RING finger ubiquitin-protein isopeptide ligase Nrdp1/FLRF regulates parkin stability and activity.

Parkin is a ubiquitin-protein isopeptide ligase. It has been suggested that loss of function in parkin causes accumulation and aggregation of its substrates, leading to death of dopaminergic neurons in Parkinson disease. Using the yeast two-hybrid screen, we isolated a RING finger protein that interacted with the N terminus of parkin in a Drosophila cDNA library. Interaction between human parkin and the mammalian RING finger protein homologue Nrdp1/FLRF, a ubiquitin-protein isopeptide ligase that ubiquitinates ErbB3 and ErbB4, was validated by in vitro binding assay, co-immunoprecipitation, and immunofluorescence co-localization. Significantly, pulse-chase experiments showed that cotransfection of Nrdp1 and parkin reduced the half-life of parkin from 5 to 2.5 h. Consistent with these findings, we further observed that degradation of CDCrel-1, a parkin substrate, was facilitated by overexpression of parkin protein. However, co-transfection of Nrdp1 with parkin reversed the effects of parkin on CDCrel-1 degradation. We conclude that Nrdp1 is a parkin modifier that accelerates degradation of parkin, resulting in a reduction of parkin activity.

Parkinson disease (PD) 1 is the second most common neurodegenerative disorder after Alzheimer disease with ϳ500,000 patients in the United States alone. PD patients experience slowness of movement, rigidity, tremor, difficulty with balance, and variable manifestation of dementia. The main pathological features of PD are the loss of the dopaminergic neurons in the substantia nigra and the presence of abnormal protein aggregates that form filamentous inclusions in neuronal cytoplasm, termed Lewy bodies or Lewy neurites (nerve fibers) in PD brains (1)(2)(3)(4). Degeneration of dopaminergic nigral neurons leads to loss of dopaminergic projections to the striatum, which represents the primary defect of neurochemical pathways in PD. Immunohistochemical examination of Lewy bodies showed that Lewy bodies contain many different proteins, including neurofilament, ubiquitin (5)(6)(7), and more recently identified synuclein and parkin (8 -11).
The majority of PD is sporadic. Rare familial forms of PD could be either autosomal recessive or autosomal dominant, suggesting that the etiology of the disorder can be complicated (3,4,8). The discoveries of genetic linkages for PD to several loci provide promises to identify mutations in ubiquitin C-terminal hydrolase (UCH)-L1, ␣-synuclein, DJ-1, and Pink1 (12)(13)(14)(15)(16)(17). Autosomal recessive juvenile Parkinsonism (AR-JP) was mapped to the long arm of chromosome 6 (6q25.2-q27) and is linked strongly to the markers D6S305 and D6S253 (18). D6S305 is deleted in one Japanese AR-JP patient (19). Using the positional cloning strategy combined with the exon-trapping technology and cDNA library screening, Kitada et al. (20) identified a gene named parkin in which exons 3-7 were deleted from this Japanese patient. They also described four other AR-JP patients from three unrelated families with a deletion of exon 4 in the parkin gene, confirming that mutations in the parkin gene appear to be responsible for the pathogenesis of AR-JP. Although most mutations in the parkin gene are thought to inactivate the gene with exon deletions (20 -25), missense point mutations were also identified from some of AR-JP patients (23,26,27) further confirming the involvement of parkin in the pathogenesis of PD.
