Commitment of 1-Methyl-4-phenylpyrinidinium Ion-induced Neuronal Cell Death by Proteasome-mediated Degradation of p35 Cyclin-dependent Kinase 5 Activator*

The dysfunction of proteasomes and mitochondria has been implicated in the pathogenesis of Parkinson disease. However, the mechanism by which this dysfunction causes neuronal cell death is unknown. We studied the role of cyclin-dependent kinase 5 (Cdk5)-p35 in the neuronal cell death induced by 1-methyl-4-phenylpyrinidinium ion (MPP+), which has been used as an in vitro model of Parkinson disease. When cultured neurons were treated with 100 μm MPP+, p35 was degraded by proteasomes at 3 h, much earlier than the neurons underwent cell death at 12–24 h. The degradation of p35 was accompanied by the down-regulation of Cdk5 activity. We looked for the primary target of MPP+ that triggered the proteasome-mediated degradation of p35. MPP+ treatment for 3 h induced the fragmentation of the mitochondria, reduced complex I activity of the respiratory chain without affecting ATP levels, and impaired the mitochondrial import system. The dysfunction of the mitochondrial import system is suggested to up-regulate proteasome activity, leading to the ubiquitin-independent degradation of p35. The overexpression of p35 attenuated MPP+-induced neuronal cell death. In contrast, depletion of p35 with short hairpin RNA not only induced cell death but also sensitized to MPP+ treatment. These results indicate that a brief MPP+ treatment triggers the delayed neuronal cell death by the down-regulation of Cdk5 activity via mitochondrial dysfunction-induced up-regulation of proteasome activity. We propose a role for Cdk5-p35 as a survival factor in countering MPP+-induced neuronal cell death.

Parkinson disease (PD) 3 is the second most common neurodegenerative disease, characterized pathologically by degenerated dopaminergic neurons and ubiquitin-positive aggregates known as Lewy bodies (1). Most cases of PD are sporadic, but a small proportion of patients with PD have the familial form. Several causative genes have been identified for familial PDs, including ␣-synuclein (2), ubiquitin C-terminal hydrolase L1 (UCH-L1) (3), and parkin, an ubiquitin ligase E3 of the ubiquitin-proteasome system (4), implicating the impairment of the ubiquitin-proteasome pathway in the pathogenesis of PD. However, the mechanisms underlying the involvement of the ubiquitin-proteasome system in the development of PD are not yet understood.
The 1-methyl-4-phenylpyrinidinium ion (MPP ϩ ), a toxic metabolite of 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP), is a neurotoxin used widely to induce dopaminergic neuronal cell death in in vitro models of PD (5). Previous studies have indicated that MPP ϩ induces neuronal cell death via several pathways, including the inhibition of complex I activity of the respiratory chain in mitochondria, leading to energy depletion, protein peroxidation, and DNA damage by producing reactive oxygen species and the induction of cytotoxic glutamate secretion (6,7). However, the precise molecular pathway resulting in neuronal cell death remains to be identified.
Cyclin-dependent kinase 5 (Cdk5) is a member of the Cdk serine/threonine kinase family. Cdk5 plays a role in a variety of neuronal activities including neuronal migration during central nervous system development (8,9), synaptic activity in matured neurons (10), and neuronal cell death in neurodegenerative diseases (11,12). Generally, when Cdk5 are activated by their respective activator cyclins, they function in cell cycle progression. However, unlike those cell cycle Cdk5, the kinase activity of Cdk5 is detected mainly in post mitotic neurons. This is because Cdk5 activators p35 and p39 are expressed predominantly in neurons (13,14). The amount of p35 is the major determinant of Cdk5 activity, and it is normally a short-lived protein degraded by the ubiquitin-proteasome pathway (15,16). However, in stressed neurons, the calcium-activated protease calpain cleaves p35 to the more stable and active form, p25 (17)(18)(19)(20)(21). Hyperactivated or mislocalized Cdk5-p25 has been implicated in the pathogenesis of numerous neurodegenerative disorders including PD and Alzheimer disease. In the case of PD, Cdk5 and p35 are found in the Lewy bodies of the dopaminergic neurons of the brain (22,23). Cdk5 is activated by p25 and is required for cell death in mouse models of PD induced with MPTP (24) or 6-hydroxydopamine (25). It has been shown that Cdk5-p25 in MPTP-treated neurons phosphorylates the survival factor, myocyte enhancer factor 2 (MEF2), to inactivate it, leading to cell death (26,27). However, further studies are required to clarify the involvement of p35 metabolism in the PD pathway.
