Association of the Cytoskeletal GTP-binding Protein Sept4/H5 with Cytoplasmic Inclusions Found in Parkinson's Disease and Other Synucleinopathies*

α-Synuclein-positive cytoplasmic inclusions are a pathological hallmark of several neurodegenerative disorders including Parkinson's disease, dementia with Lewy bodies, and multiple system atrophy. Here we report that Sept4, a member of the septin protein family, is consistently found in these inclusions, whereas five other septins (Sept2, Sept5, Sept6, Sept7, and Sept8) are not found in these inclusions. Sept4 and α-synuclein can also be co-immunoprecipitated from normal human brain lysates. When co-expressed in cultured cells, FLAG-tagged Sept4 and Myc-tagged α-synuclein formed detergent-insoluble complex, and upon treatment with a proteasome inhibitor, they formed Lewy body-like cytoplasmic inclusions. The tagged Sept4 and α-synuclein synergistically accelerated cell death induced by the proteasome inhibitor, and this effect was further enhanced by expression of another Lewy body-associated protein, synphilin-1, tagged with the V5 epitope. Moreover, co-expression of the three proteins (tagged Sept4, α-synuclein, and synphilin-1) was sufficient to induce cell death. These data raise the possibility that Sept4 is involved in the formation of cytoplasmic inclusions as well as induction of cell death in α-synuclein-associated neurodegenerative disorders.

The importance of ␣-synuclein in neurodegeneration has been established by the findings that missense mutations in the ␣-synuclein gene cause familial PD (7,8) and that transgenic animals overexpressing ␣-synuclein or its mutants (A30P or A53T) show some phenotypes resembling PD (9 -12). LBs also contain ubiquitin (13) and ubiquitin carboxyl-terminal hydrolase-L1 (14), which may be involved in the pathogenesis of PD. Genetic studies on autosomal-recessive juvenile parkinsonism (AR-JP) led to the identification of the responsible gene, parkin, which encodes ubiquitin-protein isopeptide ligase (E3) (15). Loss-of-function mutations in parkin result in the accumulation of its substrates such as synphilin-1 (16), O-glycosylated ␣-synuclein (17), and Pael receptor (18) and cause neuronal damage without formation of visible inclusions. Synphilin-1 was first identified as an ␣-synuclein-binding protein (19) and was found to accumulate in LBs (20). Thus, impairment in the ubiquitin-proteasome system and accumulation of ␣-synuclein and other proteins in the neuronal cytoplasm may represent common molecular mechanisms underlying PD and AR-JP.
Septins are a family of filament-forming guanine nucleotidebinding proteins involved in cytokinesis, exocytosis, and other cellular processes (21)(22)(23)(24). At least 10 septin genes, Sept1-Sept10, have been identified in mouse as well as human genomes (25). Although septin family members can form co-polymers, their differential expression patterns in mammalian brains indicate some functional diversity among septins (26). Three septins (Sept1, Sept2, and Sept4) are accumulated in tau-based filamentous deposits known as neurofibrillary tangles and glial fibrils in Alzheimer's disease (27). A splice variant of the Sept4 gene encodes a mitochondrial protein called ARTS (apoptosis-related protein in the transforming growth factor-␤ signaling pathway), which mediates a pro-apoptotic signal (28). These data support the hypothesis that accumulation of a class of septins may accelerate neurodegeneration. In addition, Sept5 was found to be a substrate of Parkin (29). Sept5 is associated with ␥-aminobutyric acidergic synaptic vesicles (26), and excessive expression of Sept5 was found to interfere with regulated exocytosis (22). Thus, accumulation of Sept5 may affect dopamine release in the nigrostriatal system in AR-JP (29,30).
In this study, we have conducted an immunohistochemical examination of brain tissues from patients afflicted by Parkinson's disease or two other synucleinopathies. We found that Sept4 was co-localized with ␣-synuclein in LBs/GCIs in all cases. We also analyzed the physical and functional interaction among Sept4, ␣-synuclein, and synphilin-1 in cultured cells. Our data suggest that these proteins are involved in the formation of cytoplasmic inclusions as well as induction of cell death.

