Parkin Phosphorylation and Modulation of Its E3 Ubiquitin Ligase Activity*

Mutations in the PARKIN gene are the most common cause of hereditary parkinsonism. The parkin protein comprises an N-terminal ubiquitin-like domain, a linker region containing caspase cleavage sites, a unique domain in the central portion, and a special zinc finger configuration termed RING-IBR-RING. Parkin has E3 ubiquitin-protein ligase activity and is believed to me-diate proteasomal degradation of aggregation-prone proteins. Whereas the effects of mutations on the struc-ture and function of parkin have been intensely studied, post-translational modifications of parkin and the regulation of its enzymatic activity are poorly understood. Here we report that parkin is phosphorylated both in human embryonic kidney HEK293 cells and human neu-roblastoma SH-SY5Y cells. The turnover of parkin phosphorylation was rapid, because inhibition of phosphata-ses with okadaic acid was necessary to stabilize phosphoparkin. Phosphoamino acid analysis revealed that phosphorylation occurred mainly on serine residues under these conditions. At least five phosphorylation sites were identified, including Ser 101 , Ser 131 , and Ser 136 (located in the linker region) as well as Ser 296 restriction

by the progressive loss of dopaminergic neurons in the substantia nigra pars compacta (1). Although more than 90% of PD cases occur sporadically, the study of genetic mutations has offered great insight into the molecular mechanisms of PD (2). After the discovery that mutations in the PARKIN gene cause autosomal recessive juvenile parkinsonism (3), parkin mutations have been recognized as the most common cause of hereditary PD and possibly a risk factor for idiopathic PD (4,5).
The PARKIN gene comprises 12 exons and codes for a 465amino acid protein that is widely expressed, most prominently in muscle and throughout the brain (3). The 52-kDa parkin protein comprises an N-terminal ubiquitin-like domain (aa 1-76), a unique parkin domain (aa 145-232), and two RING (really interesting new gene) fingers (aa 238 -293 and 418 -449, respectively) flanking an IBR (in-between RING) domain (aa 314 -377) at the C terminus. All of these domains appear to be functionally important, because PD mutations cluster in them (6).
Post-translational modifications often regulate enzymatic activity. Nitrosylation of parkin was recently found to occur in PD, leading to an inhibition of its ubiquitin ligase activity (23,24). Here we addressed the question whether phosphorylation of parkin occurred and, if so, by which kinases under what cellular conditions. Parkin was found to be phosphorylated on at least five serine residues. Casein kinase-1 (CK-1), protein kinase A (PKA) and protein kinase C (PKC) were identified as parkin kinases in vitro, and inhibition of CK-1 suppressed phosphorylation of parkin in cell lysates. Unfolded protein stress mediated by proteasomal inhibition or endoplasmatic reticulum (ER) stress, but not oxidative stress, reduced the overall phosphorylation of parkin. Unphosphorylated parkin isolated from eukaryotic cells or purified as recombinant fusion protein from bacteria showed a small but significant increase of autoubiquitin ligase activity, compared with parkin phosphorylated in vivo and in vitro. Thus, we suggest that modulation of the phosphorylation state of parkin has a regulatory role on its E3 ubiquitin ligase activity.

MATERIALS AND METHODS
Cell Culture, Transfection, and Establishment of Stable Transfectants-HEK293, HEK293T, and SH-SY5Y cells were cultured in Dulbecco's modified Eagle's medium with Glutamax (PAA Laboratories GmbH) supplemented with 10% fetal calf serum for HEK293 cells and HEK293T cells or 15% for SH-SY5Y cells. Cells were transfected using Lipofectamine 2000 Reagent (Invitrogen) or FuGene (Roche Applied Science) according to the supplier's instructions. Stable HEK293 and SH-SY5Y transfectants were selected with 200 or 22.5 g/ml zeocin, respectively.
Construction of cDNAs-Human full-length parkin was amplified by PCR (all primer sequences are available upon request) using a parkin cDNA construct (a gift from R. Baumeister) and cloned into the XbaI/HindIII restriction sites of pcDNA3.1 zeo(Ϫ) (Invitrogen), yielding MYC-parkin.