Parkin is one of the largest genes in the human genome. It spans more than 1.5 megabases and has 12 exons and huge introns. A 4.5-kilobase transcript is expressed in many human tissues and is abundant in the brain. The parkin gene encodes a protein of 465 amino acids with a molecular mass of about 52 kDa (20). The deduced amino acid sequence of parkin showed similarity to ubiquitin at the N terminus. The C terminus of parkin contains two RING finger motifs and in between RING finger domain. Further studies demonstrate that parkin possesses E3 ligase activity (28) and ubiquitinates at least eight substrates including ␣-synuclein, Pael receptor, CDCrel-1, tubulin, synphilin-1, synaptotagmin, cyclin E, and P38 (29 -36). In PD patients, it is presumed that the ability of parkin to ubiquitinate these substrates is diminished because of mutations or deletions in the gene, leading to their accumulation and aggregation that result in the death of nigral neurons. To identify additional potential factors that are involved in PD, we used the yeast two-hybrid screen. We report here that parkin interacts with a RING finger domain protein FLRF/Nrdp1 (37). FLRF/Nrdp1 is also an ubiquitin E3 ligase that ubiquitinates neuregulin receptors ErbB3 and ErbB4 (38,39). Our data demonstrate that FLRF/Nrdp1 promotes degradation of parkin protein and stabilizes a parkin substrate CDCrel-1. Our findings suggest that parkin and Nrdp1/FLRF may form a complex that regulate activities of E3 ligases and thereby the metabolisms of their substrates.

EXPERIMENTAL PROCEDURES
The Yeast Two-hybrid Screen-The N terminus of the Drosophila parkin (dparkin) cDNA was amplified by PCR and cloned into pGKBT7 to encode a hybrid protein containing the DNA-binding domain of Gal4. Expression of the chimeras in yeast was determined by Western blot analysis. Yeast cells of the AH109 reporter strain were sequentially transformed with the dparkin/pGKBT7 construct and then with a Drosophila expression library containing cDNAs fused to sequence of the GAL4 trans-activation domain in pACT2 vector (Clontech). Transformants were selected on S.D.-Ade/-His/-Leu/-Trp selection medium. Positive clones were tested for ␤-galactosidase and ␣-galactosidase activities.
Expression Vectors and Antibodies-For mammalian expression, hparkin is cloned into pcDNA 3.1 with a Myc tag at its N terminus, whereas N-terminal HA-tagged Nrdp1 was generated in pcDNA 3.0. Nrdp1-FLAG and Nrdp1-CT-FLAG were obtained from Dr. Alfred Goldberg, Harvard Medical School. FLAG-tagged CDCrel-1 was from Dr. W. S. Trimble, Hospital for Sick Children, Toronto, Canada. Glutathione S-transferase (GST) or GST-Nrdp1 was produced from constructs in pGEX-4T-1 vector. Anti-HA antibody 12CA5 is from Roche Applied Science. Anti-tubulin antibody is from Oncogene. Anti-FLAG M2 antibody and fluorescence-conjugated anti-Myc and anti-FLAG antibodies are from Sigma. Anti-Myc 9E10 is from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-parkin antibody was purchased from Cell Signaling Technology, Inc. (Beverly, MA).
In Vitro Interactions-For in vitro binding assay, Myc-hparkin, HA-Nrdp1, and luciferase peptides labeled with [ 35 S]methionine were generated using the Promega TNT kit and mixed and rocked in binding buffer (10 mM Tris-HCl, pH 8.0, 200 mM NaCl, 5 mM EDTA, 0.5% Nonidet P-40, 1 mM dithiothreitol, 3 mg/ml bovine serum albumin, and the proteinase inhibitors). Myc or HA antibody was employed to precipitate either Myc-hparkin or HA-Nrdp1, and the co-immunoprecipitated Nrdp1 or hparkin was separated by SDS-PAGE and then analyzed by a PhosphorImager. In a separate experiment, bacterially expressed GST-Nrdp1 was produced and purified. Glutathione-Sepharose beads bound to GST-fusion proteins were washed with binding buffer and rocked with aliquots of in vitro translated [ 35 S]methioninelabeled hparkin proteins for 1 h at 4°C in binding buffer. The beads were washed with binding buffer and then boiled in SDS-PAGE sample buffer. The proteins binding to GST-Nrdp1 beads were resolved on SDS-PAGE and detected by a PhosphorImager. Domains of hparkin that bind to Nrdp1 were mapped with hparkin C-terminal or N-terminal TNT-translated peptides.
