NEDL1, a novel ubiquitin-protein isopeptide ligase for dishevelled-1, targets mutant superoxide dismutase-1.

Approximately 20% of familial amyotrophic lateral sclerosis (FALS) arises from germ-line mutations in the superoxide dismutase-1 (SOD1) gene. However, the molecular mechanisms underlying the process have been elusive. Here, we show that a neuronal homologous to E6AP carboxyl terminus (HECT)-type ubiquitin-protein isopeptide ligase (NEDL1) physically binds translocon-associated protein-delta and also binds and ubiquitinates mutant (but not wild-type) SOD1 proportionately to the disease severity caused by that particular mutant. Immunohistochemically, NEDL1 is present in the central region of the Lewy body-like hyaline inclusions in the spinal cord ventral horn motor neurons of both FALS patients and mutant SOD1 transgenic mice. Two-hybrid screening for the physiological targets of NEDL1 has identified Dishevelled-1, one of the key transducers in the Wnt signaling pathway. Mutant SOD1 also interacted with Dishevelled-1 in the presence of NEDL1 and caused its dysfunction. Thus, our results suggest that an adverse interaction among misfolded SOD1, NEDL1, translocon-associated protein-delta, and Dishevelled-1 forms a ubiquitinated protein complex that is included in potentially cytotoxic protein aggregates and that mutually affects their functions, leading to motor neuron death in FALS.

loss of motor neurons in the spinal cord, brain stem, and motor cortex. The sporadic and familial forms of the disease have similar clinical and pathological features. About 10% of ALS cases are familial, and mutation of superoxide dismutase-1 (SOD1) is found in 20% of familial ALS (FALS) patients (1,2). Mice that express mutant SOD1 transgenes develop an age-dependent ALS phenotype independent of levels of dismutase activity, suggesting that FALS pathology is because of a toxic gain of function in SOD1 and that the abnormal protein structure of mutant SOD1 is critical in the pathogenesis of motor neuron death (3)(4)(5)(6). Recently, proteasome expression and activity have been reported to decrease with age in the spinal cord (7,8). Furthermore, mutant SOD1 turns over more rapidly than wild-type SOD1, and an inhibitor of proteasome action inhibits this turnover and thus selectively increases the steadystate level of mutant SOD1 (8). These results suggest the involvement of the ubiquitin-proteasome function in the cause of FALS. However, the biochemical nature of this gain-of-function mutation in SOD1 and the mechanism by which SOD1 mutations cause the degeneration of motor neurons have remained elusive.
We show here the identification of a novel HECT-type ubiquitin-protein isopeptide ligase (E3), NEDL1, which is expressed in neuronal tissues, including the spinal cord, and selectively binds to and ubiquitinates mutant (but not wildtype) SOD1. NEDL1 is physically associated with transloconassociated protein-␦ (TRAP-␦), one of the endoplasmic reticulum (ER) translocon components that has previously been reported to bind mutant SOD1 (9,10). Both NEDL1 and TRAP-␦ form a complex with mutant SOD1, with the binding intensity among these proteins being roughly proportionate to the rapidity of progression of the associated FALS phenotype. Immunohistochemical study has shown that NEDL1 is positive in the Lewy body-like hyaline inclusions in the spinal cord motor neurons of both FALS patients and mutant SOD1 transgenic mice. We have also found that NEDL1 targets Dishevelled-1 (Dvl1) for ubiquitination-mediated degradation and that mutant (but not wild-type) SOD1 affects the function of Dvl1. Our observations suggest that NEDL1 is a quality control E3 that recognizes mutant SOD1 to form a tight complex with the physiological targets of NEDL1 in motor neurons of FALS patients.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection-Human neuroblastoma-derived cells were grown in RPMI 1640 medium supplemented with 10% heat-inactivated fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin. COS-7 and Neuro2a cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fe-tal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin. All cells were maintained in a humidified 37°C incubator with 5% CO 2 . All transfections were carried out with LipofectAMINE Plus transfection reagent (Invitrogen) according to the manufacturer's instructions.
In some experiments, transfected cells were treated with MG-132 for 30 min at a final concentration of 40 M.