Contrary to its role in cell death progression, recent studies have also suggested a survival function for Cdk5 in maintaining survival signals or counteracting apoptotic signals. For example, Cdk5 inhibits c-Jun phosphorylation by c-Jun-N-terminal protein kinase 3, which is activated by UV irradiation (28). Cdk5 also promotes the survival of neurons by activating Akt through the well known neuregulin/phosphatidylinositol 3-kinase (PI3K) survival pathway, which leads to the down-regulation of proapoptotic factors (29). Cdk5 attenuates cell death either by up-regulating Bcl-2 through the phosphorylation of ERK (30) or by phosphorylating Bcl-2 to maintain its neuroprotective effect (31). However, whether Cdk5 acts as the anti-apoptotic factor in the PD model of neuronal cell death has not been determined.
Here, we studied the role of Cdk5-p35 in the cell death of neurons treated with MPP ϩ . We found that p35 was proteolysed in cultured neurons by either calpain or proteasomes depending on the concentration of MPP ϩ used. The proteasomal MPP ϩ -induced degradation of p35 occurred earlier and at lower MPP ϩ concentrations than did its cleavage by calpain. MPP ϩ up-regulated the overall proteasome activity in the neurons by impairing the mitochondrial protein import system. A brief MPP ϩ treatment for up to ϳ3 h was sufficient to induce delayed cell death at 24 h. The overexpression of p35 suppressed this MPP ϩ -induced cell death, and depletion of p35 increased cell death. Together, these results implicate a role for Cdk5-p35 as a survival factor in MPP ϩ -treated neurons.
Construction and Production of Lentivirus shRNA-The small interfering RNA (siRNA) targeting sequence, 5Ј-CAGC-TACCAGAGCAACATCGC-3Ј, for mouse and rat p35 were subcloned into the BamH⌱ and EcoR⌱ sites of pSIH1-H1-copGFP shRNA cloning vector. pSIH1-siLuc-copGFP encoding shRNA against luciferase was used as control construct. Recombinant viruses were produced by transient transfection into HEK293TN cells by pPACKH1 packaging plasmid mix according to the manufacturer's protocol. Infectious lentiviruses were collected after 72 h. The efficiency of knockdown was assessed by Western blotting.
Primary Cortical Neuron Culture and Transfection-All animals used in this study were housed in a temperature-controlled room under a 12-h light/12-h dark cycle with free access to food and water. All of the experiments were performed in compliance with relevant laws and institutional guidelines. Brain cortices from embryonic day 15-16 mice (ICR mice; Sankyo Laboratory Service, Tokyo, Japan) were dissected and plated at a density of 1.2 ϫ 10 6 cells/cm 2 in polyethyleneiminecoated dishes or plates with glass coverslips in Dulbecco's modified Eagle's medium and Ham's F-12 (1:1) supplemented with 5% fetal bovine serum and 5% horse serum. The medium was then changed to neurobasal medium supplemented with B-27 (Invitrogen) and 1 mM L-glutamine. FLAG-tagged p35 was expressed in neurons on day 4 in vitro (DIV) using an adenoviral vector as described previously (34). Lentivirus shRNA vector against p35 or luciferase was infected to neurons on DIV3. Transient transfection with the pAcGFP1-mito or pAcGFP1-N1 vector was performed at DIV6 using Lipofectamine 2000 reagent (Invitrogen) according to the manufacturer's protocol. Pharmacological experiments were performed on DIV6. Neurons were harvested with cold phosphate-buffered saline (PBS) by centrifugation at 1,000 ϫ g for 5 min, and the pellets were stored at Ϫ80°C until use.
Cell Death Detection Assay-Lactate dehydrogenase activity was measured using an LDH Cytotoxic Assay Kit (Wako Chemicals, Osaka, Japan) according to manufacturer's instructions. For the nuclear condensation assay, the neurons cultured on glass coverslips were fixed with 4% paraformaldehyde for 20 min, washed twice with PBS, and stained with DAPI for 10 min. The nuclei were observed under a Zeiss LSM5 confocal microscope (Carl Zeiss, Oberkochen, Germany), and the number of condensed nuclei was counted and expressed as a ratio of the total number of cells.