Immunoprecipitation and Immunoblotting-Cells were harvested 24 h after the transfection and lysed in lysis buffer (10 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, and 1% Triton X-100) with proteinase inhibitors (Sigma) on ice for 15 min. Cell lysates were centrifuged at 4°C at 16,000 ϫ g for 20 min. Immunoprecipitations were carried out using protein G-Sepharose (Amersham Biosciences) or protein A-Sepharose, anti-Myc-agarose, anti-FLAG M2-agarose affinity gel (all from Sigma) at 4°C for 2 h. Immunoprecipitates were washed with lysis buffer three times (or, in the case of radiolabeled samples, six times). Whole cell extracts and immunoprecipitates were separated by SDS-PAGE, and proteins were transferred onto PVDF membrane (Immobilon; Millipore Corp.). Enhanced chemiluminescence detection reagents (Amersham Biosciences) were used to detect immunoblot signals of the following antibodies: 9E10 monoclonal anti-Myc (Developmental Studies Hybridoma Bank, University of Iowa), monoclonal anti-V5 (Invitrogen), monoclonal anti-KDEL against glucose-regulated proteins Grp78 and Grp94 (Stressgen), polyclonal anti-parkin (Cell Signaling), and PRK8 monoclonal anti-parkin (25) (kindly provided by V. Lee).
In Vivo Phosphorylation Assay-Twenty-four h after the transfection, cells were incubated for 45 min in phosphate-free medium (Sigma), and 13-36 MBq of [ 32 P]orthophosphate was added. After 30 min (to label transfected parkin) or 2 h (to label endogenous parkin), 1 M okadaic acid (OA) was added and incubated at 37°C for 1 h. The conditioned medium was aspirated, and the cell monolayer washed twice with ice-cold phosphate-buffered saline. Cells were lysed on ice with lysis buffer, and immunoprecipitations were performed as above. Immunoprecipitates were separated by SDS-PAGE and transferred onto PVDF membrane. Autoradiography was carried out to visualize radiolabeled proteins.
Matrix-assisted Laser Desorption Ionization Time-of-flight (MALDI-TOF) Mass Spectrometry-Phosphorylation of transfected MYC-parkin fragments was induced with OA, and MYC immunoprecipitates were subjected to SDS-PAGE. Colloidal blue-stained bands of interest were in-gel digested with endoproteinase Lys-C as described (26). After overnight digestion, about 1 l was mixed with 1 l of saturated ␣-cyanocinnamic acid in 50% acetonitrile, 0.1% trifluoroacetic acid in water and applied to the MALDI target. The samples were analyzed with a Bruker Daltonics (Bremen, Germany) Ultraflex TOF/TOF mass spectrometer.
An acceleration voltage of 25 kV was used. Calibration was internal to the samples with des-Arg-bradykinin and ACTH-(18 -38) (both peptides purchased from Sigma).
Nanoelectrospray Ionization Tandem Mass Spectrometry-In order to identify which of the three possible serines was phosphorylated in the parkin peptide obtained after digestion with endoproteinase Lys-C with a monoisotopic mass of 1568.75 Da, this proteolytic product was subjected to nanoelectrospray ionization tandem mass spectrometry on a QSTAR Pulsar I quadrupole TOF tandem mass spectrometer (Applied Biosystems/MDS-Sciex, Toronto, Canada) equipped with a nanoelectrospray ion source (Proxeon, Odense, Denmark) as described (27).
Dephosphorylation by Alkaline Phosphatase-HEK293 cells and SH-SY5Y cells were transiently transfected with MYC-parkin-V5 and various portions of parkin (MYC-parN-V5, MYC-parNM-V5, and MYC-parC-V5). Twenty-four h after the transfection, 1 M OA was added, and cells were incubated for 1 h. Immunoprecipitation using anti-Mycagarose conjugate (Sigma) was performed. Immunoprecipitates were incubated at 37°C for 1 h with or without calf intestinal alkaline phosphatase (CIP) (New England Biolabs) according to the supplier's manual. Reactions were terminated by adding 2ϫ SDS sample buffer and analyzed by immunoblot.