In Vitro Ubiquitination Assays-In vitro ubiquitination assays were performed as follows. Reaction mixtures containing 0.5 g of purified glutathione S-transferase fusion proteins, 0.25 g of yeast ubiquitinactivating enzyme (E1) (BostonBiochem, Cambridge, MA), 1 l of crude lysates from Escherichia coli expressing ubiquitin carrier proteins (E2), and 10 g of bovine ubiquitin (Sigma) were incubated in 250 mM Tris-HCl (pH 7.6), 1.2 M NaCl, 50 mM ATP, 10 mM MgCl 2 , and 30 mM dithiothreitol. Reactions were terminated after 2 h at 30°C by the addition of SDS sample buffer. Samples were resolved by SDS-PAGE, transferred to membranes, and immunoblotted with anti-ubiquitin monoclonal antibody 1B3 (Medical & Biological Laboratories, Nagoya, Japan).
Yeast Two-hybrid Screening-Yeast two-hybrid screening was performed using the Gal4-based Matchmaker two-hybrid system with the cDNA libraries derived from fetal human brain (first screening) and adult human brain (second screening) (Clontech, Palo Alto, CA). Saccharomyces cerevisiae CG1945 cells were transformed with pAS2-1-NEDL1-1 (amino acids 757-1114; first screening) or pAS2-1-NEDL1-2 (amino acids 382-1448; second screening), which did not activate the transcription of lacZ alone. The transformants were subsequently transformed with the cDNA library, and the lacZ-positive colonies were selected. The plasmid DNAs were extracted from these positive colonies, and their nucleotide sequences were determined.
Cloning of Human NEDL1 cDNA-A forward primer (5Ј-GGTTTT-TAGGCCTGGCCGCC-3Ј) and a reverse primer (5Ј-CAATGAGGTA-CATGCCAATCC-3Ј) were used to amplify the 5Ј-part of the NEDL1 cDNA using cDNA libraries derived from human neuroblastoma and fetal human brain (Stratagene, La Jolla, CA) as templates. The fulllength human NEDL1 cDNA was generated by fusion of the PCRamplified fragment (nucleotides ϩ1 to ϩ68, where position ϩ1 represents the translation initiation site) and the KIAA0322 cDNA (a gift from T. Nagase, Kazusa DNA Institute). Gel electrophoresis and Western blot analysis were carried out as described above.
Expression Constructs-The mammalian expression plasmids for hemagglutinin-tagged and His 6 -tagged ubiquitin were kind gifts of D. Bohmann. The full-length NEDL1 cDNA was inserted into the mammalian expression plasmid pEF1/His (Invitrogen) or pIRESpuro2 (Clontech). cDNAs encoding wild-type and mutant forms of SOD1 were fused to the FLAG or Myc epitope tag sequence at their C termini and subcloned into pIRESpuro2. Similarly, the FLAG or Myc epitope tag sequence was attached to the C terminus of TRAP-␦. Also similarly, the FLAG or Myc epitope tag sequence was attached to the N terminus of Dvl1. Coding sequences were verified by automated DNA sequencing.
Protein Stability Experiments-Neuro2a cells were transfected with the expression plasmid for the wild-type or mutant form of SOD1 with or without the NEDL1 expression plasmid. Twenty-four hours after transfection, cycloheximide (50 g/ml) was added to the culture medium, and the cells were harvested at the indicated time points by lysis in radioimmune precipitation assay buffer. The protein concentrations were determined using the Bradford protein assay system (Bio-Rad) according to the instructions of the manufacturer.
Immunohistochemistry-The immunohistochemical studies were performed as described previously using affinity-purified rabbit anti-NEDL1 antibody (11). Patient tissues were obtained at autopsy from two FALS siblings from a Japanese family. The clinical course of the sister, who died at age 46, was 18 months (case 1), and that of the brother, who died at age 65, was 11 years (case 2) (11). The SOD1 gene was mutated with a 2-bp deletion at codon 126 (11,12). Normal spinal cord tissues were obtained from three neurologically and neuropathologically normal individuals. The same study was performed on spinal cord tissues from three normal rats and a transgenic ALS rat carrying a mutant allele of the human SOD1 gene (H46R) (13). These mice were killed at 180 days. As a negative control, some sections were incubated with anti-NEDL1 antibody that had been pre-absorbed with an excess of NEDL1 antigen. Bound antibodies were visualized by the avidin⅐biotin⅐immunoperoxidase complex method.