Western Blot Analysis, Pulse-Chase, and Co-immunoprecipitation-Epitope-tagged cDNAs (HA-Nrdp1 or Nrdp1-FLAG (38) and Myc-hparkin or FLAG-CDCrel-1) were co-transfected into C33A cells. Western blot analysis was carried out by using anti-Myc 9E10, anti-HA 12CA5, anti-FLAG M2, or anti-parkin antibodies. For co-immunoprecipitation, extracts were first reacted with antibody to the epitope of hparkin (9E10) or of Nrdp1 (HA, 12CA5) and after the complexes were washed HA-Nrdp1 or Myc-hparkin protein was detected in immunoprecipitates by Western blot analysis with 12CA5 or 9E10 antibodies. For pulsechase analysis, two methods were used. (a) Cells were chased with 12 g/ml of cycloheximide. CDCrel-1 was analyzed by Western blot using anti-FLAG M2 antibody. (b) Cells were labeled with [ 35 S]methionine and chased up to 16 h. Myc-hparkin or FLAG-CDCrel-1 was immunoprecipitated by using anti-Myc 9E10 or anti-FLAG M2 antibodies. Precipitated proteins were separated on SDS-PAGE and quantitated by a PhosphorImager. For quantification, three independent experiments were performed.
Immunofluorescence Co-localization-Human C33A cells were cultured on glass cover slips and co-transfected with Myc-hparkin and Nrdp1-FLAG. After 48 h, cells were fixed in 4% of paraformaldehyde for 20 min, washed three times with phosphate-buffered saline, permeabilized in 0.1% of Triton X-100 in phosphate-buffered saline for 20 min, and blocked with 3% milk and 3% goat serum in phosphate-buffered saline for 1 h. The cells were incubated with primary antibodies against either Myc or FLAG that had been fluorescence-conjugated (Sigma). In a separate experiment, transfected PC12 cells were treated with 50 ng/ml of nerve growth factor for 7 days, fixed, and stained with anti-Myc and anti-FLAG antibodies or anti-parkin antibody. The localization of proteins was visualized by a fluorescence microscopy.

RESULTS
To identify potential parkin substrates or modifiers, we employed the yeast two-hybrid screen. The N terminus of Drosophila parkin cDNA (1-576 bps) was cloned into pGKBT7 vector (Matchmaker III from Clontech) and verified by sequencing and Western blot analysis (data not shown). After sequential transformation of the pGBKT7 dparkin bait and the Drosophila pACT2 cDNA library, three clones that activated all three GAL4-responsive markers His3, Mel1, and LacZ reporter genes in AH109 were isolated. Sequence analysis revealed that one of these proteins is the homologue of mammalian Nrdp1/FLRF, a RING finger E3 ligase (37) that ubiquitinates neuregulin receptors ErbB3 and ErbB4 (38,39). The Drosophila protein, here referred to as dNrdp1, interacted specifically with dparkin in a yeast two-hybrid mating assay. dNrdp1 has an open reading frame of 951 base pairs, encoding a 317 amino acid protein. Nrdp1s between human and Drosophila share a high 57% identity and 78% similarity, indicating a structural and possibly functional conservation (37).
To validate whether the interaction also occurs with the human proteins, in vitro binding assays were carried out. [ 35 S]Methionine-labeled HA-Nrdp1, Myc-hparkin, and the luciferase control were generated using coupled in vitro transcription/translation TNT kit. In vitro immunoprecipitation analysis was performed from binding mixtures of hparkin and Nrdp1 or controls. Co-precipitated proteins were then separated by SDS-PAGE and analyzed using a PhosphorImager. The Nrdp1 band was clearly observed from hparkin precipitates, whereas hparkin was pulled down by HA-Nrdp1 (Fig.  1A). Similar results were obtained when FLAG was tagged to the Nrdp1 C terminus (data not shown). An in vitro interaction between hparkin and Nrdp1 was also demonstrated by mixing equivalent aliquots of the [ 35 S]-labeled hparkin with purified GST or GST-Nrdp1 fusion protein (Fig. 1B). GST or GST-Nrdp1 fusion proteins were adsorbed to glutathione-agarose beads. In vitro translated hparkin was specifically retained on the beads by the GST-Nrdp1 fusion protein but not parental GST peptide, indicating specific binding between hparkin and Nrdp1 (Fig.  1B). Further experiment by in vitro binding assays mapped the binding between hparkin and Nrdp1 to the hparkin N terminus (hp-NT) (Fig. 1B), consistent with the results from the yeast two-hybrid assays.
