Physical and Functional Interaction between Dorfin and Valosin-containing Protein That Are Colocalized in Ubiquitylated Inclusions in Neurodegenerative Disorders*

Dorfin, a RING-IBR type ubiquitin ligase (E3), can ubiquitylate mutant superoxide dismutase 1, the causative gene of familial amyotrophic lateral sclerosis (ALS). Dorfin is located in ubiquitylated inclusions (UBIs) in various neurodegenerative disorders, such as ALS and Parkinson's disease (PD). Here we report that Valosin-containing protein (VCP) directly binds to Dorfin and that VCP ATPase activity profoundly contributes to the E3 activity of Dorfin. High through-put analysis using mass spectrometry identified VCP as a candidate of Dorfin-associated protein. Glycerol gradient centrifugation analysis showed that endogenous Dorfin consisted of a 400–600-kDa complex and was co-immunoprecipitated with endogenous VCP. In vitro experiments showed that Dorfin interacted directly with VCP through its C-terminal region. These two proteins were colocalized in aggresomes in HEK293 cells and UBIs in the affected neurons of ALS and PD. VCPK524A, a dominant negative form of VCP, reduced the E3 activity of Dorfin against mutant superoxide dismutase 1, whereas it had no effect on the autoubiquitylation of Parkin. Our results indicate that VCPs functionally regulate Dorfin through direct interaction and that their functional interplay may be related to the process of UBI formation in neurodegenerative disorders, such as ALS or PD.

Amyotrophic lateral sclerosis (ALS) 1 is one of the most common neurodegenerative disorders, characterized by selective motor neuron degeneration in the spinal cord, brain stem, and cortex. Two genes, CuZn-superoxide dismutase (SOD1) and amyotrophic lateral sclerosis 2 have been identified as responsible genes for familial forms of ALS. Using mutant SOD1 transgenic mice, the pathogenesis of ALS has been partially uncovered. The proposed mechanisms of the motor neuron degeneration in ALS include oxidative toxicity, glutamate receptor abnormality, ubiquitin proteasome dysfunction, inflammatory and cytokine activation, dysfunction of neurotrophic factors, damage to mitochondria, cytoskeletal abnormalities, and activation of the apoptosis pathway (1,2).
In a previous study (3), we identified several ALS-associated genes using molecular indexing. Dorfin was identified as one of the up-regulated genes in ALS, which contains a RING-IBR (in between ring finger) domain at its N terminus and mediated ubiquitin ligase (E3) activity (3,4). Dorfin colocalized with Vimentin at the centrosome after treatment with a proteasome inhibitor in cultured cells (4). Dorfin physically bound and ubiquitylated various SOD1 mutants derived from familial ALS patients and enhanced their degradation, but it had no effect on the stability of wild-type SOD1 (5). Overexpression of Dorfin protected neural cells against the toxic effects of mutant SOD1 and reduced SOD1 inclusions (5).
Recent findings indicate that the ubiquitin-proteasome system is widely involved in the pathogenesis of Parkinson's disease (PD), Alzheimer's disease, polyglutamine disease, and Prion diseases as well as ALS (6). From this point of view, we previously analyzed the pathological features of Dorfin in various neurodegenerative diseases and found that Dorfin was predominantly localized not only in Lewy body (LB)-like inclusions in ALS but also in LBs in PD, dementia with Lewy bodies, and glial cell inclusions in multiple system atrophy (7). These characteristic intracellular inclusions composed of aggregated, ubiquitylated proteins surrounded by disorganized filaments are the histopathological hallmark of aging-related neurodegenerative diseases (8).
A structure called aggresome by Johnston et al. (9) is formed when the cell capacity to degrade misfolded proteins is exceeded. The aggresome has been defined as a pericentriolar, membrane-free, cytoplasmic inclusion containing misfolded ubiquitylated protein ensheathed in a cage of intermediate filaments, such as Vimentin (9). The formation of the aggresome mimics that of ubiquitylated inclusions (UBIs) in the affected neurons of various neurodegenerative diseases (10). Combined with the fact that Dorfin was localized in aggresomes in cultured cells and UBIs in ALS and other neurode-generative diseases, these observations suggest that Dorfin may have a significant role in the quality control system in the cell. The present study was designed to obtain further clues for the pathophysiological roles of Dorfin. For this purpose, we screened Dorfin-associated proteins using high performance liquid chromatography coupled to electrospray tandem mass spectrometry (LC-MS/MS). The results showed that Valosincontaining protein (VCP), also called p97 or Cdc48 homologue, obtained from the screening, physically and functionally interacted with Dorfin. Furthermore, both Dorfin and VCP proteins colocalized in aggresomes of the cultured cells and in UBIs in various neurodegenerative diseases.