RESULTS
Cloning and Expression of the NEDL1 E3 Gene-To detect novel molecules that are important in regulating neuronal programmed cell death, we constructed oligo-capping cDNA libraries from a mixture of three fresh human neuroblastoma tissues (stages 1 and 2) that were undergoing gradual spontaneous regression, probably by neuronal apoptosis (14). Screening of 1152 novel genes by RT-PCR revealed that 194 genes were expressed differentially in regressing neuroblastomas with favorable prognosis and in aggressive tumors with poor prognosis. Among these genes, we found a partial cDNA sequence with an HECT-like domain (Nbla0078) that partially matched the KIAA0322 gene. Because KIAA0322 lacks a 5Ј-coding region, we used a genome-based PCR procedure to clone the corresponding full-length cDNA. This is predicted to encode a protein product of 1585 amino acids with homology to NEDD4 E3 (15,16), which includes a C2 domain at the N-terminal region supposed to mediate its membrane localization in a calcium-dependent manner, two WW motifs important for protein-protein interaction through binding to specific proline-rich clusters, and a conserved catalytic HECT domain at the C terminus (Fig. 1A). We named this novel ligase, which mapped to chromosome 7p13, NEDL1 (NEDD4-like ubiquitin-protein ligase-1). We also cloned the mouse counterpart of NEDL1 cDNA, whose amino acid sequence is 78% identical to the human sequence. Tissue-specific expression of NEDL1 mRNA of ϳ10 and 7 kb in size was observed, with predominant expression in adult and fetal brains as examined by Northern blot analysis (Fig. 1B). Its expression was also weakly detected in adult kidney, where the size of the expressed transcript appeared to be Ͻ7 kb. Expression of NEDL1 in specific regions of the nervous system was further confirmed in the cerebral cortex, corpus callosum, cerebral peduncles, and spinal cord by RT-PCR (Fig. 1C). Thus, NEDL1 is a novel HECT-type E3 preferentially expressed in neuronal tissues, including the spinal cord. Using a specific anti-NEDL1 polyclonal antibody that we generated, we localized NEDL1 primarily to the cytoplasm in both intact human neuroblastoma CHP134 cells and COS-7 cells transiently expressing NEDL1 (Fig. 1D). The in vitro system containing UbcH5c or UbcH7 demonstrated that NEDL1 has a ubiquitin-protein ligase activity (Fig. 1E).
NEDL1 Physically Interacts with TRAP-␦ and Mutant SOD1-We then sought protein-binding partners of NEDL1 by yeast two-hybrid screening using the region including two WW protein interaction domains (amino acids 757-1114) as bait. Of 96 positive clones subjected to DNA sequencing, one was a full-length cDNA for TRAP-␦; this was of considerable interest, as TRAP-␦ was previously reported to bind mutant (G85R and G93A), but not wild-type, SOD1 (9). TRAP-␦ is a protein component of the translocon in the ER membrane (10). We therefore examined the interaction among NEDL1, TRAP-␦, and SOD1 by an immunoprecipitation assay after cotransfecting the corresponding expression constructs into COS-7 cells. As shown in Fig. 2 (A and B), NEDL1 was physically associated with both exogenous and endogenous TRAP-␦ probably through the region of two WW domains, as originally suggested by the result of two-hybrid screening. Surprisingly, NEDL1 bound to mutant (but not wild-type) SOD1 (Fig. 2C). Furthermore, the degree of binding between NEDL1 and different mutant SOD1 proteins was roughly proportionate to the rapidity of progression (time from clinical onset to death) of the associated FALS phenotype (17)(18)(19)(20)(21)(22)(23). For example, two mutant SOD1 proteins associated with an extremely rapid clinical course (C6F and A4V) interacted very strongly with NEDL1. By contrast, the binding of NEDL1 to other mutants was less striking and decreased proportionately to the falloff of disease severity corresponding to those mutants. Of further interest, like the NEDL1-mutant SOD1 interaction, the binding intensity between TRAP-␦ and mutant SOD1 was also dependent on the disease severity (Fig. 2D). These observations suggest that NEDL1 and TRAP-␦ are normally associated with each other, but that misfolded mutant SOD1 makes a complex with them. Such a complex is not formed with wild-type SOD1. The experiments using the in vitro translated