To investigate interactions between parkin and Nrdp1 in vivo, Myc-hparkin and HA-Nrdp1 were co-transfected into C33A cells for co-immunoprecipitation analysis. To minimize degradation of parkin by Nrdp1, we used N-terminal-tagged HA-Nrdp1 (38). Anti-HA (12CA5) antibody was used to precipitate Nrdp1 protein from C33A cell lysates. Co-immunoprecipitated hparkin was detected from cells co-transfected with Mychparkin and HA-Nrdp1 but not from control cells that were co-transfected with Myc-hparkin and a HA-vector (Fig. 2B). In a separate experiment, we employed anti-Myc 9E10 antibody to co-immunoprecipitate Myc-hparkin and HA-Nrdp1. The negative control cells transfected with Myc-vector and HA-Nrdp1 showed no detectable HA-Nrdp1 protein pulled down, whereas a clear band of HA-Nrdp1 was detected from the anti-Myc precipitates in cells co-transfected with HA-Nrdp1 and Mychparkin, demonstrating interactions between these two proteins endogenously ( Fig. 2A). To further characterize interactions between parkin and Nrdp1, C33A and PC12 cells were cultured on glass cover slips and co-transfected with Mychparkin and Nrdp1-FLAG. Immunofluorescence analysis was carried out with fluorescence-conjugated anti-Myc and anti-FLAG antibodies or anti-parkin antibody and then a fluorescence-labeled secondary antibody in cells treated or untreated with nerve growth factor. The co-localization of hparkin and Nrdp1 was visualized in cell bodies as well as neuronal processes (Fig. 3).
Because both parkin and Nrdp1 are E3 ligases (28, 31, 32, 34, 38, 39), next we tested whether Nrdp1 or parkin affects their stability or activity. We performed a pulse-chase experiment. C33A cells that were co-transfected with plasmids were metabolically labeled with [ 35 S]methionine for 3 h. [ 35 S]-labeled proteins were then chased for 0, 1.5, 3, 5, or up to 16 h in cells co-transfected with Myc-hparkin and Nrdp1-FLAG, Nrdp1-CT-FLAG or a control. Myc-hparkin was immunoprecipitated using 9E10 antibody and analyzed on SDS-PAGE gels and then a PhosphorImager (Fig. 4A). Expression of Myc-parkin was quantitated (Fig. 4B). Our results show that the half-life of hparkin is about 5 h in the absence of Nrdp1 in C33A cells consistent with the previous reports (31), whereas overexpression of Nrdp1 reduces parkin half-life to 2.5 h (Fig. 4B), indicating that Nrdp1 accelerates parkin degradation. However, in contrast to the previous report (38), the C terminus of Nrdp1 does not have dominant effects on degradation of parkin. Significantly, we also demonstrated by pulse-chase assays that accelerated degradation of hparkin downstream substrate CD-Crel-1 was abrogated with Nrdp1 overexpression (Fig. 5). In this experiment, plasmids of FLAG-CDCrel-1, Myc-hparkin, and Nrdp1-FLAG were transfected in C33A cells individually or in combination. Cells were labeled with [ 35 S]methionine and then chased up to 16 h. FLAG-CDCrel-1 was immunoprecipitated using anti-FLAG antibody M2, resolved on SDS-PAGE gels, and then quantitated in a PhosphorImager (Fig. 5, A and   B). In the absence of exogenous parkin expression, the half-life of CDCrel-1 is ϳ8 h, whereas overexpression of parkin accelerated CDCrel-1 turnover with a half-life of 5 h. However, CDCrel-1 was re-stabilized when both hparkin and Nrdp1 were co-transfected (Fig. 5, A and B), indicating that Nrdp1 affected CDCrel-1 turnover not directly but rather via modulating parkin activity. Similar results were obtained in C33A cells when cycloheximide was used in co-transfected C33A cells (Fig. 5, C  and D). We conclude that Nrdp1 not only