Phosphoamino Acid Analysis-Phosphoamino acid analysis was performed using the method by Jelinek and Weber (28). After electroblotting radiolabeled proteins onto PVDF membrane, bands were excised and hydrolyzed using 6 N HCl at 100°C for 90 min. After centrifugation, supernatants were dried in a SpeedVac concentrator. Pellets were dissolved in pH 2.5 buffer (5.9% glacial acetic acid, 0.8% formic acid, 0.3% pyridine, and 0.3 mM EDTA) and spotted onto thin layer chromatography plates (Merck) together with unlabeled phosphoamino acid markers (1 g each of Ser(P), Thr(P), and Tyr(P); Sigma). One-dimensional high voltage electrophoresis was performed at 20 mA for 50 min. Radioactive phosphoamino acids were identified by autoradiography and co-migration with the ninhydrin-stained standards.
In Vitro Phosphorylation Assays-Recombinant rat CK-1 ␦, recombinant ␣-subunit of human CK-2 and recombinant human Akt1/PKB protein kinase were used for in vitro phosphorylation assays according to the supplier's instructions (Cell Signaling). The catalytic subunit of PKA purified from bovine heart (gift from V. Kinzel) was used in a buffer containing 20 mM Tris, pH 7.5, 5 mM magnesium acetate, and 5 mM dithiothreitol. PKC purified from rat brain (Biomol) was used in a similar buffer to PKA supplemented with 1 M phorbol-12,13-dibutyrate (PDBu), 0.5 mM CaCl 2 , and 100 g/ml phosphatidylserine under mixed micellar conditions. Fusion proteins GST-parkin, GST-parN, GST-parC, and various serine-to-alanine mutations in GST-parN or GST-parC were used as substrates. The reaction was started by adding 10 M [␥-32 P]ATP (250 M [␥-32 P]ATP in the case of Akt1/PKB) and allowed to proceed for 30 min at 30°C.

FIG. 1. Parkin is phosphorylated both in HEK293 cells and SH-SY5Y cells. A, HEK293 cells (upper panels)
and SH-SY5Y cells (lower panels) stably expressing MYC-parkin were labeled with [ 32 P]orthophosphate in the presence or absence of OA, and then cell lysates were immunoprecipitated with 9E10. Proteins separated by 12% SDS-PAGE were transferred onto PVDF membrane. Autoradiography was carried out in order to visualize phosphorylated parkin (left panels). Afterward, blots were probed with polyclonal anti-parkin or 9E10 anti-Myc (right panels). B, endogenous parkin in HEK293 cells (upper panels) and SH-SY5Y cells (lower panels) was labeled with [ 32 P]orthophosphate and immunoprecipitated with polyclonal anti-parkin. After the autoradiography (left panels), blots were probed with PRK8 monoclonal anti-parkin (right panels). DMSO, Me 2 SO.
Alternatively, cell lysates were used to phosphorylate fusion proteins of GST carrying parkin, parN, and parC. HEK293 cells were lysed in a buffer containing 20 mM Tris, pH 7.5, 5 mM magnesium acetate, 5 mM dithiothreitol, and 0.5% Triton X-100. In case of PKC, 0.5 mM CaCl 2 , 1 M PDBu, and 100 g/ml phosphatidylserine were added. After centrifugation at 14,000 ϫ g at 4°C for 10 min, GST-parkin, GST-parN, GST-parC, or various serine-to-alanine mutants were added to supernatants. Phosphorylation reactions were started by adding [␥-32 P]ATP and allowed to proceed at 30°C for 30 min in the presence of 4 M OA and in the presence or absence of 5 M hymenialdisine (donated by L. Meijer), 5 M H-89 (Biomol), or 5 M GF 109203X (Biomol). After the reaction, precipitations with glutathione-Sepharose (Amersham Biosciences) were carried out at 4°C for 2 h. Precipitates were washed four times with phosphate-buffered saline and eluted by 2ϫ SDS sample buffer.