proteins suggested that association of mutant SOD1 and TRAP-␦ was direct (data not shown). It therefore appears that mutant SOD1 forms tightly bound protein complexes with NEDL1 and TRAP-␦ and that the tightness of binding in the complex is determined in part by FIG. 1. Amino acid sequence, brain-specific expression, and subcellular localization of NEDL1 E3. A, alignment of conserved amino acid sequences of human NEDL1 (hNEDL1) and its mouse homolog (mNEDL1). Numbers on the right indicate the number of residues to the initiator methionine. The C2 domain (shaded), two WW domains (dashed boxes), and the HECT domain (solid box) are indicated. B, brain-specific expression of NEDL1 mRNA. Total RNAs derived from the indicated adult (left panel) and fetal (right panel) human tissues were analyzed by Northern blotting using a 32 P-labeled human NEDL1 cDNA restriction fragment as a probe. Control hybridization with a human ␤-actin cDNA probe verified the equal amount of RNA loaded. C, expression of NEDL1 in human brain subsections. Total RNA from the cerebral cortex, corpus callosum, cerebral peduncles, spinal cord, or placenta was subjected to RT-PCR using specific primers for NEDL1 or glyceraldehyde-3-phosphate dehydrogenase (GAPDH). RT-PCR analysis for NEDL1 in the placenta provided a negative control. Amplification of glyceraldehyde-3-phosphate dehydrogenase was used as an internal control. D, confocal microscopic images of human neuroblastoma CHP134 cells (left panel) and COS-7 cells transfected with an expression plasmid for NEDL1 (right panel). Cells were subjected to immunofluorescence analysis using rabbit anti-NEDL1 polyclonal antibody, followed by fluorescein isothiocyanate-conjugated anti-mouse IgG. E, in vitro ubiquitination assays showing that NEDL1 has a ubiquitin-protein ligase activity. The degree of ubiquitination was increased in an NEDL1-dependent manner. In this assay, yeast ubiquitinactivating enzyme (E1), bacterially expressed ubiquitin carrier protein (E2; UbcH5C or UbcH7), and bacterial lysates were incubated in the presence or absence of increasing amounts of glutathione S-transferase (GST)-NEDL1. Polyubiquitinated bacterial proteins appeared to migrate in a high molecular mass complex. Ub, ubiquitin.
properties of the mutant enzyme that also modulate disease severity of the resulting ALS phenotype. Such complexes do not form in cells with wild-type SOD1.
Determination of the Interaction Domains-We next examined the domains of NEDL1 required for formation of the SOD1⅐NEDL1⅐TRAP-␦ complex. We generated various con-structs of NEDL1 with deletions of each domain. Fig. 3 shows the results of immunoprecipitation assay for the association between deletion mutants of NEDL1 and mutant SOD1(G93A). Mutant SOD1 bound weakly to NEDL1 lacking WW domain-1 (Fig. 3A), suggesting that WW domain-1 and its surrounding portion are the region involved in their interaction. Immuno- precipitation analysis using the specific regions of NEDL1 clearly showed that the region between the C2 domain and WW domain-1 (CW linker region) is necessary for binding to mutant SOD1(G93A). Mutant SOD1(A4V) was also associated with NEDL1 through the same region, and TRAP-␦ bound to the two WW domains of NEDL1 (data not shown).

NEDL1 Ubiquitinates Mutant SOD1 for Degradation Depending on the Disease Severity of FALS-Because NEDL1 is
an E3, we next tested whether it ubiquitinates TRAP-␦ and mutant SOD1 for degradation. As shown in Fig. 4A, NEDL1 clearly ubiquitinated mutant SOD1(A4V), but not TRAP-␦ (data not shown). Furthermore, the degree of ubiquitination of mutant SOD1 by NEDL1 was dependent on the disease severity of FALS (A4V Ͼ G93A Ͼ H46R) (Fig. 4A). Fig. 4B shows the time course of degradation of wild-type and mutant SOD1 in the presence or absence of NEDL1. As reported previously (46), mutant SOD1 was degraded more rapidly than wild-type SOD1. NEDL1 did not affect wild-type SOD1 degradation. As expected from the co-immunoprecipitation and ubiquitination analyses, degradation of mutant SOD1 was stimulated by NEDL1 proportionately to the disease severity of FALS caused by the particular SOD1 mutant (A4V Ͼ G93A Ͼ H46R Ն wild-type). Thus, NEDL1 targeted mutant SOD1 for ubiquitinmediated degradation in the cell in parallel with the binding intensity.
