin. For monitoring the GFP pull-downs the following antibodies were used: monoclonal mouse anti-GFP antibody (Roche) and monoclonal mouse anti-Flag M2-HRP conjugated antibody (Sigma). Anti-mouse, rabbit and chicken HRP labelled secondary antibodies (Jackson ImmunoResearch Laboratories) and anti-guinea pig (Invitrogen) were used; and anti-mouse, rabbit and sheep IR 680 and IR800-coupled antibodies (LI-COR Biosciences).
UAS-BirA and UAS-( Bio Ub) 6 -BirA  and their recombination with GMR-GAL4 flies for the study of ubiquitin proteomics has been previously described . FLAG-tagged Parkin wild-type (ParkinWT) and ParkinLD flies were generated at Bestgene using the pUASattb constructs described above. Both UAS-Parkin WT and UAS-Parkin LD lines were independently crossed with GMR-GAL4,UAS-( Bio Ub) 6 -BirA to finally generate: GMR-GAL4,UAS-( Bio Ub) 6 -BirA/CyO;UAS-Parkin WT and GMR-GAL4,UAS-( Bio Ub) 6 -BirA/CyO;UAS-Parkin LD . GMR-GAL4/CyO;UAS-BirA/TM6 and GMR-GAL4,UAS-( Bio Ub) 6 -BirA/CyO flies were additionally used as controls. UAS-GFP, elav-GAL4, GMR-GAL4, Tub-GAL4, Da-GAL4, Ple-GAL4 flies were obtained from Bloomington Drosophila Stock Center. UAS-GFP CL1 flies were obtained from  and park 25 /TM6b GFP-w + and UAS-park were obtained from . Flies were grown in 12 h light-dark cycles at 25 °C and were fed with wheat flour and yeast food (1% agar, 5.5% dextrose, 3.5% wheat flour, 5% yeast, 0.25% Nipagen, 0.4% Propionic acid and 0.02% Benzalkonium Chloride in distilled H2O).
Flies of indicated ages and genotypes were anesthetised with CO2 on a pad, and 20 flies (10 male and 10 female) were randomly selected. After an hour of recovery, flies were transferred to a climbing vial and ability to climb was scored as followed. Flies were gently tapped to the bottom and the number of flies that reached the 10 cm mark at 30 s was counted three times, with 30 s interval.
One hundred newborn flies of the indicated genotypes were maintained in wheat flour and yeast food in 12 h light-dark cycles at 25 °C. The vials were changed every 2–3 days and the number of flies alive was counted.
Fly head extract preparation and cell lysis
Six heads (three males and three females) of adult flies were cut and homogenised in 60 μL of 4x Laemmli buffer with DTT. Samples were centrifuged 1 min at maximum spin and the supernatant was recovered. Cells were harvested using “Hot Lysis buffer” (2% SDS, 50 mM NaF and 1 mM EDTA at 110 °C; Fig. 8). Protein concentrations were determined by BCA protein assay (Pierce).
Cell culture, siRNA knockdown and transfection
hTERT-RPE1-Parkin cells [32, 49] were cultured in Dulbecco’s modified Eagle’s medium (DMEM)/F12 with 10% FBS, 1% non-essential amino acids and 1% penicillin/streptomycin. Reverse transfection was performed using Lipofectamine RNAi-MAX (Invitrogen) and carried out for 72 h according to manufacturer’s instruction. Further 165 h double-hit knockdown was executed by transfecting again previously siRNA treated cells at 72 h. Cells were transfected with the following siRNA at a final concentration of 20nM: Non-Targeting siRNA oligo #1 (NT1) (ONTARGETplus: NT1; 5’-UGGUUUACAUGTCGACUAA-3’) and VPS35 (SMARTpool ONTARGETplus VPS35 siRNA oligo#5: 5’-GAACAUAUUGCUACCAGUA-3’; oligo#6: 5’-GAAAGAGCAUGAGUUGUUA-3’; oligo#7: 5’-GUUGUAAACUGUAGGGAUG-3’; oligo#8: 5’-GAACAAAUUUGGUGCGCCU-3’) from Dharmacon.