Cell Culture and Transfection-All media and reagents for cell culture were purchased from Invitrogen. HEK293 cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal calf serum, 5 units/ml penicillin, and 50 g/ml streptomycin. HEK293 cells at subconfluence were transfected with the indicated plasmids using Fu-GENE6 reagent (Roche Applied Science). To inhibit cellular proteasome activity, cells were treated with 1 M MG132 (benzyloxycarbonyl-Leu-Leu-Leu-al; Sigma) for 16 h after overnight post-transfection. Cells were analyzed at 24 -48 h after transfection.
Protein Identification by LC-MS/MS Analysis-FLAG-Dorfin WT was expressed in HEK293 cells (semiconfluent in a 10-cm dish) and then immunoprecipitated by anti-FLAG antibody. The immunoprecipitates were eluted with a FLAG peptide and then digested with Lys-C endopeptidase (Achromobacter protease I). The resulting peptides were analyzed using a nanoscale LC-MS/MS system as described previously (16). The peptide mixture was applied to a Mightysil-PR-18 (1-m particle, Kanto Chemical Corp., Tokyo) column (45 ϫ 0.150 mm ID) and separated using a 0 -40% gradient of acetonitrile containing 0.1% formic acid over 30 min at a flow rate of 50 nl/min. Eluted peptides were sprayed directly into a quadrupole time-of-flight hybrid mass spectrometer (Q-Tof Ultima; Micromass, Manchester, UK). MS and MS/MS spectra were obtained in data-dependent mode. Up to four precursor ions above an intensity threshold of 10 cps were selected for MS/MS analysis from each survey scan. All MS/MS spectra were searched against protein sequences of Swiss Prot and RefSeq (NCBI) using batch processes of the Mascot software package (Matrix Science, London, UK). The criteria for match acceptance were the following: 1) when the match score was 10 over each threshold, identification was accepted without further consideration; 2) when the difference of score and threshold was lower than 10 or when proteins were identified based on a single matched MS/MS spectrum, we manually confirmed the raw data prior to acceptance; 3) peptides assigned by less than three y series ions and peptides with ϩ4 charge state were all eliminated regardless of their scores.
Recombinant Proteins and Pull-down Assay-We used pMALp2 (New England BioLabs) and pMALp2T (Factor Xa cleavage site of pMALp2 was replaced with a thrombin recognition site) to express fusion proteins with MBP. To produce the full-length (residues 1-838) Dorfin (MBP-Dorfin full ), N-terminal (residues 1-367) Dorfin (MBP-Dorfin N ), and C-terminal (residues 368 -838) Dorfin (MBP-Dorfin C ), the PCR fragments were amplified from pcDNA4/HisMax-Dorfin (4) by using the appropriate PCR primers with restriction sites (FbaI and HindIII) and then ligated into pMAL-p2 vectors. To produce the MBP-Parkin protein, full-length PARKIN cDNA was inserted into the EcoRI-NotI sites of pMALp2T. All of the MBP-tagged recombinant proteins were purified from Escherichia coli BL21-codon-plus. The detail of the purification method of MBP-tagged proteins was described previously (17). Recombinant GST fusion VCP WT and VCP K524A proteins were also generated from E. coli lysate and purified with glutathione-Sepharose. Recombinant His-VCP WT and His-VCP K524A proteins were purified from insect cells using baculovirus. The detail of purification of these recombinant VCP proteins was described previously (15). Binding experiments were performed with proteins carrying different tags. His-or GST-VCP were mixed with MBP fusion proteins: MBP-Dorfin full , -Dorfin N , -Dorfin C , -Parkin, and -mock. His-VCP and GST-VCP proteins were precipitated by Ni 2ϩ -nitrilotriacetic acid-agarose (Qiagen, Valencia, CA), and glutathione-Sepharose (Amersham Biosciences), respectively. Binding was performed with 1-3 g of each protein in 300 l of binding buffer (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 5 mM MgCl 2 , 10% glycerol, 0.5 mg/ml bovine serum albumin, 1 mM dithiothreitol) for 1 h at 4°C. Then 15 l of beads were added and incubated for 30 min. The beads were washed by binding buffer three times and eluted with sample buffer and analyzed by SDS-PAGE followed by Western blotting using specific antibodies.