In Vitro Ubiquitination Assay-FLAG-parkin was transfected into HEK293T cells. Twenty-four h after the transfection, 1 M OA was added to cells and incubated for 50 min. Cells were harvested and lysed in lysis buffer. Alternatively, GST-parkin immobilized on glutathione-Sepharose, which was phosphorylated by CK-1, PKA, or PKC was used. Immunoprecipitates using anti-FLAG M2-agarose (Sigma) or immobilized phosphorylated GST-parkin were washed three times with lysis buffer and once with ubiquitination buffer (50 mM Tris-HCl, pH 7.4, and 5 mM MgCl 2 ). One mM dithiothreitol, 2 mM ATP, 100 ng of E1 (AFFINITI), 2 g of UbcH7 (AFFINITI or MBL) and 5 g of ubiquitin biotinylated using the EZ-Link Sulfo-NHS biotinylation kit (Pierce) were added to the immunoprecipitates. The reactions were conducted at 30°C for 90 min and terminated by adding 2ϫ SDS sample buffer. Reaction mixtures were resolved by 10% SDS-PAGE, and immunoblot was carried out using anti-FLAG M2 monoclonal antibody (Sigma) or streptavidin-peroxidase polymer (Sigma). Autoubiquitination (biotinylated) of parkin was quantified by densitometric scanning of the streptavidin-peroxidase developed blots. Image analysis was done using NIH Image version 1.62 (available on the World Wide Web at rsb.info.nih.gov/nih-image).

RESULTS
Parkin Is Constitutively Phosphorylated at the N Terminus and the C Terminus-In order to examine whether or not parkin is phosphorylated, we carried out in vivo phosphorylation assays. Stable HEK293 and SH-SY5Y transfectants expressing MYC-parkin were labeled with [ 32 P]orthophosphate. Phosphorylation was stabilized with OA, which inhibits phosphoprotein phosphatase 1, 2A, and 2B. Analysis of Myc-immunoprecipitated parkin by autoradiography revealed that full-length MYC-parkin undergoes phosphorylation both in HEK293 cells and SH-SY5Y cells (Fig. 1A). OA treatment also increased the [ 32 P]orthophosphate incorporation into endogenous parkin present at low levels in HEK293 and SH-SY5Y cells (Fig. 1B).
To assess which portion of parkin is phosphorylated, we generated Myc-tagged N-terminal parkin (aa 2-144) (parN) and C-terminal parkin (aa 294 -465) (parC) constructs ( Fig.  2A) and established their stable transfectants. Stable transfectants from the middle portion of parkin (aa 145-293) could not be established, possibly due to folding difficulties of the polypeptide. Both N-terminal and C-terminal parkin fragments were phosphorylated (Fig. 2B). We also noted in the anti-Myc immunoprecipitates phosphorylated protein bands that were possibly cleavage products derived from full-length parkin (29 -31) and resembled the parN fragments in terms of 32 PO 4 incorporation and band shift. Retarded electrophoretic motility of the phosphorylated bands was evident for N-terminal parkin fragments, whereas such a mobility shift was not observed in parC (Fig. 2B). Thus, parkin is phosphorylated in both N terminus (with mobility shift) and C terminus (without mobility shift).
In order to further prove that parkin is phosphorylated and the observed electrophoretic motility shift of parN is caused by the covalent incorporation of phosphate, parkin immunoprecipitates were dephosphorylated with CIP. For this set of experiments, HEK293 and SH-SY5Y cells were transiently transfected with double-tagged MYC-parkin-V5, MYC-parN-V5, and MYC-parC-V5 (Fig. 3A). Transiently transfected cells were treated with OA and lysed, and the washed Myc-agarose immunoprecipitates were incubated with or without CIP. Then samples were subjected to 10 -15% Tris/glycine gel electro- phoresis, and immunoblots were probed with anti-V5 antibody. MYC-parN-V5 immunoprecipitated from OA-treated cells showed retarded electrophoretic mobility compared with controls (Fig. 3B). Full-length MYC-parkin-V5 also showed slight band retardation after OA treatment. These phosphorylationinduced band shifts were observed both in HEK293 and SH-SY5Y cells, indicating similar phosphorylation patterns in nonneuronal neuronal cells. The phosphorylation-induced band retardations were reversed upon CIP treatment (Fig. 3B), con-firming that phosphorylation in the N terminus of parkin occurs along with a characteristic band shift. Electrophoretic mobility shift caused by the incorporation of covalent phosphate is frequently observed in phosphorylated proteins (e.g. tau (32) and the C-terminal fragment of presenilin-1 (33)). On the other hand, band shifts were hardly observed in parNM and parC, despite the fact that parC was phosphorylated in vivo (Fig. 2). Note that a band shift is not always a consequence of phosphorylation (34).  3, 5, 6, 8, 9, 11, and 12) or Me 2 SO (lanes 1, 4, 7, and 10) was added and cultured for 2 h. Immunoprecipitation (IP) was carried out using MYC-agarose beads. Immunoprecipitates were incubated in the presence (lanes 3, 6, 9, and 12) or absence (lanes 1, 2, 4, 5, 7, 8, 10, and 11) of CIP. Samples were subjected to 10 -20% Tris-glycine gel (Invitrogen) or 15% SDS-PAGE. *, unspecific band. The band shifts of full-length parkin from SH-SY5Y cells are better seen on a short exposure (inset). C, parkin expression levels in each transient transfectant treated with Me 2 SO (lanes 1, 3, 5, and 7) or OA (lanes 2, 4, 6, and 8) were confirmed by immunoblotting (IB) of cell lysates using anti-V5 antibody (lower panels).