Human neuroblastoma SH-SY5Y cells were cultured in DMEM supplemented with 10% FBS (Thermo Scientific) and 1:100 penicillin/streptomycin at 37 °C. 300,000 cells were seeded in a 6 well-plate and incubated overnight under serum starvation. The following day, medium was removed, replaced by fresh DMEM and cells were transfected with 1 μg of FLAG-Ub, 1 μg of YFP-VPS35 together with 1 μg of pcDNA3.1 (control) or 1 μg of hParkin or 1 μg of hParkin (C431S) using Lipofectamine 3000 (Invitrogen) for 72 h according to manufacturer’s instruction.
GFP beads pull-down assay
Transfected cells were lysed in 500 μl of lysis buffer [50 mM Tris–HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.5% Triton, 1 × Protease Inhibitor cocktail (Roche Applied Science), 50 mM N-ethylmaleimide (NEM, from Sigma)] and collected for centrifugation at 14,000 × g for 10 min. The supernatant was mixed with 25 μL of GFP-Nanotrap beads (Chromotek GmbH) that had been pre-washed with dilution buffer (10 mM Tris–HCl pH 7.5, 150 mM NaCl, 0.5 mM EDTA, 1 × Protease Inhibitor cocktail, 50 mM NEM). The mixture was incubated at RT for 150 min with gentle rolling and centrifuged for 2,700 × g for 2 min. The supernatant was removed and the beads washed once with dilution buffer, three times with washing buffer (8 M Urea, 1% SDS in PBS) and once with 1% SDS in 1× PBS. The bound proteins were eluted with sample loading buffer (250 mM Tris–HCl pH 7.5, 40% glycerol, 4% SDS, 0.2% BPB) by heating at 95 °C for 10 min. Eluted samples were run into 4–15% Tris–Glycine gels.
Western blotting and silver staining
For fly samples the amount of sample loaded largely was equivalent to one head per lane but was optimized for each antibody. Generally, 4–15% gradient TGX gels (Bio-Rad) were used and proteins were transferred to PVDF membranes using the iBlot system (Invitrogen). Following primary and secondary antibody incubation, membranes were developed using ECL kit (Biorad Clarity). Alternatively, western blots were scanned using the LI-COR Odyssey system (LI-COR Biosciences). Silver staining was performed with a SilverQuest kit (Invitrogen) following the manufacturer’s instructions. Dual-colour westerns were prepared by assigning independent colour channels to two independent westerns developed in the same membrane.
The BioUb pulldown was performed as previously described [39, 40, 43] using GMR-GAL4,UAS-( Bio Ub) 6 -BirA/CyO; UAS-Parkin WT , GMR-GAL4,UAS-( Bio Ub) 6 -BirA/CyO;UAS-Parkin LD , GMR-GAL4/CyO;UAS-BirA/TM6 and GMR-GAL4, UAS-( Bio Ub) 6 -BirA/CyO flies. Briefly, 2–5 day old adult flies were fragmented by flash freezing in liquid nitrogen and shaking. A pair of sieves with a nominal cut-off of 710 and 425 μm was used to separate the heads from the rest of the fragments. About 0.5 g of heads were homogenized in 2.9 mL Lysis buffer (8 M Urea, 1% SDS and 50 mM N-ethylmaleimide in PBS, including a protease inhibitor mixture from Roche Applied Science). 250 μL of NeutrAvidin-agarose beads (Thermo Scientific) were used to purify BioUbiquitinated material and after stringent washes with a succession of buffers containing 8.4 M urea, 6 M Gdn-HCl, 1 M NaCl and 2%SDS, beads were finally eluted with 125 μL Elution Buffer (4X Laemmli Buffer and 100 mM of DTT). Recovered volume was ~150 μL with an average yield of 20–40%.
Eluted samples from three different experiments were briefly run on a SDS-PAGE, which was then cut in four slices, to separate the avidin band from the other proteins in the gel. Proteins were digested using automated in-gel digestion protocol . The digested peptides were cleaned-up on Stage-tips . The eluted peptides were resuspended in 0.1% formic acid and separated on a 15 cm reverse phase column (75 μm inner diameter, 3 μm reprosil beads, Dr. Maisch, GmbH, in house packed) using 5 to 50% acetonitrile gradient (VWR). Peptide ionization was performed on a Proxeon ionsource and sprayed directly into the mass spectrometer (Q-Exactive, Thermo Scientific). The MS was recorded with mass resolution of 60000, while the MS2 spectra were collected with a resolution of 35000 using a fixed fill time of 120 ms. For the analysis of the recorded mass spectra the MaxQuant software package (version 18.104.22.168) was used with 1% FDR for both peptides and proteins . Searches were performed using the Andromeda search engine against the Uniprot Drosophila melanogaster database (downloaded 15.08.2015). As a fixed modification, cysteine carbamidomethylation was selected and as variable modifications, methionine oxidation, protein N-terminal acetylation and di-glycine addition on the ε-aminogroup of lysines. Two missed trypsin (full specificity) cleavages were allowed and the MaxQuant label free quantification for proteins was activated. The identified ubiquitination sites were checked visually using the Max-Quant View program and identification artefacts were removed as previously described .