EXPERIMENTAL PROCEDURES
Tissues-Postmortem brain samples were obtained from five patients with PD (age: range, 66 -79 years; mean, 74.6 years), two with DLB (69 and 69.0 years), five with MSA (71-78 and 75.0 years), and four with non-neurological diseases (68 -81 and 74.5 years) from the Department of Neurology, Kyoto University Hospital. The brain specimens were used for neuropathological investigation after informed consent was obtained from the patients' relatives. The diagnoses of PD, DLB (pure form), and MSA were established on the basis of clinical and neuropathological data according to widely accepted criteria (31)(32)(33). After fixation with 4% paraformaldehyde, tissues were dissected. The midbrain and cerebellum were embedded in paraffin and sliced (6 m in thickness). The frontal lobe, pons, caudatoputamen, and thalamus were freeze-sectioned (20 m). Mouse brains were obtained from adult male C57Bl/6 mice (25-30 g) after deep anesthesia with sodium pentobarbital (50 mg/kg, intraperitoneal) and transcardial perfusion with phosphate-buffered saline (PBS), using procedures approved by the Animal Use and Care Committee of Kyoto University.
Plasmids-The coding region of human ␣-synuclein cDNA was cloned in-frame between the EcoRI and XhoI sites of the pCMV-Myc vector (Clontech) to express ␣-synuclein tagged with the Myc epitope at its amino terminus (Myc-␣-syn). The coding region of mouse Sept4 cDNA was cloned between the EagI and SalI sites of the pFLAG-CMV-2 vector (Sigma) to express Sept4 tagged with FLAG epitope at its amino terminus (FLAG-Sept4). The coding region of Sept2 was cloned between the HindIII and XbaI sites of the same vector to express FLAG-Sept2. The coding region of human synphilin-1 was cloned into pcDNA3.1/V5-His-TOPO expression vector (Invitrogen) to express synphilin-1 tagged with V5 epitope at its carboxyl terminus (V5-synphilin-1 (34)) (a generous gift from Dr. P. J. McLean). The missense mutations, A30P in ␣-synuclein and G154V in Sept4, were introduced into the respective wild type cDNAs using the QuikChange site-directed mutagenesis kit (Stratagene). The integrity of each insert cDNA was confirmed by DNA sequencing.
Cell Lines, Transfection, and Fractionation of Cellular Components-Two mouse cell lines, NIH3T3 and N18, were transiently transfected with expression vectors using LipofectAMINE-2000 (Invitrogen). After 24 or 48 h, the cells were rinsed with cold PBS and then lysed using Nonidet P-40 lysis buffer (50 mM Tris-HCl, pH 7.6, 1% Nonidet P-40, 150 mM NaCl, 5 mM EDTA). Fractionation of the subcellular components and subsequent immunoprecipitation were carried out as described previously (18) with minor modifications. In brief, the cell lysates were fractionated by centrifugation at 15,000 ϫ g at 4°C for 30 min, and each fraction was adjusted to the same volume with SDS sample buffer, sonicated (see below), and heat denatured. Equal volumes of these samples were subjected to immunoblot assay, and the amounts of immunoreactive proteins were estimated by densitometry using NIH Image software (version 1.62).
Co-immunoprecipitation-N18 cells (ϳ2 ϫ 10 5 ) transfected with 2 g of each plasmid were incubated for 24 h and fractionated as described above. The supernatant fraction was directly immunoprecipitated using anti-FLAG antibodies (Sigma) and protein G-Sepharose beads (Amersham Biosciences). Proteins in the Nonidet P-40-insoluble fraction were solubilized in 1% SDS with the aid of sonication for 20 s and diluted with 10 volumes of Nonidet P-40 lysis buffer. After centrifugation at 15,000 ϫ g at 4°C for 30 min, supernatants from the Nonidet P-40-insoluble fractions were immunoprecipitated. The beads bearing immune complexes were washed three times with 25 volumes of lysis buffer followed by four washings with PBS. The proteins on the beads were solubilized and analyzed by immunoblot assay. For co-immunoprecipitation of endogenous proteins, a human brain sample was homogenized in 4 volumes of PBS containing 320 mM sucrose, 0.1% Triton X-100, and 1 mM phenylmethylsulfonyl fluoride. After centrifugation at 37,000 ϫ g at 4°C for 20 min, the supernatant fraction was subjected to immunoprecipitation with anti-hemagglutinin (HA), anti-␣-synuclein (Affiniti Research Products, Exeter, United Kingdom), or anti-Sept4 antibodies. Following five washings with PBS, the precipitates were subjected to immunoblot assay using anti-␣-synuclein antibodies.