Glycerol Gradient Centrifugation-Cultured cells or mouse tissues were homogenized in 1 ml of PBS with protease inhibitor (Complete Mini; Roche Applied Science). Supernatants (1 mg of protein for cultured cells, 5 mg of protein for mouse tissues, and 0.1 mg of recombinant His-VCP protein) were used as the samples after 10,000 ϫ g centrifugation for 20 min. The samples (1.0 ml) were loaded on top of a 34-ml linear gradient of glycerol (10 -40%) prepared in 25 mM Tris-HCl buffer, pH 7.5, containing 1 mM dithiothreitol in 40 PA centrifuge tubes (Hitachi, Tokyo), and centrifuged at 4°C and 80,000 ϫ g for 22 h using a Himac CP100␣ centrifuge system (Hitachi). Thirty fractions were collected from the top of the tubes. Two hundred l of each fraction was precipitated with acetone, and the remaining pellet was lysed with 50 l of sample buffer and then used for SDS-PAGE followed by Western blotting.
Immunological Analysis-Cells (4 ϫ 10 5 in a 6-cm dish) were lysed with 500 l of lysis buffer (50 mM Tris-HCl, 150 mM NaCl, 1% Nonidet P-40, and 1 mM EDTA) with protease inhibitor mixture (Complete Mini) 24 -48 h after transfection. The lysate was then centrifuged at 10,000 ϫ g for 10 min at 4°C to remove debris. A 10% volume of the supernatants was used as the "lysate" for SDS-PAGE. When immunoprecipitated, the supernatants were precleared with protein A-Sepharose (Amersham Biosciences), and specific antibodies, anti-FLAG (M2), anti-Myc (9E10), or anti-Dorfin (Dorfin-30) were then added and then incubated at 4°C with rotation. Immune complexes were then incubated with protein A-Sepharose for 3 h, collected by centrifugation, and washed four times with the lysis buffer. For protein analysis, immune complexes were dissociated by heating in SDS-PAGE sample buffer and loaded onto SDS-PAGE. The samples were separated by SDS-PAGE (12% gel or 4 -12% gradient gel) and transferred onto a polyvinylidene difluoride membrane. Finally, Western blotting was performed with specific antibodies.
Immunohistochemistry-HEK293 cells grown on glass coverslips were fixed in 4% paraformaldehyde in PBS for 15 min. Then the cells were blocked for 30 min with 5% (v/v) normal goat serum in PBS, incubated for 1 h at 37°C with anti-HA antibody (12CA5), washed with PBS, and incubated for 30 min with Alexa 496-nm anti-mouse antibodies (Molecular Probes, Inc., Eugene, OR). The coverslips were washed and mounted on slides. Fluorescence images were obtained using a fluorescence microscope (DMIRE2; Leica, Bannockburn, IL) equipped with a cooled charge-coupled device camera (CTR MIC; Leica). Pictures were taken using Leica Qfluoro software.
Pathological Studies-Pathological studies were carried out on 10% formalin-fixed, paraffin-embedded spinal cords and brain stems filed in the Department of Neurology, Nagoya University Graduate School of Medicine. The specimens were obtained at autopsy from three sporadic cases of ALS and four sporadic PD patients. The spinal cord and brain stem specimens of these ALS and PD cases were immunohistochemically stained with antibodies against Dorfin (Dorfin-41) and VCP. Dou-ble staining of identical sections was performed as described previously (7). In immunofluorescence microscopy, Alexa-488-and Alexa-546-conjugated secondary antibodies (Molecular Probes) were used. All human and animal studies described in this report were approved by the appropriate Ethics Review Committees of the Nagoya University Graduate School of Medicine.