affects hparkin stability but also activity on its downstream substrates. DISCUSSION Models for parkin and its substrates synuclein and Pael-R have been generated (40 -42). Although phenotypes in mice models have been documented as mild (40,41,43), it has been shown that manifestations in Drosophila are more dramatic (42,44). Knock-out of parkin in flies induces mitochondria defects that mimic pathological phenotypes of PD patients (44). The down-regulation of parkin expression in Drosophila that overexpresses synuclein or Pael-R results in more significant overexpressed Nrdp1 or suppressed expression of Nrdp1. Regulation on parkin expression as well as on biological functions of parkin by Nrdp1 can then be investigated by crossing these animals with Drosophila models for parkin, synuclein, and Pael-R.
One of the most important questions about our findings is whether Nrdp1 plays a role in the pathogenesis of Parkinson disease. We have demonstrated that overexpression of Nrdp1 reduces the parkin half-life from 5 to 2.5 h. It is reasonable to presume that increased Nrdp1 activity in brain would significantly reduce the level of parkin protein, resulting in accumulation of parkin substrates (Fig. 6). Although there is no direct evidence that the Nrdp1 locus on chromosome 12q13 (37) is linked to familiar PDs, our results in this report and the fact that significant reduction of parkin results in more death of dopaminergic neurons in Drosophila with overexpression of ␣-synuclein or Pael-R (45) raise a possibility that Nrdp1 may modify phenotypes of Parkinson disease by reducing the level of parkin protein. It is also possible that Nrdp1 is directly involved in the pathogenesis of the disease.
Parkin contains two RING domains and is a ubiquitin E3 ligase. Parkin ubiquitinates its substrates and facilitates their degradation. It is widely suggested that loss of function in parkin causes accumulation and aggregation of these substrates, resulting in death of dopaminergic neurons. However, recent studies in mice with parkin knock-out demonstrated that absence of parkin protein has no effects on stability of its substrates, namely CDCrel-1, Pael-R, and synuclein (43). In Drosophila, overexpression of parkin reduced severity of phenotypes caused by overexpression of synuclein and Pael-R by unknown mechanisms other than by simply reduction of its substrate proteins (45). Furthermore, no or few Lewy bodies or Lewy neurites were observed in PD patients with parkin mutations although AR-JP patients with mutations in the parkin show a severe loss of DA neurons. All these results indicate that besides promoting degradation of its substrates, parkin may have other cellular functions in cell death. Under this assumption, we hypothesize that regulation of parkin and Nrdp1 complexes may play significant roles in the pathogenesis of PD by modulating signal pathways including the epidermal growth factor pathway (ErbB receptors) (Fig. 6). Therefore, it would be imperative to examine expression levels as well as distribution of both parkin and Nrdp1 proteins in PD and control brains.
FIG. 6. Proposed model for the role of interaction between parkin and Nrdp1 in neurodegeneration. Increased activity of Nrdp1 (ϩNrdp1) reduces the level of parkin protein, resulting in accumulation of parkin substrates and cell death. Decreased activity of Nrdp1 (ϪNrdp1) may protect cells by stabilizing parkin protein. On the other hand, parkin may regulate signal pathways that are involved in cell death by modulating Nrdp1 activity on its substrates such as ErbB3 or ErbB4.