Immunofluorescent Staining of Cultured Cells and Scoring of Dead Cells-NIH3T3 or N18 cells transfected with three plasmids (each expressing Myc-␣-syn, FLAG-Sept4, or V5-synphilin-1) in various combinations were incubated in the absence or presence of 10 -20 M lactacystin for 24 h and immunostained as described previously (21). Mouse monoclonal antibodies against Myc (Sigma), V5 (Invitrogen), or FLAG or rabbit polyclonal antibodies against ␣-synuclein, FLAG (Sigma), or ubiquitin (Sigma) were used as primary antibodies. Fluorescein isothiocyanate-or rhodamine-conjugated anti-mouse IgG or anti-rabbit IgG was used as secondary antibodies. Dead cells were scored after brief staining with 0.2% trypan blue (Invitrogen) or with 1.0 g/ml annexin V-enhanced green fluorescent protein (Clontech) plus 2.5 g/ml propidium iodide (Clontech).
Ubiquitination Assay-NIH3T3 cells (ϳ5 ϫ 10 4 ) transfected with 1 g of FLAG-Sept4 expression vector were incubated in growth medium for 24 h and then in medium with or without 20 M lactacystin for an additional 16 h. Fractionated cell lysates were analyzed directly, or after immunoprecipitation with anti-FLAG antibodies, by immunoblot assay using antibodies against ubiquitin or FLAG. Statistics-Quantitative data were collected in at least three independent experiments. Significance of difference between data groups was assessed by Student's t test using StatView II software (version 5.0 for Macintosh, SAS Institute). p values smaller than 0.05 were considered significant.

Expression of Six Septins in Human and Mouse Brains-We
first examined the expression of six septins in the normal human frontal lobe and mouse whole brain by immunoblot assay using specific antibodies (Fig. 1). All six septins were detected in the human and mouse brains. The anti-Sept4 antibody (H5C-2) detected three major bands of 53, 51, and 49 kDa in the human brain; this antibody does not recognize the 32-kDa pro-apoptotic Sept4 variant, ARTS (28), because ARTS lacks the carboxyl-terminal region of Sept4 where the epitope for H5C-2 resides.
Sept4 Is Commonly Associated with ␣-Synuclein-based Cytoplasmic Inclusions-Having established the expression of six septins in the normal mouse brain and human frontal lobe, we next studied the expression of these septins in the postmortem brain samples from patients with neurodegenerative disorders. Notably, in all of the PD cases we tested (n ϭ 5), nigral LBs were found to be stained positive with the anti-Sept4 antibody (H5C-2) (Fig. 2, A and B). Preadsorption of H5C-2 with its cognate antigen abolished this staining (data not shown). Double staining of serial sections indicated that 70 -80% of ␣-synuclein-positive LBs were Sept4-positive (data not shown). Distribution of these two proteins, however, looked different; Sept4-staining was low in the corona of LB where ␣-synuclein staining was most intense (Fig. 2, C and D). Sept4 was also found in another type of ␣-synuclein-positive structures termed pale bodies (Fig. 2, C and D, arrowheads) (3). Such aberrant structures containing Sept4 were not found in nigral neurons in normal control brains (n ϭ 4; data not shown). All six septins examined are expressed in normal substantia nigra (26), and yet none of them except Sept4 was found to be accumulated in LBs, suggesting selective incorporation of Sept4 into LBs (data not shown).
Sept4 was also detected in the cortical and nigral LBs in brains with DLB (n ϭ 2, Fig. 2, E-G). The Sept4-positive cortical LBs frequently found in the small-to-medium-size, nonpyramidal neurons in deeper cortical layers lacked distinct core (Fig. 2, E and F), whereas the nigral LBs (Fig. 2G) were similar in morphology to those in PD (see above). Immunofluorescent double staining revealed that most of the ␣-synuclein-positive inclusions found in DLB brains were Sept4-positive. In cortical LBs, distribution of Sept4 and ␣-synuclein appeared to be tightly overlapped (Fig. 2, H-J). Again, none of the septins examined except Sept4 was found to accumulate in the nigral and cortical LBs in DLB brains.