RESULTS
Identification of Dorfin-associated Protein in the Cells-In an effort to identify protein(s) that physically interacts with Dor-fin in the cells, FLAG-Dorfin was expressed in HEK293 cells and then immunoprecipitated by anti-FLAG antibody. The immunoprecipitates were eluted with a FLAG peptide and then digested with Lys-C endopeptidase (Achromobacter protease I), and the cleaved fragments were directly analyzed using a highly sensitive "direct nanoflow LC-MS/MS" system as described under "Materials and Methods." Following data base search, a total of 13 peptides were assigned to MS/MS spectra obtained from the LC-MS/MS analyses for the FLAG-Dorfin-

FIG. 1. In vivo interaction between Dorfin and VCP.
A, FLAG-Dorfin and HA-VCP are co-expressed in HEK293 cells. FLAG-mock vector was used as a negative control. The amounts of HA-VCP in 10% of the lysate used are shown (Lysate); the rest was subjected to immunoprecipitation (IP) with anti-FLAG (M2) antibody. Following immunoblotting (IB) with anti-HA (12CA5) antibody revealed that HA-VCP was coimmunoprecipitated with FLAG-Dorfin. B, 5 mg of protein of various mouse tissues (brain, liver, kidney, and muscle) and 1 mg of protein of cultured cells (HEK293, HeLa, and Neuro2a) were each homogenized in 1 ml of PBS. Supernatants were fractionated by 10 -40% glycerol gradient centrifugation followed by separation into 30 fractions using a fraction collector. Immunoblotting using anti-Dorfin, anti-VCP, and anti-Parkin antibodies was performed on the fractions (Fr.), including fractions 2-17. Endogenous Dorfin was co-sedimented with VCP in the fractions with a molecular mass of around 400 -600 kDa. The positions of co-migrated molecular mass markers are indicated above the panels. C, immunoprecipitation with polyclonal anti-Dorfin antibody (anti-Dorfin-30) was performed on fractions 11-14 collected by glycerol gradient centrifugation analysis, where endogenous Dorfin was seen in B. As a negative control, immunoprecipitation with nonimmune rabbit IgG was used on the same fractions. associated complexes. These peptide data identified nine proteins as candidates for Dorfin-associated proteins. One of these identified proteins was VCP that has been proposed to have multiple functions, such as membrane fusion or endoplasmic reticulum-associated degradation (ERAD) (18 -22). In the next step, we examined the relationship between Dorfin and VCP, because the latter has been reported to be linked to various aspects of neurodegeneration (15).
Dorfin Interacts with VCP in Vivo-To verify the interaction between Dorfin and VCP, FLAG-Dorfin and HA-VCP were transiently overexpressed in HEK 293 cells. Immunological analyses revealed that HA-VCP was co-immunoprecipitated with FLAG-Dorfin but not with FLAG-mock (Fig. 1A), confirming their physical interactions in the cells. To determine whether endogenous Dorfin forms a complex, the lysate from mouse brain homogenate was fractioned by glycerol density gradient centrifugation. Each fraction was immunoblotted with anti-Dorfin antibody. The majority of endogenous Dorfin was co-sedimented with VCP around a size of 400 -600 kDa, although endogenous Parkin, which is another RING-IBR type E3 ligase (12), existed in the fractions of much lighter molecular weight (M r ) (Fig. 1B, top panels). Moreover, Dorfin was sedimented in the fractions of 400 -600 kDa in other tissues, such as the liver, kidney, and muscle of mouse, and various cultured cells including Neuro2a, HeLa, and HEK293 cells (Fig. 1B, bottom panels). To determine whether endogenous Dorfin interacts with VCP, immunoprecipitation using polyclonal anti-Dorfin antibody (Dorfin-30) was performed on the fractions shown in Fig. 1B, top panels. Endogenous VCP was co-immunoprecipitated with endogenous Dorfin in the fractions of high M r (fractions (Fr.) 13 and 14). No apparent band was observed when precipitated with rabbit IgG (Fig. 1C).