Immunohistochemistry-One of the characteristic cytopathological changes of mutant SOD1-linked FALS is the formation of neuronal Lewy body-like hyaline inclusions (LBHIs) that contain aggregates of SOD1 and ubiquitin (24). We therefore performed immunostaining to determine whether the NEDL1 protein is included within the LBHIs of the spinal cord motor neurons obtained from two siblings with FALS caused by frameshift 126 mutation of SOD1 (11,12). One case had neuropathological findings compatible with FALS with posterior column involvement, whereas the other had multisystem degeneration in addition to motor neuron disturbance. We also performed NEDL1 immunostaining in specimens obtained from mutant SOD1(H46R) transgenic mice at 180 days, by which time they show clinical motor signs in the hind limbs (13). The specificity of the NEDL1 staining was confirmed by pretreating the specimens with an excess of NEDL1 antigen. NEDL1 immunoreactivity in the spinal cords of the human control cases was identical to that of normal mice: immunoreactivity was identified predominantly in the cytoplasm of the neurons of the spinal cords (Fig. 5, A and B). The LBHIs in the anterior horn cells of two FALS patients and transgenic mice showed equivalent immunoreactivity for NEDL1. Although the intensity of NEDL1 immunoreactivity in neuronal LBHIs varied, most of the LBHIs were immunoreactive for NEDL1 (Fig.  5, C and D). The reaction products were generally restricted to the cores of the core and halo-type LBHIs that showed eosinophilic cores with pale peripheral halos upon hematoxylin and eosin staining (Fig. 5C); by contrast, immunopositive NEDL1 in ill defined LBHIs was distributed throughout the inclusions (Fig. 5D). NEDL1 immunoreactivity in the residual neurons in humans and mice was identified primarily in cell bodies. Thus, NEDL1 immunostaining was clearly positive in the FALSrelated LBHIs that were also positive for ubiquitin and SOD1 (data not shown).
NEDL1 Targets Dishevelled-1 for Ubiquitin-mediated Protein Degradation-We next hypothesized that the physiological function of NEDL1 to mediate ubiquitination is interfered with  5. NEDL1 immunohistochemical analyses. A, immunohistochemical analysis of NEDL1 in normal human spinal cord. NEDL1positive anterior horn cells are evident (arrow), although the immunoreactivity for NEDL1 is somewhat faint. There was no counterstaining. Magnification ϫ520. B, NEDL1 immunohistochemistry in normal mouse spinal cord. Normal anterior horn cells are positive for NEDL1 (arrow). The section was counterstained with hematoxylin. Magnification ϫ750. C, immunostaining for NEDL1 in spinal cord LBHIs from an FALS patient with a frameshift 126 mutation in the SOD1 gene. The NEDL1-positive reaction products were mostly restricted to the cores of the core and halo-type LBHIs (arrowheads). In the LBHI-bearing neurons and residual neurons, the antibody to NEDL1 also stained the neuronal cell body. There was no counterstaining. Magnification ϫ540. D, NEDL1 immunostaining in a spinal cord LBHI from an SOD1(H46R) transgenic mouse. An ill defined LBHI in the SOD1(H46R) transgenic mouse was positive for NEDL1; this ill defined LBHI shows a diffuse staining pattern (arrowhead). The staining intensity in the residual neurons stained by anti-NEDL1 antibody varied from neuron to neuron. The section was counterstained with hematoxylin. Magnification ϫ770.