Our previous studies indicate that Sept4 is expressed in both neurons and glial cells (26). Consistent with this observation, we found Sept4-positive GCIs in several regions including the cerebellum, pons, caudatoputamen, and thalamus in brains with MSA (n ϭ 5, Fig. 3). These findings demonstrate common and selective association of Sept4 with the inclusions in synucleinopathies.
Physical Interaction between Sept4 and ␣-Synuclein in the Brain and in Cultured Cells-To test the possible interaction between Sept4 and ␣-synuclein in vivo, human brain homogenate was immunoprecipitated with anti-HA (negative control), anti-␣-synuclein, or anti-Sept4 antibodies and then subjected to immunoblot analysis with anti-␣-synuclein antibodies (Fig.  4A). A fraction of ␣-synuclein was consistently pulled down with Sept4, indicating direct or indirect physical interaction between Sept4 and ␣-synuclein in vivo.
To analyze the molecular nature of this interaction, mouse cell lines (N18 neuroblastoma and NIH3T3 fibroblast) were co-transfected with a plasmid vector expressing ␣-synuclein tagged with Myc epitope (Myc-␣-syn) and a second vector expressing either the normal Sept4, a mutant Sept4 (Sept4 G154V ), or the normal Sept2, each tagged with the FLAG epitope (Fig.  4, B and C). We chose to use short epitope tags (Myc and FLAG) to reduce the probability of generating insoluble fusion proteins, as has been reported for the green fluorescent protein tag (34). Sept4 G154V has a missense mutation in its "P-loop" domain and shows reduced guanine nucleotide-binding activity (21,28,35). The transfected cells were lysed in a buffer containing 1% Nonidet P-40, and the subcellular components were separated into Nonidet P-40-soluble and -insoluble fractions. We subjected each fraction directly or after immunoprecipitation with anti-FLAG antibodies to the immunoblot assay using anti-Myc antibodies (to detect Myc-␣-syn) or anti-FLAG antibodies (to detect FLAG-septins). From the Nonidet P-40-soluble fraction, Myc-␣-syn co-precipitated with FLAG-Sept4 (Fig.  4B, middle panel, lane 4), indicating physical interaction between tagged ␣-synuclein and Sept4 in cultured cells. Myc-␣syn co-precipitated with FLAG-Sept4 G154V or FLAG-Sept2 was less than that co-precipitated with FLAG-Sept4 (Fig. 4B, middle panel; compare lanes 4, 6, and 8) even though the total amount of Myc-␣-syn in the sample of FLAG-Sept4 G154V -or FLAG-Sept2-transfected cells was comparable with, if not greater than, that in the sample of FLAG-Sept4-transfected cells (Fig. 4B, top panel; compare lanes 6 and 8 with lane 4). This may reflect the difference in the levels of FLAG-tagged septins present in these samples and/or the difference in their ability to be incorporated into the molecular complexes. Nevertheless, the wild type FLAG-Sept4 construct was the most efficient among the three septin constructs we tested in bring-  ing down Myc-␣-syn from the Nonidet P-40-soluble fraction.
FLAG-Sept4 and Myc-␣-syn Become Detergent-insoluble when Co-expressed-The above observations suggest that FLAG-Sept4 and Myc-␣-syn are soluble when expressed alone but both become insoluble when they are co-expressed. To confirm this model, NIH3T3 cells transfected with plasmids in various combinations were lysed in buffer containing 1% Nonidet P-40, and partition of each protein into the supernatant (S) and pellet (P) fractions was estimated by immunoblot assay (Fig. 5). Consistent with the above observation (Fig. 4C), coexpression of Myc-␣-syn brought more FLAG-Sept4 into the pellet fraction (Fig. 5, compare panels 1 and 2). In contrast, Myc-␣-syn showed little effects on the solubility of FLAG-Sept4 G154V or FLAG-Sept2 under these conditions (Fig. 5, compare panels 4 and 5, and panels 7 and 8). Reciprocally, coexpression of FLAG-Sept4, but not FLAG-Sept4 G154V or FLAG-Sept2, brought more Myc-␣-syn into the insoluble fraction (Fig.  5, compare panels 3, 6, and 9). These data demonstrate the bidirectional nature of the interaction between FLAG-Sept4 and Myc-␣-syn in bringing these proteins into the Nonidet P-40-insoluble fraction.