Mutations of RING Finger Domain of Dorfin Results in Loss of Dorfin-VCP Interactions-Next, we examined whether transfected Dorfin (FLAG-Dorfin WT ) and its RING mutant (FLAG-Dorfin C132S/C135S ), in which the two Cys residues at positions 132 and 135 within the RING finger domain were substituted for Ser residues, form a complex. The results showed overexpression of FLAG-Dorfin WT in high molecular fractions (Fr. in Fig. 2), whose peak was between fractions 10 and 12, whereas overexpressed FLAG-Dorfin C132S/C135S did not consist of high molecular weight complex. Overexpression of FLAG-Dorfin WT or FLAG-Dorfin C132S/C135S did not change the sedimentation pattern of VCP ( Fig. 2A). Furthermore, immunoprecipitation analysis showed that FLAG-Dorfin WT , but not FLAG-Dorfin C132S/C135S , could interact with HA-VCP in HEK293 cells (Fig. 2B).
Dorfin Interacts with VCP in Vitro-To confirm the direct binding between Dorfin and VCP and to determine the exact portion of Dorfin that interacts with VCP in vitro, we performed pull-down assays using recombinant proteins. Recombinant MBP-Dorfin or its deletion mutants (i.e. MBP-Dorfin N and MBP-Dorfin C ) and the same molar of recombinant His-VCP or GST-VCP were mixed and incubated for 1 h at 4°C. MBP-mock protein was used as a negative control in these experiments. A small portion of MBP-Dorfin full or Dorfin C (Cterminal substrate-recognizing domain) bound to both His-VCP and GST-VCP, whereas MBP-mock, MBP-Dorfin N (Nterminal RING-IBR domain), and MBP-Parkin did not bind to His-VCP or GST-VCP (Fig. 3A). We next determined the number of Dorfins that bind one hexamer of VCP. To investigate this issue, we incubated His-VCP with increasing amounts of MBP-Dorfin full , MBP-Dorfin N , MBP-Dorfin C , MBP-mock, or MBP-Parkin. As shown in Fig. 3B, the amount of binding portion of MBP-Dorfin full and -Dorfin C pulled down with His-VCP was not saturated below the even molar ratio. The pulldown experiments using excess amounts of MBP-Dorfin full revealed that MBP-Dorfin full was saturated at the even molar ratio (Fig. 3C). As reported previously (15), recombinant His-VCP sedimented in high molecular weight fractions, indicating that it formed a hexamer in vitro (Fig. 3D). These findings indicated that six Dorfin molecules were likely bind to a VCP complex in vitro. Subcellular Localization of Dorfin and VCP in HEK293 Cells-In previous studies, we showed that exogenous and endogenous Dorfin resided perinuclearly and was colocalized with Vimentin in cultured cells treated with a proteasome inhibitor (4). The staining patterns of Dorfin were indistinguishable from those of the aggresome, namely a pericentriolar, membrane-free, cytoplasmic inclusion containing misfolded ubiquitylated proteins packed in a cage of intermediate filaments (4). VCP immunostaining was also observed throughout aggresomes in cultured neuronal cells when induced by treatment with a proteasome inhibitor (15). In order to examine the subcellular localization of Dorfin and VCP, GFP-Dorfin and HA-VCP were co-expressed in HEK293 cells. Without proteasome treatment, GFP-Dorfin-expressing cells showed granular fluorescence in the cytosol, and the HA-VCP-expressing cells showed diffuse and uniform cytoplasmic staining (Fig. 4A). Treatment with MG132 (1 M, 16 h) resulted in accumulation of both GFP-Dorfin and HA-VCP and perinuclear colocalization as a clear large protein aggregate that mimics aggresomes (Fig. 4B).
Colocalization of Dorfin and VCP in the Affected Neurons of ALS and PD-In previous studies, immunostaining of Dorfin and VCP was independently noted in LBs of PD, and the peripheral staining pattern of both proteins in LBs was similar (7,23). To confirm the immunoreactivities of Dorfin and VCP in the affected neurons in ALS and PD, we performed a doublelabeling immunofluorescence study using a rabbit polyclonal anti-Dorfin antibody (Dorfin-41) and a mouse monoclonal VCP antibody on the postmortem samples of ALS and PD. In the ALS spinal cords, both proteins were colocalized in the LB-like inclusions (Fig. 5, A-F). The margin of LBs in PD was intensely immunostained for Dorfin and VCP, and merged images confirmed their strong colocalization (Fig. 5, G-L). Dorfin and VCP were also positive in Lewy neurites in the affected neurons of PD (Fig. 5, M-O).