Ubiquitin chain linkage quantification
Samples from three technical replicates were loaded on an 8% stacking gel and proteins compressed in one band. Proteins were digested in-gel using trypsin, and purified as described above. The peptides were separated on a 25 cm reverse phase column (75 μm inner diameter, 3 μm reprosil beads, Dr. Maisch, GmbH, in house packed) using 5 to 50% acetonitrile gradient (VWR). The peptides were ionized on a Nano3 ion source (ABSciex) and directly sprayed into a QTRAP 5500 triple-quadrupole mass spectrometer, run in MRM mode to detect branched ubiquitin peptides and their synthetic isotopically labelled counterparts [54, 55]. Each sample was injected as technical triplicates. Two transitions for each peptide, i.e., K6, K11, K48, K63, were acquired. After the measurement peaks were integrated and light/heavy ratios calculated using the Skyline software package  reflecting the abundance of the linkage type compared to the internal heavy standard.
Bioinformatic and statistical analysis
Non-ubiquitinated/background proteins were excluded by a 4-fold change threshold of the average LFQ intensity between BioUb, BioUb + ParkinWT or BioUb + ParkinLD against the BirA sample. LFQ intensity values of ubiquitinated proteins were imported to Perseus software (http://www.perseus-framework.org/) and the imputation tool was used to replace missing values. Subsequently, values were grouped in categories and two T-tests were performed: BioUb + ParkinWT vs BioUb and BioUb + ParkinWT vs BioUb + ParkinLD. Proteins with a p-value smaller than 0.05 and a fold change bigger or smaller than 2 in any of these two tests were selected for further filtering. Proteins with less than two unique peptides in all the three BioUb + ParkinWT (enriched) and BioUb + ParkinLD (reduced) were excluded. To further exclude false positives, all unique peptide profiles of these proteins were analysed. Peptides that appeared only in one condition were excluded and the average peptide intensity was calculated for each peptide in each condition. On the one hand, average of all peptide intensities between BioUb + ParkinWT vs BioUb or BioUb + ParkinLD were calculated and only the proteins whose total fold change was bigger (enriched) or smaller (reduced) than two were further selected. On the other hand, the average of all fold changes of peptide intensities between BioUb + ParkinWT vs BioUb or BioUb + ParkinLD minus the SEM (Standard Error of the Mean) was required to be bigger or smaller than two. Only the proteins that successfully passed all the tests were selected as: Parkin substrates (35) and less ubiquitinated in a Parkin - dependent manner (2).
A two sample T-Test for Parkin proteomics was performed using Perseus software applying default parameters: t-test; S0; Side = both; threshold p-value = 0,05. Imputation tool was used as default: width = 0,3; down shift = 1,8; mode = separately for each column. Other statistical analysis was performed with Graphpad Prism6. For the GO Term identification g:Cocoa tool of g:Profiler web server was used using archive “Ensembl 79, Ensembl Genomes 26 (r1395)”. p-values are indicated as *p ≤ 0.05, **p ≤ 0.01 and ***p ≤ 0.001. Functional connectivity analysis was performed using STRING (http://string-db.org/) web tool.
For GFP pull-downs protein bands were analysed and quantified by densitometry using ImageJ software. Statistical significance in Western blotting quantification was evaluated using an analysis of variance (ANOVA) complemented by Tukey’s honest significance difference test (Tukey’s HSD) performed in GraphPad PRISM software. Statistical significance levels are annotated as ns, not significant; **, p < 0.01 (mean ± SEM, n = 4).