Identification of Parkin Phosphorylation Sites-To identify the phosphorylated residues of parkin in OA-treated cells, phosphoamino acid analysis was carried out. Stable transfectants of MYC-parkin in HEK293 cells and SH-SY5Y cells as well as MYC-parN and MYC-parC in HEK293 cells were labeled with [ 32 P]orthophosphate in the presence of OA. Myc immunoprecipitates were subjected to SDS-PAGE and then transferred onto PVDF. Proteins were eluted from the excised bands and hydrolyzed. Phosphoamino acid analysis revealed that parkin was mainly phosphorylated at serine residues in OA-treated HEK293 cells and SH-SY5Y cells (Fig. 4). Some minor threonine phosphorylation was observed, whereas tyrosine phosphorylation was not evident under these conditions.
There are 30 serine residues in parkin, 14 of them in parN and 4 in parC. We carried out site-directed mutagenesis to identify phosphorylated serine sites. Selected serines with high phosphorylation probability identified with the NetPhos 2.0 prediction algorithm were substituted by alanine in order to generate unphosphorylatable forms. Since there are only 4 serines in parC (Ser 296 , Ser 378 , Ser 384 , and Ser 407 ), we mutagenized all of them. These serine-to-alanine mutants in parN and parC as well as wild type with Myc tag at the N terminus and V5 tag at the C terminus were transiently transfected into HEK293 cells, and then cells were labeled with [ 32 P]orthophosphate in the presence or absence of OA. Cell lysates were subjected to immunoprecipitation using Myc-agarose beads. Phosphate incorporation of S101A, S131A, and S136A mutations under the treatment of OA were reduced compared with wild type parN (Fig. 5A). In the case of [S101A]parN, the shifted band was no longer detected. Thus, Ser 101 was found to be the responsible phosphate acceptor for the motility shift of parN. 32 PO 4 incorporation into the lower band was detectable but reduced for the S101A, S131A, and S136A mutants of parN. We also carried out mass spectrometry to determine phosphorylation sites in parN and confirmed that Ser 131 and Ser 136 (Fig. 5, B and C) were unambiguously phosphorylated, with weaker signals for Ser 131 than for Ser 136 . The phosphorylation of Ser 101 was not detected by mass spectrometry analysis, since Lys-C proteolysis of parN could not provide complete coverage of the protein sequence.
In the case of parC, S296A was the only mutant that revealed a small but reproducible reduction of phosphate incorporation compared with wild type parC in the presence of OA (Fig. 5D). No difference of phosphate incorporation between wild type parC and mutants S384A and S407A was detected. However, Ser 378 was clearly identified as phosphorylation site using an alternative assay (see below). Taken together, we discovered that Ser 101 , Ser 131 , and Ser 136 in the parkin N terminus as well as Ser 296 (Fig. 5) and Ser 378 (see Fig. 7B) in the parkin C terminus are phosphorylated.
We analyzed further which kinases were responsible for selective phosphorylation of residues Ser 101 , Ser 131 , Ser 136 , Ser 296 , and Ser 378 . In vitro phosphorylation assays were performed using GST fusion proteins harboring serine-toalanine mutations. Phosphorylation by CK-1 was reduced in GST-[S101A]parN (Fig. 6B) and GST-[S378A]parC (Fig. 6C). We detected slightly reduced phosphorylation of GST-[S101A]parN, GST-[S131A]parN and GST-[S136A]parN by PKA, which indicates that PKA might possibly phosphorylate these serine sites. On the other hand, PKC-mediated phosphate incorporation was not reduced in the serine-to-alanine mutants investigated (Fig. 6, B and C). Thus, PKC was not responsible for phosphorylation of these sites.