Localization and Biological Activity of Overexpressed FLAG-Sept4 and Myc-␣-syn in Cultured Cells-To determine the subcellular localization of FLAG-Sept4 and Myc-␣-syn, we performed immunofluorescent staining of transfected cells using anti-FLAG and anti-Myc antibodies (Fig. 6A). FLAG-Sept4 and Myc-␣-syn were diffusely distributed in the cytoplasm with apparent accumulation and co-localization at the cortex or periphery of the cells (Fig. 6A, top row). After treatment with a proteasome inhibitor, lactacystin, a subset (ϳ30%) of the cells contained punctate aggregates at the cell periphery and/or rods in the cytoplasmic protrusions (Fig. 6A, middle row). Interestingly, a fraction (ϳ5%) of doubly transfected, lactacystintreated cells contained large, double-positive structures reminiscent of LBs (Fig. 6A, bottom row). Such a structure was not found in cells expressing FLAG-Sept4 alone or Myc-␣-syn alone. Aggregates were rare in the cells expressing Myc-␣-syn alone (Fig. 6B), FLAG-Sept4 alone (Ͻ0.5%, data not shown), or Myc-␣-syn plus FLAG-Sept4 G154V or FLAG-Sept2 (Ͻ5%, data not shown).
Interestingly, a dye exclusion assay using trypan blue showed that the cells expressing both FLAG-Sept4 and Myc-␣syn were more sensitive to cytotoxicity by lactacystin as compared with the cells expressing one of the proteins alone or transfected with empty vectors (Fig. 6C). Separate experiments employing annexin V-enhanced green fluorescent protein and propidium iodide to stain dying cells also showed similar results (data not shown). Hence, FLAG-Sept4 and Myc-␣-syn were co-operative in making cells vulnerable to the stress induced by proteasome inhibition.
The also contained ubiquitin (Fig. 7A), and ubiquitinated FLAG-Sept4 was found predominantly in detergent-insoluble fractions (Fig. 7, B-E). Thus, FLAG-Sept4 can be ubiquitinated; this may occur in insoluble molecular complexes, or once ubiq-uitinated, FLAG-Sept4 may rapidly be incorporated into such complexes.
We also generated and tested the activities of a Myc-␣-syn expression plasmid harboring a missense mutation (A30P) that has been found in familial PD cases (8). Interestingly, the amounts of FLAG-Sept4 and Myc-␣-syn in the anti-FLAGimmunoprecipitates from Nonidet P-40-soluble fractions were slightly but consistently greater when Myc-␣-syn A30P was used (Fig. 8A). Furthermore, cells co-expressing FLAG-Sept4 and Myc-␣-syn A30P were more sensitive to lactacystin toxicity than cells expressing FLAG-Sept4 plus wild type Myc-␣-syn (Fig. 8C).
Effects of V5-Synphilin-1-It has been reported that synphilin-1 induces cytoplasmic inclusions when co-expressed with ␣-synuclein in HEK293 cells (19). When Myc-␣-syn and synphilin-1 tagged with another oligopeptide epitope (V5) were co-expressed in NIH3T3 cells, ␣-syn/V5 double-positive cyto-  1  and 3). On the immunoblots treated with anti-FLAG antibodies (to detect Sept4), broad smears in the high molecular mass range, which probably represent FLAG-Sept4 molecules ubiquitinated to various degrees, were found only in the Nonidet P-40-insoluble fraction (C, lane 3). The discrete bands in the high molecular mass range of the Nonidet P-40-soluble fraction (C, lanes 1 and 2) are probably nonspecific, because they were also seen in vector-transfectants (data not shown). Immunoprecipitation of FLAG-Sept4 with anti-FLAG antibodies followed by immunoblot detection with anti-ubiquitin (D) or anti-FLAG (E) more clearly demonstrated that lactacystin treatment increases the amount of ubiquitinated FLAG-Sept4 only in the Nonidet P-40-insoluble fractions (D and E, lane 3). plasmic inclusions formed (Fig. 9A), supporting the previous finding. Interestingly, when FLAG-Sept4 was additionally expressed in this system, larger inclusions with LB-like morphology formed (Fig. 9B, upper panels). These round inclusions were frequently accompanied by satellite inclusions with irregular shapes (Fig. 9B, lower panels), which are reminiscent of pale bodies, putative precursors of LB found in PD neurons (see Fig. 2, C and D) (3). These irregular inclusions were FLAG (Sept4)-positive but V5 (synphilin-1)-negative (Fig. 9B, lower panels). FLAG-Sept4 and V5-synphilin-1 could not be co-immunoprecipitated, and they did not form inclusions in the absence of Myc-␣-syn (data not shown).