Reverse Transcription-Polymerase Chain Reaction-Total RNA was isolated from cultured neurons using Isogen RNA extraction kit (Nippon Gene, Tokyo, Japan). Single-stranded cDNAs were generated from 1 g of total RNA using Superscript First-strand synthesis system for reverse transcription-PCR (Invitrogen) according to manufacturer's instructions. PCR was carried out with following primers: 5Ј-GCTCTG-CAGGGATGTTATCTCC-3Ј and 5Ј-CTTCTTGTCCTCCT-GACCACTC-3Ј for p35 and 5Ј-ATGGTGAAGGTCGGTGT-GAACG-3Ј and 5Ј-TGGTGAAGACGCCAGTAGACTC-3Ј for glyceraldehyde-3-phosphate dehydrogenase. The amount of cDNAs and the number of cycles were chosen with in a linear range of amplification.
Preparation of Whole Cell Extract and Western Blotting-Neurons were sonicated in sample buffer (30 mM Tris-HCl, pH 6.8, 5% glycerol, 1.6% SDS, and 2.5% ␤-mercaptoethanol) and boiled for 5 min. Protein concentrations were determined with Coomassie protein assay reagent (Pierce) using bovine serum albumin as the standard. Equal amounts of each protein sample were applied to a 10 or 12.5% polyacrylamide gel for SDS-PAGE. Tris/Tricine SDS-PAGE with 10% acrylamide gel was used to detect the phosphorylation status of p35 (24).
Visualization of Mitochondria-Mitochondria in the neurons were visualized by either immunofluorescent staining with anti-Tom20 antibody or transfection with either pAcGFP1mito or pSu9-EGFP. Fluorescence was monitored and images were taken using a Zeiss LSM5 confocal microscope. For the quantification of mitochondrial phenotype, segmented mitochondria after MPP ϩ treatment was scored when at least 80% of the filamentous mitochondria were disintegrated. For the pAcGFP1-mito transfected neurons, the total fluorescence intensity of the mitochondria in each cell body of the GFPpositive neurons was measured, and the mean fluorescence intensity (total fluorescence intensity/cell body area) was evaluated using ZEN imaging software. GFP-mito-positive neurons were classified into three levels according to the mean GFP fluorescence intensity in the mitochondria.
Glycerol Gradient and Measurement of Proteasome Activity-Cell extract was collected by centrifugation at 14,000 ϫ g for 30 min after brief sonication in extraction buffer (20 mM Tris-HCl, pH 7.5, 2 mM ATP, 5 mM MgCl, 1 mM DTT). Equal amounts of protein from each sample were used for the assay. For glycerol gradient, 0.7 mg of protein was loaded on top of a linear 35-10% (v/v) glycerol in extraction buffer and centrifuged at 83,000 ϫ g for 22 h at 4°C with a RPS-40T swing rotor (Hitachi, Tokyo, Japan), and fractions were collected. Each extract and fraction was incubated in reaction buffer (50 mM Tris-HCl, pH 8.5, 1 mM EDTA, 1 mM ATP, 4.5 M Suc-LLVY-AMC) for 0.5 or 3.5 h at 37°C. The fluorescence of the resulting cleaved substrate peptide was measured using a fluorescence spectrophotometer (RF-5000, Shimadzu, Tokyo) with emission at 440 nm and excitation at 350 nm.
Subcellular Fractionation-Cultured neurons were disrupted in fractionation buffer (20 mM Hepes, pH 7.5, 150 mM NaCl, 20 mM KCl, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, 0.32 M sucrose, and protease inhibitors) by 20 passages through a 27-gauge syringe. The homogenates were then centrifuged at 1,000 ϫ g for 5 min. The supernatant was centrifuged twice more at 16,000 ϫ g for 10 min, and the final supernatant was used as the cytoplasmic fraction. The pellets were washed once with Hepes buffer and collected by centrifugation at 16,000 ϫ g for 20 min. The pellet was used as the mitochondria-enriched fraction after brief sonication in fractionation buffer. Fractionation was confirmed by Western blotting with anti-Tom20, anti-cytochrome c, anti-GFP, and anti-actin antibodies.