Excessive Parkin activity in developing neurons causes climbing defects and reduced life span
Parkin substrate identification in adult Drosophila neurons
Mass Spectrometry analysis of Parkin - dependent ubiquitination identified novel putative Parkin substrates in Drosophila neurons
The most robust Parkin responders identified in Drosophila neurons
Parkin preferentially generates K6 ubiquitin chains in Drosophila neurons
Validation of Parkin candidate substrates
Parkin-dependent Vps35 ubiquitination could not be validated in flies by western blotting, due to the lack of available working antibodies for Drosophila Vps35. We instead performed in human cells an in vivo ubiquitination assay we recently developed [42, 43, 57, 58]. The main advantage of this method is that, due to the very stringent washes performed, the entire ubiquitin signal detected corresponds exclusively to the protein of interest and cannot be derived from other proteins in the cell. We tested this assay on SH-SY5Y neuroblastoma cells, which allowed us to further test whether Parkin-dependent ubiquitination of VPS35 is conserved in human cells.
VPS35 knockdown does not affect Parkin - dependent mitophagy in YFP-Parkin expressing RPE1 cells
Our results confirm a linkage between Parkin and the PD-associated protein VPS35, which has been previously revealed by genetic studies . Both the proteomic data and the in vivo ubiquitination assay coupled to subsequent western blot validation suggest that VPS35 may be a substrate of Parkin, in line with previous studies that analysed Parkin substrates following acute mitochondrial depolarisation . The mechanism that links VPS35 to PD is unclear, although recent evidence indicates that VPS35 regulates mitochondrial homeostasis by regulating fission-fusion dynamics [60, 61] and the trafficking of mitochondrial-derived vesicles (MDV) [62, 63]. We and others have recently used retinal pigment epithelial cells (RPE1) that stably over-express YFP-Parkin as a convenient system for analysis of Parkin-dependent mitophagy [32, 49]. We tested whether VPS35 is necessary for PINK1/Parkin - dependent mitophagy in these cells following acute mitochondrial depolarisation. However, siRNA-mediated knockdown of VPS35 had no effect upon mitophagy as judged by the time-dependent loss of several mitochondrial marker proteins (Fig. 8c).
Parkin regulates a delicate balance between survival and cell death
Our finding that neuronal Parkin over-expression during neurodevelopment leads to multiple Parkinsonian-like defects conflicts with the simple notion that Parkin is neuroprotective . However, recent cell culture and in vivo studies have shown that Parkin activity can also sensitize towards cell death in certain contexts [30, 31, 32, 33]. We observed that the severity of Parkin over-expression phenotype is associated with its E3 ubiquitin ligase activity in a dosage dependent manner, which is further augmented during aging. Nevertheless, the fact that flies over-expressing ParkinLD still have a mild phenotype, suggests that this mutant shows a dominant-negative effect that is not linked to the ligase function. Further in vivo experiments are necessary to unravel Parkin function at different developmental stages in different tissues.
Functional roles of identified substrates
Identification of Parkin substrates is crucial in order to comprehend the still elusive aetiology of PD. While early studies stressed the notion that Parkin ubiquitinates and degrades several aggregation-prone toxic proteins, currently Parkin is mainly associated with mitophagy, through ubiquitination of several OMM proteins [35, 36, 64, 65, 66]. Parkin is predominantly held in an inactive conformation at steady state  and PINK1 dependent phosphorylation of both ubiquitin and Parkin is necessary for its activation [19, 20, 21, 22, 23, 24, 25]. Mitochondrial depolarisers have been proven to efficiently activate Parkin through PINK1 accumulation, but most cell lines show low levels of Parkin, precluding detectable levels of Parkin-dependent mitophagy and ubiquitination. Moreover, the ubiquitinated fraction of proteins is usually low due to their fast proteasomal degradation and/or the activity of deubiquitinating enzymes (DUBs) . Therefore, most studies have been performed under conditions of Parkin over-expression using non-neuronal in vitro cell culture models, usually employing proteasome inhibitors and mitochondrial depolarisers. However, these acute conditions are not encountered in physiological settings. In the present study we have sought for in vivo Parkin substrates in Drosophila neurons, using a fly model that co-expresses constitutively biotinylated ubiquitin together with FLAG-tagged Parkin. While it has been shown that tagging human Parkin can lead to some constitutive activation , Drosophila Parkin has an extended N-terminal sequence relative to the human protein and addition of a further short sequence tag is not predicted to impact on the structure (Additional file 8: Figure S8). If such a scenario was extant, it would simply shift an equilibrium and potentially allow more sensitive detection of Parkin substrates without the recourse to acute drug treatments used in previous studies [27, 34, 36, 37, 38]. Treatments with mitochondrial depolarising drugs will lead to PINK1 accumulation and activation of Parkin specifically at mitochondria. Our study provides a complementary resource, which may capture some Parkin substrates that are PINK1-independent. The notion of alternative activation routes for Parkin has been previously proposed  and it is supported by data suggesting a role for Parkin in xenophagy that might be PINK1-independent .