We further analyzed whether CK-1, PKA and PKC phosphorylate GST-parN and GST-parC in cell lysates. Fusion proteins were incubated with extracts prepared from HEK293 cells, [␥-32 P]ATP, and OA in the presence or absence of selective inhibitor of CK-1 (hymenialdisine), an inhibitor of PKA (H-89), and an inhibitor of PKC (GF 109203X). PDBu was used to stimulate PKC activity. Consistent with the result from in vitro phosphorylation assays (Fig. 6), we found that GST-parN and GST-parC were phosphorylated by cellular extracts in the absence of inhibitors (Fig. 7, A and B). Both parN and parC phosphorylation was completely inhibited with the CK-1 inhibitor hymenialdisine (Fig. 7, A and B), but not in the presence of the PKA inhibitor H-89. Stimulation of PKC with PDBu enhanced phosphorylation of parN and parC, and this effect was reversed with the PKC inhibitor GF 109203X.
Consistent with the in vitro phosphorylation assays (Fig. 6), each of the serine-to-alanine mutants (S101A, S131A, S136A, S296A, and S378A) showed reduced phosphorylation in cell lysates (Fig. 7, A and B). The incorporation of phosphate was completely abolished when GST-[S378]parN was incubated with HEK293 lysates. Phosphate incorporation into the S296A mutant was also reduced in this assay, but to a lesser extent. Taken together, parkin is phosphorylated by CK-1, PKA, and PKC. CK-1 is a strong candidate kinase for phosphorylation of Ser 101 in the parkin N terminus and Ser 378 in the parkin C terminus (Fig. 7C).
Cellular Modulation of Parkin Phosphorylation-After having identified experimental conditions affecting parkin phosphorylation, we studied how cellular stress would influence parkin phosphorylation. Parkin has been shown to reduce oxidative stress (35) and protect against unfolded protein stress mediated by proteasome inhibitors (19) and ER stress (8,11,18). Thus, we have exposed HEK293 cells stably transfected with MYC-parkin-V5 to hydrogen peroxide, the proteasome inhibitor MG132, and the glycosylation inhibitor tunicamycin, concomitant with stabilization of phosphorylation by OA. As an alternative means to induce ER stress, cells were transfected with PaelR-FLAG, an aggregation-prone ER protein that mediates ER stress upon overexpression (11). Indeed, aggregation of PaelR-FLAG was readily detected on FLAG-probed Western blots (Fig. 8A).
ER stress was confirmed by determination of the co-regulated ER chaperones Grp78 and Grp94 (36,37). Tunicamycin treatment caused induction of GRP78 mRNA, as evidenced by RT-PCR, as well as Grp78 and Grp94 proteins, as evidenced by immunoblotting (Fig. 8B). PaelR overexpression caused an induction of these ER chaperones in HEK293 cells, an effect that was greatly reduced in parkin transfectants (Fig. 8B). Parkin expression also attenuated tunicamycin-induced glucose-regulated protein induction (Fig. 8B), consistent with previous reports that parkin protects cells against ER stress (8,11).
[ 32 P]Orthophosphate incorporation into V5-immunoprecipitated MYC-parkin-V5 was found to be specifically reduced under conditions of protein folding stress. ER stress elicited by tunicamycin treatment and by PaelR overexpression greatly reduced overall parkin phosphorylation (Fig. 8). Proteasomal inhibition with MG-132 reduced parkin phosphorylation, but less than ER stress (Fig. 8). In contrast, oxidative stress mediated by hydrogen peroxide exposure did not affect parkin phosphorylation, although cytotoxicity became apparent after prolonged exposure to 60 M H 2 O 2 . No ER stress was evidenced after H 2 O 2 treatment (Fig. 8B). Thus, protein folding stress specifically reduces parkin phosphorylation.