Dorfin Ubiquitylates Mutant SOD1 in Vivo-Unlike the wild-type form, mutant SOD1 proteins are rapidly degraded by the ubiquitin-proteasome system. Consistent with our previous results (5), SOD1 G93A and SOD1 G85R were polyubiquitylated, and co-expression with FLAG-Dorfin WT enhanced polyubiquitylation of these mutant SOD1s compared with co-expression with FLAG-BAP, a negative control construct (Fig. 6A). Boiling with 1% SDS-containing buffer did not change the level of ubiquitylated mutant SOD1, indicating that mutant SOD1 itself was ubiquitylated by Dorfin (Fig. 6B). We also performed the same in vivo ubiquitylation assay using Neuro2a cells to examine for E3 activity of Dorfin in neuronal cells. The enhanced polyubiquitylation of these mutant SOD1s by Dorfin was observed in Neuro2a cells as well as in HEK293 cells (Fig.  6C). FLAG-Dorfin C132S/C135S did not enhance polyubiquitylation of mutant SOD1s, indicating that this RING finger mutant form was functionally inactive (Fig. 6D).
VCP K524A Suppresses the E3 Activity of Dorfin-VCP has two ATPase binding domains (D1 and D2). A D2 domain mutant, VCP K524A , induces cytoplasmic vacuoles, which mimics vacuole formation seen in the affected neurons in various neurodegenerative diseases (11,15). The D2 domain represents the major ATPase activity and is essential for VCP function (11). The ATPase activity of VCP K524A is much lower than that of VCP WT , and VCP K524A caused accumulation of polyubiquitylated proteins in the nuclear and membrane fractions together with elevation of ER stress marker proteins due to ERAD MBP-Dorfin full , MBP-Dorfin N , MBP-Dorfin C , and MBP-Parkin with increasing amounts (molar ratio to VCP: 0.25, 0.5, and 1.0). The amounts of MBP fusion Dorfin derivatives and His-VCP in 10% of the samples used are shown (10% input). C, 2 g of His-VCP was incubated with MBP-Dorfin full with increasing amounts (molar ratio to VCP: 0.25, 0. 5, 1, 2, and 4). The amounts of MBP-Dorfin full and His-VCP in 10% of the samples used are shown (10% input). D, His-VCP protein (0.5 g) was fractionated by 10 -40% glycerol gradient centrifugation followed by separation into 30 fractions using a fraction collector. Immunoblotting using anti-VCP antibody was performed on the selected fractions (fractions 2-17). *, The molar ratio was calculated by the amount of VCP monomers, not VCP complexes.

FIG. 4. Subcellular localization of GFP-Dorfin and HA-VCP in HEK293 cells treated or untreated with a proteasome inhibitor.