The fact that there is no singular pathway controlled by substrates described here suggests that under in vivo physiological conditions Parkin may have broader roles beyond the control of mitophagy. For example, we note prominent representation of endosomal-associated proteins, including three sub-units of the v-ATPase required for endosome and lysosomal acidification, two proteins associated with the endosomal sorting complexes required for transport (ESCRT) machinery (ALiX/PDCD6IP, VPS4) and the PD-associated gene and retromer component VPS35. Moreover, proteins involved in protein synthesis, metabolic process, cell death, calcium signaling and immune response are also present (Additional file 10: Table S2). Further study of Parkin substrates will reveal whether they act in an unexplored common pathway.
Parkin monoubiquitinates substrates and preferentially generates K6 ubiquitin chains
A single ubiquitin can be attached to a unique (monoubiquitination) or several (multiple-monoubiquitination) lysine residues of substrate proteins. Ubiquitin can be generated from these stubs to form chains of different topologies (M-1, K6, K11, K27, K29, K33, K48, K63). This provides a complex “ubiquitin code” that regulates a myriad of biological processes. Whilst K48 regulates protein turnover, through proteasomal degradation, and K63 is mainly involved in endosomal trafficking and DNA repair, the function of the “non-canonical” chains is less well defined [78, 79]. Parkin has been reported to be able to promote monoubiquitination as well as multiple chain types such as K6, K11, K27, K48 and K63 ubiquitin chains in diverse contexts through interaction with different E2s [64, 80, 81, 82, 83]. Moreover, Parkin has been shown to autoubiquitinate itself via K6 ubiquitin chains and the deubiquitinase USP8 has been suggested to remove these K6 chains to regulate Parkin stability and its recruitment to mitochondria . Furthermore, USP15, USP30 and USP35 have been shown to deubiquitinate Parkin substrates and/or regulate mitophagy following acute mitochondrial depolarisation in cells [32, 38, 85, 86]. While USP15 has been reported to have a preference for K48 and K63 , USP30 has been reported to favour K6 chains disassembly .
In contrast to our earlier work using this same bioUb pulldown and Western blotting approach [39, 40, 42, 43], most of the specific proteins we have validated here by western blotting were found to be monoubiquitinated. This agrees with previous observations demonstrating that Parkin is able to generate monoubiquitinated conjugates [81, 82]. Our ubiquitin chain analysis showed that Parkin preferentially promotes K6 chain formation in contrast to some literature reports indicating that Parkin promotes a more diverse set (K6, K11, K48 and K63) after mitochondrial depolarisation [83, 88]. However, in line with our observations, recent ubiquitin chain analysis showed that K6 is preferentially formed in whole lysates from Parkin over-expressing HEK and SH-SY5Y cells after mitochondrial depolarisation .
Implications of Parkin-dependent VPS35 ubiquitination in neurodegeneration
Despite the fact that the exact molecular mechanism leading to PD is unclear, the set of genes linked to PD can be divided in two main pathways. Parkin, DJ-1, PINK1 and FBXO7 are linked to mitochondrial homeostasis [65, 66, 89, 90], while α-synuclein (SNCA), LRRK2, VPS35, EIF4G1 and ATP13A2 are involved in endosomal-lysosomal trafficking [7, 8, 9]. LRRK2, RAB7L1 and VPS35 have been shown to act in a common pathway [91, 92] and VPS35 has recently been shown to genetically interact with EIF4G1 in a pathway upstream of α-synuclein . Furthermore, VPS35 impairment leads to lysosomal dysfunction and α-synuclein pathology [94, 95].