OA-stabilized Phosphorylation of Parkin Suppresses E3 Activity-In order to examine whether phosphorylation can affect the enzymatic activity of parkin, in vitro ubiquitination assays were performed. Since it has been reported that parkin is autoubiquitinated (7-9), we used parkin itself as a substrate. For this purpose, HEK293T cells were transfected with FLAG-parkin as well as phosphoserine mutants. After OA or control treatment, cell lysates were prepared, and 5. Parkin is phosphorylated at serines 101, 131, 136, and  296 in in vivo phosphorylation assays. A, serines 101, 131, and 136 in MYC-parN-V5 were individually mutagenized to alanine. Each construct was transiently transfected into HEK293 cells and an in vivo phosphorylation assay was carried out with [ 32 P]orthophosphate in the presence or absence of OA. Cells were harvested, and immunoprecipitation using Myc-agarose beads was performed (upper panels). The anti-FLAG immunoprecipitates were added for reconstitution of an in vitro ubiquitination assay. The formation of high molecular weight smears of biotin-ubiquitin in the FLAG immunoprecipitates revealed autoubiquitination of parkin. Overall phosphorylation upon OA treatment of wild-type parkin-transfected cells caused some reduction of parkin autoubiquitination (Fig. 9A).
To provide more quantitative measures of the effect of phosphorylation on parkin activity, we conducted autoubiquitination assays using recombinant GST-parkin phosphorylated in vitro with CK-1, PKA, and PKC. Parkin phosphorylation by these kinases reduced parkin activity (Fig. 9B). In vitro phosphorylation of GST-parkin decreased its autoubiquitination activity by 24 Ϯ 8% (n ϭ 4) in the case of CK-1, by 44 Ϯ 5% (n ϭ 3) in the case of PKA, and by 39 Ϯ 12% (n ϭ 3) in the case of PKC (Fig. 9C). Thus, phosphorylation of parkin appears to down-regulate its ubiquitin ligase activity.
In the attempt to identify individual regulatory phosphorylation sites within parkin, we investigated FLAG-tagged constructs of the phosphorylation site serine-to-alanine and -aspartate mutants identified above. None of the individual phosphorylation sites investigated appeared to exert a unique regulatory role, as evidenced from densitometric quantification of the autoubiquitinated parkin bands (results not shown). Thus, if parkin E3 activity is regulated by phosphorylation, it must arise from multiple sites. DISCUSSION Here we demonstrate that parkin is phosphorylated both in nonneuronal and neuronal cell lines. Parkin appears to be dephosphorylated rapidly under steady state conditions, because the phosphate incorporation was hardly observed with-out stabilization with OA. At least 5 serine residues were identified as phosphorylation sites. The kinases CK-1, PKA, and PKC were found to phosphorylate parkin. In cells exposed to the Parkinson's disease relevant protein folding stress (38), overall parkin phosphorylation decreased. Unphosphorylated parkin tended to be more active. These findings suggest that phosphorylation of parkin contributes to the regulation of its ubiquitin ligase activity upon unfolded protein stress.
Phosphoamino acid analysis revealed that serine sites are mainly phosphorylated in OA-treated cells. Threonine residues may also be phosphorylated in parkin, because a weak signal was clearly detected. Site-directed mutagenesis combined with in vivo and in vitro phosphorylation assays led to the identification of Ser 101 , Ser 131 , Ser 136 , Ser 296 , and Ser 378 as phosphorylation sites in parkin. The corresponding serine-to-alanine mutants showed reduced, but not abolished incorporation of phosphate. Thus, multiple phosphorylation sites exist in parkin.