Measurement of Complex I Activity-The collected neurons were briefly sonicated in isolation buffer (20 mM Hepes, pH 7.4, 10 mM KCl, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, 0.32 M sucrose, and protease inhibitors). The homogenates were centrifuged at 1,000 ϫ g for 10 min, and the supernatants were centrifuged again at 9,000 ϫ g for 15 min to isolate the mitochondria-enriched fraction. The pellets were suspended with mitochondrial isolation buffer. The mitochondria-enriched fraction (30 g of protein) was added to assay buffer (25 mM Hepes, pH 7.4, 8 mM sodium phosphate, 10 mM MgSO 4 , 136 mM NaCl, 1.5 mM KH 2 PO 4 , 1 mM KCl, 0.25% bovine serum albumin, 150 M NADH, 50 M decylubiquinone). The decrease in absorbance at 340 nm, which represents the oxidation of NADH, was measured after incubation for 7 min.
Determination of ATP Levels-ATP levels were measured by a firefly bioluminescence assay. The neurons were lysed as whole cell extracts by sonication in 50 mM Tris-HCl (pH 7.4) and centrifuged at 1,000 ϫ g for 10 min, and the supernatants were collected. Cell extracts with the same amount of protein were added to the measurement buffer (0.3 M Tris-HCl, pH 7.4, 0.2 M glycine, 20 mM MgSO 4 , 2 mM EDTA, 0.04 mg/ml luciferase, 0.25 mM luciferin) and incubated for 2 min. Luminescence was detected with the ChemiDoc XRS System (Bio-Rad Laboratories).
Statistical Evaluation-All of the experiments were performed a minimum of three times with similar results; representative results are shown in the Figs. 1-8. The statistical significance of the data was tested using Student's t test to compare the two conditions. Differences were considered to be significant at p Ͻ 0.05 (*) or p Ͻ 0.01 (**).

Degradation of p35 Is Induced by MPP ϩ before Neuronal Cell
Death-We first examined the cytotoxicity of MPP ϩ using cultured cortical and striatal neurons. Cortical neurons were treated with various concentrations of MPP ϩ (100 M to 1 mM) for 0 to 24 h. In these cells, lactate dehydrogenase activity increased slightly at 3 and 6 h and markedly at 12 and 24 h with higher concentrations of MPP ϩ (Ͼ 250 M) (Fig. 1A). When we measured apoptotic cell death by nuclear condensation, neurons undergoing cell death increased to 40% after 6 h of treatment with 1 mM MPP ϩ (Fig. 1B). However, at lower concentra-tions of MPP ϩ (100 M), no increase in lactate dehydrogenase activity or in apoptotic nuclei was seen until 12 h after treatment ( Fig. 1, A and B). We evaluated the effect of MPP ϩ on p35 expression by Western blotting. p35 was cleaved to p25 in the 500 M and 1 mM MPP ϩ treatments. p25 was detected 6 h after the 1 mM MPP ϩ treatment, when apoptotic nuclei were observed (Fig. 1C). In contrast, 100 M MPP ϩ reduced p35 dramatically at 3 h, without cleavage to p25, before any neuronal death was detected. Further, to see the effect of MPP ϩ on the reduction in p35, neurons were treated with lower concentrations of MPP ϩ (5-100 M). The reduction in p35 was observed even at 5 M MPP ϩ (supplemental Fig. S1A). Similar results were observed with cultured striatal neurons (supplemental Fig. S2, A and B).
To see whether the reduction in p35 was induced at transcription levels, we measured mRNA transcript level by reverse transcription-PCR. p35 mRNA remained unaffected at 3 h MPP ϩ treatment (Fig. 1D), indicating that p35 is reduced post-translationally. In all cases, the expression of Cdk5 remained unchanged (Fig. 1C), but a reduction in p35 was expected to be accompanied by the down-regulation of the kinase activity of Cdk5. We used the phosphorylation status of tau protein at Ser-404, a Cdk5 phosphorylation site, to assess the in situ kinase activity of Cdk5. Concomitant with the decrease in p35, Ser-404 phosphorylation decreased without any change in the amount of total tau protein (Fig. 1E). These results indicate that Cdk5 activity is down-regulated by MPP ϩ through the degradation of p35.