Expression of V5-synphilin-1 itself was cytotoxic, especially when combined with lactacystin treatment (Fig. 9C, compare bars 2 and 4; p ϭ 0.0037). Although additional expression of Myc-␣-syn had little effect on cytotoxicity (Fig. 9C, compare bars 3 and 4 with bars 5 and 6), additional expression of Myc-␣-syn plus FLAG-Sept4 resulted in significant enhancement of cytotoxicity, especially when combined with lactacystin treatment (Fig. 9C; compare bars 4 and 8, p ϭ 0.046). The effect of V5-synphilin-1 in enhancing the cytotoxicity exerted by FLAG-Sept4 and Myc-␣-syn was significant both in the absence and presence of lactacystin ( Fig. 9C; compare bars 7 and 9, p ϭ 0.029, and bars 8 and 10, p ϭ 0.015). Taken together, these findings raise the possibility that Sept4 has the potential to cooperate with ␣-synuclein and synphilin-1 to induce cytoplasmic inclusions and cell death. DISCUSSION An important finding in this study is that Sept4 is accumulated in cytoplasmic inclusions found in three major synucle-inopathies, PD, DLB, and MSA. In nigral LBs, Sept4 is found predominantly in the central cores and ␣-synuclein in the peripheral portions. Possible explanations for this finding include: (i) indirect association between Sept4 and ␣-synuclein; (ii) selective association of one of the proteins with a subset of the other due, for instance, to differential modification or conformational change; and (iii) epitope masking in specific area of LBs. In cultured cells, FLAG-Sept4 and Myc-␣-syn were cooperatively incorporated into the Nonidet P-40-insoluble fraction, and upon proteasome inhibition or expression of V5-synphilin-1, they participated in the formation of LB-like cytoplasmic inclusions and promotion of cell death. Taken together, our findings in vivo and in vitro support the idea that endogenous Sept4 and ␣-synuclein are actively involved in LB formation and promotion of cell death. Furthermore, our finding that pale body-like cytoplasmic aggregates did not contain V5-synphilin-1 (Fig. 9) suggests that interaction between Sept4 and ␣-synuclein may be an early event during LB formation. In fact, a recent report indicates that synphilin-1 is negative in pale bodies (36).
Results with overexpression of tagged proteins in cultured cells cannot be directly extrapolated to the pathological conditions in the brain. Such experiments in vitro, however, have their own advantages, and if supported by pathological evidence, they may shed light on the role of particular molecules in human disorders. The most obvious and important limitation of pathological evidence is that it completely depends on the availability of samples and that it is always retrospective. Furthermore, immunohistological evidence heavily depends on the quality of the antibodies used; thus the possibility of crossreaction or epitope masking is often hard to exclude. Antibodies against epitope tags (e.g. FLAG, Myc, and V5) are high in affinity and specificity, so that clear-cut experimental evidence can be obtained. Although the effects of the extra peptide on the conformation of tagged protein cannot be completely excluded, we found no evidence that the tagged proteins tended to be insoluble when expressed alone. Our immunohistochemical evidence (Figs. 2 and 3) as well as the data obtained with specific antibodies (Fig. 4A) make it further unlikely that the observed interactions between the tagged proteins represent mere artifacts.
The primary structure of Sept4 is unique among septins in that it has an extended amino-terminal region (37). The largest form of Sept4 (478 amino acid residues) is more than 100 residues longer than Sept5 (369 residues). This extended region, which shares no homology with any other protein in the current data base, may confer unique biochemical properties on Sept4. In contrast, the remaining portion of Sept4 is highly homologous (78% identical) to Sept5, a known target for Parkin (29,30). Our data indicate that upon proteasome inhibition, Myc-␣-syn and ubiquitinated FLAG-Sept4 accumulate in the detergent-insoluble fraction (Fig. 7), which probably corresponds, at least in part, to the visible cytoplasmic inclusions (Fig. 6). Whether the FLAG-Sept4 ubiquitination in this system is mediated by endogenous Parkin or by some other ubiquitin protein ligase remains to be clarified.