The finding that Parkin ubiquitinates VPS35 is of considerable interest, as it could represent a connection between mitochondrial homeostasis and endo-lysosomal pathways, unifying the PD-associated genes in a common network . In fact, recent studies in Drosophila reported that Vps35 genetically interacts with parkin . Our identification of VPS35 as a Parkin substrate strengthens this association. Moreover, VPS35 has been found to be involved in mitochondrial homeostasis [60, 61], therefore it is tempting to consider that Parkin-dependent ubiquitination of VPS35 may regulate mitophagy. However, our results show that Parkin-dependent mitophagy is not affected by VPS35 knockdown, suggesting that VPS35 is not an essential regulator of Parkin-dependent mitophagy in a mammalian cell system subjected to acute depolarisation. Nor do we see any changes in VPS35 levels following Parkin activation (Fig. 8c) or with control RPE-1 cells which do not express Parkin (not shown). Alternatively, Parkin - dependent-ubiquitination of VPS35 could be involved in the regulation of mitochondrial-derived vesicles (MDV), which have been reported to traffic defective mitochondrial proteins to peroxisomes and lysosomes in a more specific manner prior to the whole organelle impairment [62, 63, 97]. A further possibility is that Parkin ubiquitinates and regulates endo-lysosomal pathway components through the ubiquitination of several proteins, including VPS35 and others described here. During the preparation of this manuscript Parkin has been reported to modulate endo-lysosomal trafficking through Rab7 ubiquitination, but also to influence the distribution of VPS35 and the levels of the retromer cargo protein ci-M6PR . Other recent papers have also linked Parkin to endo-lysosomal trafficking [99, 100, 101, 102]. The involvement of endocytic trafficking in PD is gaining momentum with the finding that two main PD-associated genes, PINK1 and LRRK2, converge and lead to the phosphorylation of common Rab proteins, small GTPases involved in membrane trafficking [103, 104, 105].
Interestingly, both VPS35 and Parkin mutations are also associated with Alzheimer’s Disease (AD) [106, 107, 108]. VPS35 interacts with the amyloid precursor protein (APP) sorting receptor, SorL1, which is also an AD causing gene, to regulate APP processing and Aβ production [109, 110, 111, 112, 113]. Parkin has recently been identified as an AD risk gene  and decreased Parkin levels have been found in AD patient cerebrospinal fluids . Remarkably, Parkin over-expression has been shown to ameliorate AD symptoms by increasing the clearance of amyloid loads [115, 116], which may suggest that defects in a common endo-lysosomal pathway are shared between PD and AD. In this scenario, Parkin may have a role in endosomal trafficking control through ubiquitination of VPS35, with a potential impact on both PD and AD. Certainly, further work is required in order to explore this hypothesis.
Constitutive expression of biotinylated ubiquitin has been proven to be a successful approach to study ubiquitination in vivo [39, 40, 43]. Here we extend this method to identify Parkin substrates in adult Drosophila neurons. We report 35 substrates, eight of which have been previously detected and 27 are novel. Notably we identified VPS35 as an in vivo Parkin substrate, both in flies and in human neuroblastoma SH-SY5Y cells, reinforcing a connection between mitochondrial homeostasis and endo-lysosomal pathways in PD and AD.
We would like to thank So Young Lee (CIC bioGUNE, Spain) for her contribution to the generation of the FLAG-Parkin flies. We also would like to acknowledge Jon D Lane (Bristol, UK) for providing us hTERT-RPE1-YFP-Parkin cell line and Alex Whitworth (Cambridge, UK) for kindly providing park 25 /TM6b GFP-w + and UAS-park flies and Drosophila Parkin antibody. We also thank Jesus Mari Arizmendi and Kerman Aloria at the University of Basque Country (UPV-EHU) for insightful discussions on MS data analysis.
A.M. is a recipient of a CIC bioGUNE/Liverpool University studentship. U.M. is a recipient of a MINECO grant (SAF2013-44782-P) co-financed with FEDER funds.
Availability of data and materials
All data generated or analysed during this study are included in this published article [and its supplementary information files].
AM performed all experiments except the MS analysis, which was performed by OP and GD, and the GFP pulldown assay, which was performed by BL. AM, BL, JR, SU, MJC and UM designed and interpreted the experiments. AM, JR and UM performed the bioinformatic analysis. JDS, SU, GD, MJC and UM contributed reagents/materials/analysis tools. JDS and SU gave critical input and comments upon the manuscript. All authors have been involved in drafting the manuscript and revising it critically for important intellectual content. All authors read and approved the final manuscript.
The authors declare that they have no competing interests.