NEDL1 Targets Mutant Superoxide Dismutase-1
by mutant SOD1. To test this hypothesis, we again performed yeast two-hybrid screening to obtain NEDL1-interacting molecules using the large region of NEDL1 (amino acids 382-1448) as bait. Of 396 His and ␤-galactosidase double-positive clones, 282 clones were subjected to DNA sequencing, and we identified Dvl1 (three clones). Human Dvl1 is a 670-amino acid protein with three conserved domains: a DIX domain, which is required for canonical Wnt/T-cell factor signaling; a PDZ domain, which is a target of both Stbm and casein kinase I binding; and a DEP domain, which is responsible for Dvl membrane localization during planar cell polarity signaling (25)(26)(27). Between the DEP domain and C-terminal end, there are three proline-rich clusters unique to mammalian Dvl1, which presumably act as the WW domain recognition sites. All three clones (clones 2-13, 1-56, and 1-77) contain the DEP domain and proline-rich clusters, suggesting that NEDL1 interacted with Dvl1 in the C-terminal half (Fig. 6A). In Neuro2a cells, NEDL1 co-immunoprecipitated with Dvl1 (Fig. 6B) and ubiquitinated it for degradation (Fig. 6, C and D). Thus, Dvl1 may be one of the physiological targets of NEDL1 E3. As recent studies strongly suggest that the cytotoxicity of SOD1 mutants is responsible for their aggregate properties, incorporating other proteins essential for cells into their aggregates (28), we examined the association between mutant SOD1 and Dvl1, both of which interact with NEDL1. Of interest, Dvl1 bound to mutant SOD1(A4V), and complex formation was increased in the presence of NEDL1 roughly proportionately to the disease severity of FALS caused by the particular SOD1 mutant (Fig.  6E). Dvl1 is known to transduce not only the Wnt/␤-catenin/Tcell factor pathway, but also the JNK/c-Jun pathway (27). Therefore, we next examined whether the Dvl1-induced phosphorylation of c-Jun at Ser 63 was affected by the tight complex formation induced by inclusion of mutant SOD1. As shown in Fig. 6F, c-Jun phosphorylation induced by overexpression of Dvl1 was significantly suppressed by coexpression with mutant SOD1(A4V) in COS-7 cells. DISCUSSION Our present results demonstrate that a novel HECT-type NEDL1 E3, which is preferentially expressed in neuronal tissues, specifically targets mutant forms of SOD1 for ubiquitination-mediated protein degradation. NEDL1 is also associated with TRAP-␦ localized at the ER translocon. The TRAP complex has recently been shown to facilitate the initiation of protein translocation in a substrate-specific manner (29). The NEDL1⅐TRAP-␦ complex recognizes mutant (but not wild-type) SOD1, with a binding intensity that broadly parallels the disease severity of FALS. NEDL1 immunoreactivity was detected in the FALS-related LBHIs in the spinal cord ventral horn motor neurons, suggesting that, although mutant SOD1 is ubiquitinated for degradation by NEDL1, the mutant SOD1⅐NEDL1⅐TRAP-␦ complex aggregates within the LBHIs. It is also conceivable that fragmentation of the Golgi apparatus reported in ALS patients and transgenic mice might be related to this aggregation (30,31). These findings suggest possible hypotheses for the role of NEDL1 in the pathogenesis of FALS: 1) NEDL1, alone or with TRAP-␦, ubiquitinates and aggregates mutant SOD1, thereby decreasing the function of mutant SOD1; 2) NEDL1 and TRAP-␦ form aggregates with mutant SOD1 that induce fragmentation of the Golgi apparatus, leading to neuronal apoptosis; 3) formation of these aggregates causes dysfunction of NEDL1 and/or TRAP-␦, and this, in turn, induces disturbances that ultimately cause motor neuron death; and 4) the mutant SOD1⅐NEDL1⅐TRAP-␦ aggregates trap and inactivate unknown factor(s) such as molecular chaperones whose normal function is important for motor neuron viability.