Whereas MPP ϩ -induced cleavage of p35 to p25 is well documented (18 -21), its MPP ϩ -induced degradation has not yet been reported. Before identifying the molecular mechanism of the MPP ϩdependent degradation of p35, we investigated the physiological significance of brief MPP ϩ treatment. First, we assessed the in vivo effect of MPP ϩ on levels of p35 by injecting MPP ϩ directly into the cerebral cortex or striatum of juvenile rat brains. p35 decreased in the both cortex and striatum without the generation of p25 3 h after MPP ϩ injection ( Fig. 2A). We also examined the effect of short term exposure to MPP ϩ on neuronal cell death by replacing the culture medium with medium lacking MPP ϩ 3 h after treatment with MPP ϩ . Even after the MPP ϩ was washed out, p35 continued to decrease to the levels observed in neurons treated for 6 or 21 h ( Fig. 2B; also see supplemental Fig. S1B). Although the neurons did not undergo cell death at 3 h in the presence of 100 M MPP ϩ , they underwent cell death when they were cultured for an additional 18 h in the absence of MPP ϩ (Fig. 2C). These results suggest that a brief MPP ϩ treatment is sufficient to induce the delayed cell death of cultured neurons at around 24 h.
MPP ϩ Induces Proteasome-and Calpain-dependent Proteolysis of p35 in a Concentration-and Time-dependent Manner-Cortical neurons were treated with MPP ϩ in the presence of either a proteasome inhibitor (MG132 or epoxomicin) or a calpain inhibitor (ALLN or calpain inhibitor IV), to determine which protease is responsible for the reduction in p35 by proteolysis. The reduction in p35 without cleavage to p25 in 100 Cell lysates from 0, 100, 250, and 500 M and 1 mM MPP ϩ -treated neurons were subjected to SDS-PAGE followed by Western blotting with anti-p35 antibody C19 for p35 and p25, anti-Cdk5 antibody DC17, and anti-actin antibody. Actin was used as the loading control. D, p35 mRNA levels in MPP ϩ -treated neurons. Total RNA was isolated from neurons treated with 100 M MPP ϩ or PBS for 3 h. p35 mRNA levels were analyzed by reverse transcription-PCR using glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as an internal control (left panel). Quantification is shown in the right panel. Bars represent the means Ϯ S.E. of three independent experiments (n ϭ 3). E, in situ Cdk5 activity assessed by the phosphorylation of tau at Ser-404. Neurons were treated with 100 M MPP ϩ for 0 to 6 h, and the phosphorylation of tau at Ser-404, total tau, p35, and Cdk5 were detected by Western blotting (left panel). Actin was used as the loading control. Right panel, quantification of pSer-404. Ratio of pSer-404 is expressed as a percent of that in the pretreatment after normalization against total tau. Bars represent the means Ϯ S.E. of three independent experiments (n ϭ 3; *, p Ͻ 0.05; **, p Ͻ 0.01; Student's t test).
M MPP ϩ was suppressed by the proteasome inhibitors but not by the calpain inhibitors (Fig. 3A), indicating that the degradation of p35 induced by 100 M MPP ϩ is proteasome-dependent. In contrast, the conversion of p35 to p25 induced by 1 mM MPP ϩ was suppressed by calpain inhibitors (Fig. 3B). The latter result is consistent with previous reports that p35 is cleaved to p25 by calpain when neuronal cell death is induced by MPP ϩ or MPTP (26).
Neither Oxidative Stress nor Glutamate Is Involved in the Proteasomal Degradation of p35 Induced by MPP ϩ -MPP ϩ is an oxidative stressor (7). We suspected that proteins such as p35, which have a rapid turnover rate, would be easily damaged by oxidation or nitration followed by proteasomal degradation. To explore this possibility, neurons were treated with MPP ϩ in the presence of an antioxidant, either NAC or PBN. Both reagents failed to suppress the degradation of p35 (Fig. 4A). Furthermore, neither TRIM nor L-NAME, both of which are nitric-oxide synthase inhibitors, prevented the degradation of p35 (Fig. 4B). These results indicate that oxidative stress is not a mediator of the MPP ϩ -induced degradation of p35.