GFP-Dorfin and HA-VCP were co-expressed transiently in HEK 293 cells. Cells were treated with (B) or without (A) 1 M MG132 for 16 h. HA-VCP was stained with anti-monoclonal HA antibody (12CA5). Nuclei were stained with 4Ј,6-diamidino-2-phenylindole (DAPI). Without the treatment of MG132, GFP-Dorfin was spread through the cytosol, and it appeared like small aggregations. HA-VCP was also seen mainly in the cytosol and partly colocalized with GFP-Dorfin (A). After treatment with 1 M MG132 for 16 h, both GFP-Dorfin and HA-VCP showed perinuclear accumulation and colocalization and appeared as clear large protein aggregates (B; arrows). inhibition, whereas its expression level, localization, and complex formation were indistinguishable from those of VCP WT (11). In order to examine the functional effect of VCP on Dorfin, VCP WT , VCP K524A , or LacZ was co-expressed with SOD1 G85R , FLAG-Dorfin, and HA-Ub in HEK293 cells. Co-expression with VCP K524A showed a marked decline of polyubiquitylation of SOD1 G85R compared with co-expression with VCP WT or LacZ (Fig. 7A, top and middle). Since Dorfin physically interacts with mutant SOD1s (5), we next investigated whether this decline of polyubiquitylation of SOD1 G85R was mediated by reduced affinity between SOD1 G85R and Dorfin. Immunoprecipitation by anti-FLAG antibody showed that VCP K524A did not change affinity between SOD1 G85R and Dorfin (Fig. 7A,  bottom). Neither VCP WT nor VCP K524A changed the level of polyubiquitylation protein in the total lysate (Fig. 7B). To clarify whether this negative effect of VCP K524A is specific for Dorfin, we assessed the autoubiquitylation of FLAG-Parkin in the presence of VCP WT , VCP K524A , or LacZ. Co-expression of VCP K524A did not decrease autoubiquitylation of FLAG-Parkin compared with co-expression of LacZ or VCP WT (Fig. 7C). We performed the same experiments using Neuro2a cells to see whether VCP K524A suppress the E3 activity of Dorfin in neu-ronal cells. The marked decline of polyubiquitylation of SOD1 G85R by VCP K524A expression was also seen in Neuro2a cells (Fig. 7D). DISCUSSION UBIs in the affected neurons are histopathological hallmarks in various neurodegenerative disorders (8). Dorfin is an E3 ligase, which can ubiquitylate mutant SOD1s and synphilin-1 (5,24). These substrates and Dorfin were identified in UBIs in various neurodegenerative diseases, such as LB-like inclusions in ALS and LBs in PD and dementia with Lewy bodies (7). This finding suggests that Dorfin may play a crucial role in the process of generating inclusions in the affected neurons. In the present study, we identified VCP as one of the Dorfin-associated proteins using mass spectrometry, and VCP-Dorfin physical interaction was confirmed by an immunoprecipitation experiment using FLAG-Dorfin and HA-VCP overexpressed in HEK293 cells (Fig. 1A). VCP is an essential and highly conserved protein of the AAA-ATPase family, which is considered to have diverse cellular functions, such as membrane fusion (25)(26)(27), nuclear trafficking (28), cell proliferation (29,30), and the ERAD pathway (18 -22). Many reports have implied that VCP is involved in the pathogenesis of various neuromuscular diseases. VCP has been implicated as a factor that modifies the progress of polyglutamine-induced neuronal cell death (15). In addition, histopathological studies revealed positive staining for VCP in UBIs in PD and ALS with dementia (23). VCP is also associated with MJD protein/atxin-3, in which abnormal expansion of polyglutamine tracts causes Machado-Joseph disease/spinocerebellar ataxia type 3 (31). VCP is also required for the degradation of ataxin-3 in collaboration with E4B/Ufd2a, a ubiquitin chain assembly factor (E4) (32). Recent studies have indicated that missense mutations in the VCP gene cause inclusion body myopathy associated with Paget's disease of bone and frontotemporal dementia, which is characterized by the presence of vacuoles in the cytoplasm in muscle fibers (33).
Our results showed that endogenous Dorfin formed a 400 -600-kDa complex in various tissues and various cultured cells (Fig. 1B). Dorfin is a ϳ91-kDa protein; therefore, this high M r complex should include Dorfin-associated proteins, although the possibility that Dorfin itself oligomerizes in the cell cannot be excluded. Glycerol gradient centrifugation analysis and immunoprecipitation experiments in the present study showed that endogenous Dorfin interacted with endogenous VCP in a complex of approximately 600 kDa, possibly including a Dorfin molecule and a hexametric form of VCP (Fig. 1C).
The first RING mutant of Dorfin, in which Cys at positions 132 and 135 changed to Ser, was prepared. This mutant Dorfin, Dorfin C132S/C135S , could not ubiquitylate mutant SOD1s (Fig.  6D). Glycerol gradient centrifugation analysis revealed that Dorfin C132S/C135S did not form a high M r complex, whereas exogenous wild type Dorfin (Dorfin WT ) formed a high M r complex similar to endogenous Dorfin ( Fig. 2A). Furthermore, an immunoprecipitation experiment using Dorfin WT and Dorfin C132S/C135S revealed that Dorfin WT could interact with VCP, whereas Dorfin C132S/C135S could not (Fig. 2B).