Sept4 and synphillin-1 share three properties in common: they can be co-immunoprecipitated with ␣-synuclein (19), ubiquitinated by Parkin (16) or unknown ubiquitin ligase, and concentrated in the LB cores (20). In the cell culture model, we demonstrated that FLAG-Sept4 and V5-synphilin-1 can colocalize in the Myc-␣-syn-based cytoplasmic inclusions and that the three proteins can synergistically exert cytotoxicity. The relationship between inclusion body formation and cell death in synucleinopathies is currently unknown, but we can envisage two possibilities: 1) Sept4, ␣-synuclein, and synphilin-1 syner-FIG. 8. Co-immunoprecipitation of Sept4 and wild type or A30P ␣-synuclein. A and B, NIH3T3 cells were co-transfected with the indicated expression vectors. After a 36-h incubation, the cells were lysed with buffer containing 1% Nonidet P-40 and fractionated into soluble (A) and insoluble (B) fractions. Each fraction was subjected to immunoprecipitation with anti-FLAG antibodies (IP FLAG) followed by immunoblot detection using anti-Myc to detect Myc-␣-synuclein (middle panels) or anti-FLAG to detect FLAG-Sept4 (lower panels). The asterisk indicates the position of the immunoglobulin heavy chain. These experiments were repeated twice with similar results. C, NIH3T3 cells co-transfected with the indicated expression vectors were treated with lactacystin for 24 h and then stained with trypan blue. The proportion of blue cells among a total of 1000 cells was scored under a microscope. Error bars represent S.D. from three independent experiments. *, p ϭ 0.045 comparing cells expressing wild type and A30P ␣-synuclein. The expression levels of the total FLAG-Sept4 and Myc-␣syn proteins, as detected by immunoblot assay using anti-FLAG and anti-Myc antibodies, respectively (bottom two panels), were not greatly different between the two transfectant populations. gistically form insoluble aggregates or inclusions that trigger cell death; or 2) Sept4, ␣-synuclein, and synphilin-1 form soluble toxic complexes that trigger cell death unless enclosed into inclusions. The fact that AR-JP shows earlier onset than the other types of PD and that neurons affected in AR-JP do not contain LBs (38) is consistent with the second model. Recent reports that soluble ␣-synuclein complexes are linked to neurotoxicity in dopaminergic neurons (39) and that inclusion body formation and cell death are mutually exclusive phenomena in PC12 cells (40) also support this model. Our data on the Myc-␣-syn A30P mutant (Fig. 8) indicate an apparent correlation between the increase in Nonidet P-40-soluble FLAG-Sept4⅐Myc-␣-syn complexes and cytotoxicity, raising the possibility that the Nonidet P-40-insoluble material itself is not the toxic entity. This finding may provide an additional support to the second model. However, ␣-synuclein aggregates/inclusions seem to mechanically damage neurons in an ␣-synuclein transgenic mice model (10 -12, 41). More indirect mechanisms are also possible; accumulated Sept4, ␣-synuclein, and synphilin-1 could sequester their ubiquitin-protein ligases (e.g. Parkin) into the inclusions (42). Alternatively, sequestration of Sept4 from the cytoplasm to inclusions could affect neuronal function by destabilizing actin cytoskeleton (24). In fact, impaired actin turnover is implicated in neurodegenerative diseases such as primary dystonia (43) and Alzheimer's disease (44). These hypotheses based on our findings should be tested in vivo in future studies.
We reported previously that Sept4 and two other septins are present in the tau-based neurofibrillary tangles in Alzheimer's disease (27). Because PD and Alzheimer's disease show considerable overlap in their clinicopathological features, it has been proposed that common molecular mechanisms may underlie synucleinopathies and tauopathies (41,45). In this regard, Sept4 is unique in its association with the aberrant protein depositions in both types of diseases. Further studies on the biochemical properties and physiological functions of Sept4 may therefore provide important insights into the common mechanism underlying diverse neurodegenerative disorders.