The degradation of p35 is induced in cortical neurons by glutamate treatment (35). Because MPP ϩ treatment stimulates glutamate secretion (36), we examined whether increased glutamate secretion mediates p35 degradation. First, we confirmed in this study that p35 degradation is induced with NMDA, an agonist of the ionotropic glutamate receptor, and that its degradation is suppressed by AP5, an antagonist of the NMDA receptor (supplemental Fig. S3, A and B). We then tested the effect of AP5 on MPP ϩ -induced p35 degradation. AP5 did not suppress the degradation of p35, suggesting that the glutamate pathway is not involved in MPP ϩ -induced p35 degradation (Fig. 4C).
Impaired Complex I Activity Is Associated with the Proteasomal Degradation of p35-MPP ϩ is an inhibitor of complex I in the mitochondria. To evaluate the relationship between complex I inhibition and the degradation of p35, we used rotenone, a more specific complex I inhibitor. Rotenone at concentrations of 40 to 100 nM induced the degradation of p35 as did MPP ϩ , although higher concentrations of 250 or 400 nM rotenone also induced p35 cleavage to p25 (Fig. 5A). Having observed that a mitochondrial defect might be associated with the degradation of p35, we examined the effect of MPP ϩ on mitochondrial integrity. We first checked the mitochondrial morphology, which is an indication of mitochondrial impairment (37). After MPP ϩ treatment for 3 h, the tubular and filamentous forms of the mitochondria had changed to a fragmented form (Fig. 5, B and C). It is noteworthy that the fragmented mitochondria were still distributed in the extended neurites, suggesting that the neurites are not affected, at least morphologically, by MPP ϩ at this time. Because the fragmentation of mitochondria is a hallmark of mitochondria-dependent apoptosis, we investigated whether cytochrome c, a proapoptotic factor, is released from mitochondria into the cytoplasm with MPP ϩ treatment. Cultured neurons were subfractionated into cytosolic and mitochondria-enriched fractions, and cytochrome c was detected by Western blotting. Cytochrome c was not increased in the cytosol of neurons treated with MPP ϩ for 1 or 3 h (Fig. 5D), suggesting that mitochondrial disintegration had not yet occurred at 3 h. These results also strongly support the assumption that the degradation of p35 takes place before neurons undergo cell death.
Because we considered that the mild impairment of complex I by MPP ϩ would trigger the degradation of p35, we measured complex I activity of the respiratory chain by quantifying NADH oxidation. The oxidative activity decreased to ϳ70 and 40% of control value after treatment with MPP ϩ for 1 and 3 h, respectively (Fig. 5E). Because complex I activity is an indispensable part of ATP production in the respiratory system, its inhibition would result in ATP depletion. Conversely, ATP is needed for proteasome-dependent protein degradation. These apparently paradoxical results forced us to measure the ATP concentration when p35 is degraded by proteasomes. The ATP levels were maintained at almost the control levels in neurons treated with MPP ϩ for 1 or 3 h (Fig. 5F). These results suggest that neurons can still perform ATP-dependent proteolysis by proteasomes after a short exposure to MPP ϩ , even though complex I activity is impaired to some extent.
Dysfunction of the Mitochondrial Import System Leads to the Up-regulation of Total Proteasome Activity-Mitochondrial preproteins are degraded in the cytoplasm by proteasomes when they are not properly imported into the mitochondria (38). We assumed that a brief treatment with MPP ϩ would impair the mitochondrial protein import system, resulting in the accumulation of mitochondrial proteins in the cytoplasm and the subsequent activation of proteasomes. To verify this assumption, cortical neurons were transfected with the pAcGFP1-mito vector, which encodes GFP tagged with the mitochondrial targeting sequence of subunit VIII of human cytochrome c oxidase at the N terminus. In the control neurons, GFP-mito was brightly visible in the mitochondria 12 h after transfection. However, no accumulation of GFP-mito in the mitochondria was observed in the MPP ϩ -treated neurons (Fig. 6, A  and B), supporting our assumption of impaired protein import. We confirmed this result biochemically. GFP-mito was detected predominantly in the mitochondrial fraction of the control neurons at 12 h but decreased markedly when the neurons were treated with MPP ϩ (Fig.  6C). Similar results were obtained with EGFP tagged with the mitochondrial targeting sequence of ATPase subunit 9 (data not shown). Curiously, however, GFP-mito was not detected in the cytosolic fraction in either the control or MPP ϩtreated neurons. We suspected that MPP ϩ inhibited overall protein expression. However, this was shown to be unlikely by the fact that the expression of the control GFP from the same pAcGFP1-N1 plasmid was not affected by MPP ϩ (supplemental Fig. S4). Because mitochondrial proteins are degraded very quickly in the cytoplasm when they are not imported into mitochondria, we concluded that GFPmito might not persist long enough to be detected in the cytosolic fraction.