CK-1, PKA, and PKC were identified as putative parkin kinases. Specifically, CK-1 is one kinase to phosphorylate Ser 101 and Ser 378 , because mutations of these sites to alanine strongly reduced incorporation of phosphate. However, the possibility cannot be excluded that other kinases are involved in phosphorylation at these serine sites. CK-1 is an unexpected kinase to phosphorylate parkin because there is no CK-1 recognition consensus sequence ((D/E)XX(S/T)) in the amino acid sequence of parkin. CK-1 is ubiquitously expressed and involved in various important cellular processes, including signal transduction. We also observed a slightly reduced incorporation of phosphate by PKA in S101A, S131A, and S136A, which means that PKA may contribute to phosphorylation of these FIG. 6. CK-1, PKA, and PKC phosphorylate parkin. A, GST-parkin, GST-parN, and GST-parC were incubated with purified CK-1, CK-2, PKA, PKC, and Akt/PKB1 in the presence of [␥-32 P]ATP. Reaction mixtures were subjected to 12% SDS-PAGE. Phosphorylated fusion proteins were detected by autoradiography (upper panels). Phosvitin, histone, or GSK fusion protein was used as control substrate to confirm activity of CK-1/CK-2, PKA/PKC, or Akt/PKB1, respectively. Equal fusion protein loadings were shown by Coomassie staining (lower panels). B and C, in vitro phosphorylation assays were carried out with CK-1 (left panels) PKA (middle panels), and PKC (right panels) using the indicated GST-parN (B) or GST-parC (C) substrates. Experimental procedures were carried out as described in A. Equal substrate protein loading was confirmed by Coomassie staining (lower panels). sites. PKC is able to phosphorylate parkin but is not responsible for the serine sites investigated here.
Since three identified phosphorylation sites were located in the linker region, and it has been reported that the ubiquitin-like domain interacts with Rpn10, a subunit of 19 S in proteasome (39), we also investigated whether parkin phosphorylation could affect proteasome activities. Each serine-to-alanine or -aspartate mutants still had almost the same level of activities as vector controls (data not shown). The RING-IBR-RING motif of parkin is important to interact with E2 co-enzymes. However, S296A and S378A did not consistently show reduced levels of ubiquitination by in vitro ubiquitination assays (data not shown). Although the RING-IBR-RING motif is crucial for parkin ubiquitin ligase function, single site phosphorylation in this domain appears to have no effect on autoubiquitination. Nevertheless, OAmediated overall phosphorylation of parkin slightly reduced its E3 enzymatic activity. The regulation of parkin E3 activity must be due to multiple phosphorylation sites.
ER stress (but not oxidative stress) was found to specifically reduce parkin phosphorylation levels. Specifically, we found that OA-stabilized phosphorylation of parkin or phosphorylation by identified parkin kinases caused a small but significant reduction of parkin autoubiquitination activity. More in vivo work is needed to elucidate if and how parkin phosphorylation affects the activity and recognition of the various substrates of the E3 ubiquitin ligase parkin. In fact, some of the polyubiquitin signals detected in the in vitro E3 assay (Fig. 9) may arise from ubiquitination of co-purified, parkin-associated ubiquitin ligase substrates.
Because CK-1 appeared to be a major parkin kinase (Fig. 7), further investigations of signal transduction events involving CK-1 might be particularly revealing for the regulation of parkin E3 ubiquitin ligase activity. More generally, the involvement of parkin phosphorylation in the ER unfolded protein stress response (Fig. 8) might contribute to the understanding of parkin as dopaminergic neuron survival factor.
Taken together, we suggest that the reduced phosphorylation of parkin in ER stressed cells contributes to the up-regulation of parkin E3 ubiquitin ligase activity, which is believed to suppress cytotoxicity due to unfolded protein stress. FIG. 9. Phosphorylation down-regulates parkin autoubiquitination activity. A, cellular extracts from transfected HEK293T cells in the presence or absence of OA and Me 2 SO were subjected to immunoprecipitation (IP) using FLAG M2-agarose. Immunoprecipitates were incubated with E1, E2 (UbcH7), and biotinubiquitin at 30°C for 90 min reaction (rxn) time (upper panel). Reaction mixtures were subjected to 10% SDS-PAGE and transferred onto PVDF membrane. Ubiquitinated parkin proteins were analyzed by immunoblot (IB) with streptavidin-peroxidase polymer (upper panel). Total lysates were analyzed by immunoblot with anti-FLAG antibody (lower panel). B, GST-parkin immobilized on glutathione-Sepharose was phosphorylated by CK-1, PKA, or PKC. The same experimental procedure was carried out as described in A. C, signal intensities of the streptavidin-binding biotinylated ubiquitin smears generated in experiments performed as described in B were quantified by densitometric scanning and expressed as a percentage of unphosphorylated parkin activity. Error bars, S.E. of four (CK-1) and three (PKA and PKC) experiments, respectively.