To further understand the role of NEDL1 in motor neuron death, we searched for the physiological targets of NEDL1 and identified Dvl1. As expected, Dvl1 is ubiquitinated for degradation by NEDL1. Surprisingly, however, Dvl1 also interacts with mutant SOD1 in the presence of NEDL1 roughly proportionately to the disease severity of FALS caused by the particular SOD1 mutant. Dvl1, an essential multimodule signal transducer localized in the cellular cytosol and cytoskeleton, mediates planar cell polarity signaling as well as canonical Wnt/␤-catenin signaling (27,32). In mammals, three Dvl family members have so far been reported, and the level of Dvl1 expression is high in neuronal tissues (33). As far as we know, NEDL1 is the first E3 for Dvl1, interacting with the C-terminal region containing three proline-rich clusters. A recent report suggests that Dvl1 regulates microtubule stability through inhibition of glycogen synthase kinase-3␤ (34). Because cytoskeletal abnormalities have been reported in ALS motor neurons (35), it is possible that the effect of mutant SOD1 on NEDL1mediated Dvl1 degradation is involved in the motor neuron death. Furthermore, Dvl1 is abundant in the postsynaptic membrane region at the neuromuscular junction (36) that is reported to be involved in several neurodegenerative disorders (37,38). Of interest, Dvl1 is mapped to chromosome 1p36, which is a commonly deleted region in many human cancers, including neuroblastoma (39). As NEDL1 is highly expressed in neuroblastomas with favorable prognosis, which have a tendency to differentiate and/or regress, NEDL1 may be involved in the regulation of neuronal differentiation and survival possibly by controlling Dvl1. NEDL1, TRAP-␦, mutant SOD1, and Dvl1 appear to form a complex roughly proportionately to the disease severity of FALS caused by the particular SOD1 mutant. Our present observations strongly suggest that NEDL1 may be a quality control E3 recognizing misfolded mutant SOD1 (40). The association between mutant SOD1 and NEDL1 may induce the conformational change in the NEDL1 protein to increase the binding intensity with other physiological targets such as TRAP-␦ (not ubiquitinated) and Dvl1 (ubiquitinated). This may lead to tight complex formation especially when the proteasome activity is impaired. It has been reported that the expression and function of proteasomes decrease with age in the spinal cord (7). Okado-Matsumoto and Fridovich (41) have also found that complex formation between mutant SOD1 and heat shock proteins leads to protein aggregates. Because our data show that the ER translocon component TRAP-␦ is involved, aggregate formation may occur at the sites of the ER or Golgi apparatus or even at other cellular sites. The complex formation including NEDL1 and mutant SOD1 may conversely affect the physiological function of NEDL1, as demonstrated by a decrease in Dvl1-induced phosphorylation of c-Jun.
Recently, the RING finger-type E3 Dorfin has been reported to ubiquitinate mutant SOD1 for degradation (42). However, NEDL1 and Dorfin appear to be different in several aspects. First, NEDL1 is expressed specifically in neuronal tissues, including the spinal cord, whereas Dorfin is ubiquitously expressed in most human tissues. Second, both interaction between NEDL1 and mutant SOD1 and ubiquitination of the latter by NEDL1 roughly parallel the disease severity caused by the particular SOD1 mutant, whereas Dorfin similarly ubiquitinates mutant forms of SOD1. In addition, we have identified Dvl1 and TRAP-␦ as cellular target proteins of NEDL1, whereas the physiological targets of Dorfin have never been reported. It is probable that there are some other E3 ligases targeting mutant SOD1. However, the molecular characteristics, including tissue-specific expression, subcellular localization, and age-dependent expression, might be important in the development of the FALS phenotype.
In conclusion, we have identified a novel neuronal E3 (NEDL1) that interacts with TRAP-␦ and also binds to and ubiquitinates Dvl1 for degradation. Strikingly, NEDL1 targets and ubiquitinates mutant (but not wild-type) SOD1 for degradation. NEDL1 may normally function in the quality control of cellular proteins by eliminating misfolded proteins such as mutant SOD1, possibly via a mechanism analogous to that of ER-associated degradation (43)(44)(45). NEDL1 appears to complex tightly with mutant SOD1, Dvl1, and TRAP-␦, forming aggregates with species of mutant SOD1 that have escaped ubiquitin-mediated degradation. The NEDL1 function that affects the activities of the target proteins may also be modulated by mutant SOD1. All of these might contribute to the pathogenesis of FALS; further elucidation of the molecular mechanism of formation of this complex and its pathogenicity may provide insights into motor neuron death in ALS as well as possible new therapeutic strategies for ALS.