Consent for publication
Ethical approval and consent to participate
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Parkin over-expression in neurodevelopment results in Parkinsonian-like defects: Empty vials containing 5 male and 5 female, 0–3 days after eclosion flies of elav-GAL4/CyO;UAS-Parkin WT in the left side and elav-GAL4/CyO;UAS-Parkin LD in the right side. (M4V 4550 kb)
- 7.Gan-Or Z, Dion PA, Rouleau GA. Genetic perspective on the role of the Autophagy-Lysosome Pathway in Parkinson disease. Autophagy. 2015;0:00.Google Scholar
- 15.Riley BE, Lougheed JC, Callaway K, Velasquez M, Brecht E, Nguyen L, et al. Structure and function of Parkin E3 ubiquitin ligase reveals aspects of RING and HECT ligases. Nat Commun. 2013;4. [cited 2015 Aug 21]. Available from: http://www.nature.com/ncomms/2013/130617/ncomms2982/full/ncomms2982.html.
- 19.Narendra DP, Jin SM, Tanaka A, Suen D-F, Gautier CA, Shen J, et al. PINK1 is selectively stabilized on impaired mitochondria to activate Parkin. PLoS Biol. 2010;8. [cited 2015 Aug 20]; Available from: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2811155/.
- 35.Sandebringand and Cedazo-Mínguez A. J. Parkin- An E3 Ubiquitin Ligase with Multiple Substrates - parkin-an-e3-ubiquitin-ligase-with-multiple-substrates-2161-0460.S10-002.pdf [Internet]. 2012 [cited 2015 Oct 5]. Available from: http://www.omicsonline.org/parkin-an-e3-ubiquitin-ligase-with-multiple-substrates-2161-0460.S10-002.pdf
- 50.Kanashova T, Popp O, Orasche J, Karg E, Harndorf H, Stengel B, et al. Differential proteomic analysis of mouse macrophages exposed to adsorbate-loaded heavy fuel oil derived combustion particles using an automated sample-preparation workflow. Anal Bioanal Chem. 2015;407:5965–76.PubMedCrossRefGoogle Scholar
- 53.Beaudette P, Popp O, Dittmar G. Proteomic techniques to probe the ubiquitin landscape. Proteomics. 2015;16(2):273-87.Google Scholar
- 60.Tang F-L, Liu W, Hu J-X, Erion JR, Ye J, Mei L, et al. VPS35 deficiency or mutation causes dopaminergic neuronal loss by impairing mitochondrial fusion and function. Cell Rep. 2015.Google Scholar
- 74.Anthony M., Sylvia Christakos. Specific reduction of calcium-binding protein (28-kilodalton calbindin-D) gene expression in aging and neurodegenerative diseases [Internet]. 1990 [cited 2015 Sep 22]. Available from: http://www.pnas.org/content/87/11/4078.long
- 92.Linhart R, Wong SA, Cao J, Tran M, Huynh A, Ardrey C, et al. Vacuolar protein sorting 35 (Vps35) rescues locomotor deficits and shortened lifespan in Drosophila expressing a Parkinson’s disease mutant of Leucine-Rich Repeat Kinase 2 (LRRK2). Mol Neurodegener. 2014;9:23.PubMedPubMedCentralCrossRefGoogle Scholar
- 95.Tang F-L, Erion JR, Tian Y, Liu W, Yin D-M, Ye J, et al. VPS35 in dopamine neurons is required for endosome-to-Golgi retrieval of Lamp2a, a receptor of chaperone-mediated autophagy that is critical for α-synuclein degradation and prevention of pathogenesis of Parkinson’s disease. J Neurosci Off J Soc Neurosci. 2015;35:10613–28.CrossRefGoogle Scholar
- 100.Matheoud D, Sugiura A, Bellemare-Pelletier A, Laplante A, Rondeau C, Chemali M, et al. Parkinson’s disease-related proteins PINK1 and Parkin repress mitochondrial antigen presentation. Cell. 2016;0. [cited 2016 Jul 14]; Available from: http://www.cell.com/article/S0092867416305906/abstract.
- 104.Steger M, Tonelli F, Ito G, Davies P, Trost M, Vetter M, et al. Phosphoproteomics reveals that Parkinson’s disease kinase LRRK2 regulates a subset of Rab GTPases. Elife. 2016;5. pii: e12813.Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.