Our in vitro study using recombinant proteins showed that full-length (MBP-Dorfin full ) and the C terminus of Dorfin (MBP-Dorfin C ) directly interacted with VCP, whereas the MBP-Dorfin N mutant, containing the entire RING finger domain (amino acid residues 1-367), did not bind to VCP (Fig. 3A). This finding was unexpected, since in vivo binding analysis suggested that Dorfin could interact with VCP at the RING finger domain. It is plausible that certain structural changes in Dorfin C132S/C135S might render the C-terminal VCP-binding portion incapable of accessing VCP molecules. This may explain the result that Dorfin C132S/C135S did not form a high M r complex.  The amount of Dorfin bound with VCP was saturated at even molar ratio in vitro (Fig. 3, B and C). Since VCP exists as a homohexamer (Fig. 3D), the in vivo observed size of ϳ600 kDa appears to be too small for the Dorfin-VCP complex if one VCP molecule binds to more than one Dorfin as shown in in vitro experiments. However, it is noteworthy that the size of molecules estimated by glycerol density gradient centrifugation analysis used in this study is not accurate and sufficient to discuss the molecular interaction of Dorfin and VCP in the cells. To date, various adaptor proteins, with which VCP forms multiprotein complexes, have been identified, such as Npl4, Ufd1 (18,20), Ufd2 (34), Ufd3 (35), p47 (36), or SVIP (37). Although our in vitro study showed direct physical interaction between Dorfin and VCP, the environment with those adaptor proteins might reflect in vivo conditions. This also may explain the apparent discrepancy of the Dorfin-VCP binding fashions between in vivo and in vitro analyses.
Treatment with a proteasomal inhibitor causes the translocation of endogenous VCP and Dorfin to the aggresome in cultured cells (4,15). Our results showed that these two proteins indeed colocalized perinuclearly in the aggresome following treatment with a proteasomal inhibitor (Fig. 4). Furthermore, we were able to demonstrate both Dorfin and VCP immunoreactivities in LB-like inclusions in ALS and LBs in PD (Fig. 5). In the majority of LBs, indistinguishable peripheral staining patterns were observed with both anti-Dorfin and anti-VCP antibodies. These results confirmed that both Dorfin and VCP are associated with the formation processes of aggresomes and inclusion bodies through physical interaction.
We showed here that co-expression of VCP K524A resulted in a marked decrease of ubiquitylation activity of Dorfin compared with co-expression of VCP WT or control. On the other hand, VCP K524A failed to decrease autoubiquitylation activity of Parkin. VCP K524A did not change the level of polyubiquitylated protein accumulation in the cell lysate in this study (Fig. 7). Knockdown experiments using the RNA interference technique showed accumulation of polyubiquitylated proteins (38). Combined with the observation that inhibition of VCP did not decrease the general accumulation of polyubiquitylated proteins, our results indicated that the E3 regulation function of VCP may be specific to certain E3 ubiquitin ligases such as Dorfin. VCP is an abundant protein that accounts for more than 1% of protein in the cell cytosol and is known to have various chaperone-like activities (39); therefore, it may function as a scaffold protein on the E3 activity of Dorfin. The localization of Dorfin and VCP in UBIs in various neurodegenerative disorders indicates the involvement of these proteins in the quality control system for abnormal proteins accumulated in the affected neurons in neurodegenerative disorders.
Since the unfolded protein response and ERAD are dynamic responses required for the coordinated disposal of misfolded proteins (40), the ERAD pathway can be critical for the etiology of neuronal cell death caused by various unfolded proteins. VCP is required for multiple aspects of the ERAD system by recognition of polyubiquitylated proteins and translocations to the 26 S proteasome for processive degradation through the VCP-Npl4-Ufd1 complex (18,41). Our results suggest the involvement of Dorfin in the ERAD system, which is related to the pathogenesis of neurodegenerative disorders, such as PD or Alzheimer's disease. Further study including Dorfin knockout and/or knockdown models should examine the pathophysiology of Dorfin in association with the ERAD pathway or other cellular functions. Such studies should enhance our understanding of the pathogenetic role of Dorfin in neurodegenerative disorders.