We then investigated whether the dysfunction of the mitochondrial import system induces the specific degradation of p35 by ubiquitination. Ubiquitination is the process by which target proteins are covalently attached to ubiquitin by E3 ligase and resulting ubiquitinprotein conjugates are then proteolyzed by 26S proteasome. Unfortunately, E3 ligase for p35 has not been identified yet. Nevertheless, we examined the ubiquitination of p35 in MPP ϩtreated neurons. In this experiment, neurons were treated with MPP ϩ in the presence of MG132 to protect p35 from proteasomal degradation. p35 was immunoprecipitated with anti-p35 antibody, and ubiquitination was detected with anti-polyubiquitinated conjugates antibody (FK2). Ubiquitination of p35 was clearly detected in neurons treated with MG132, but ubiquitinated p35 was significantly reduced in MPP ϩ -treated neurons (Fig. 7A). Because this was an unexpected result, we then tested several aspects that might be involved in the ubiquitin-dependent degradation of p35. The Cdk5 inhibitor roscovitine inhibits the glutamate-induced, proteasome-dependent degradation of p35 by inhibiting the phosphorylation of p35 by Cdk5 (35). However, p35 degradation by proteasomes was not affected by roscovitine in MPP ϩ -treated neurons (Fig. 7B). We checked  whether the phosphorylation of p35 would occur after MPP ϩ treatment by examining its phosphorylation-dependent electrophoretic mobility shift on Tris/Tricine PAGE (34). MPP ϩ treatment in the presence of MG132 did not induce any mobility shift of p35 (Fig. 7D). We also determined the phosphorylation status of p35 at Ser-8, a Cdk5 phosphorylation site. MPP ϩ treatment did not increase the phosphorylation of p35 at Ser-8 (Fig. 7C). These results suggest that the MPP ϩ -dependent degradation of p35 is ubiquitination-independent and is different from the glutamate-induced degradation of p35.
We then checked the proteasome activity in MPP ϩ -treated neurons. Proteasome activity was measured using Suc-LLVY-MCA as the substrate. Proteasome activity increased by ϳ20% over the control levels in neurons treated with MPP ϩ for 1 or 3 h (Fig. 7E). To determine whether 26S or 20S proteasome activity is increased or whether the ratio of their activity is affected by MPP ϩ treatment, they were separated by glycerol gradient centrifugation. Two peaks of proteasome activity were confirmed to be 20S at fraction 17 and 26S at fraction 13 by their increased and decreased activity, respectively, in the presence of SDS (Fig. 7F, lower  panel). The 20S proteasome activity, which is responsible for ubiquitin-independent degradation, was increased slightly (ϳ125%) in MPP ϩ -treated neurons, whereas the 26S proteasome activity was not (Fig. 7F, upper panel). To examine whether this up-regulated proteasome activity by MPP ϩ treatment was due to increased expression of proteasome components, we detected Rpt6 as a subunit of the 19 S regulator and the ␣and ␤-subunits of the 20 S proteasome by Western blotting. The up-regulated proteasome activity was not due to the increased expression of proteasome (Fig. 7G). These results suggest that MPP ϩ treatment upregulates the proteasome activity in neurons, preferentially increasing the 20S proteasome activity, and resulting in the ubiquitin-independent degradation of p35.
Cdk5 Acts as a Survival Factor in MPP ϩ -induced Neuronal Cell Death-We investigated the physiological consequence of MPP ϩ -induced p35 degradation because the degradation induced by increased proteasome activity may not be specific to p35. To examine the impact of the down-regulation of Cdk5 activity on neuronal cell death, we initially overexpressed p35 using the adenovirus expression system in neurons and then treated them with MPP ϩ . To exclude the possible involvement of Cdk5-p25 in neuronal cell death, we limited the MPP ϩ treatment to only 3 h. In this experimental paradigm, the expression of p35 was kept at high levels for 3 to 21 h (Fig. 8A), and the generation of p25 was suppressed (Fig. 8A). The infection procedure itself did not induce the cell death of the neurons (data not shown). The numbers of neurons undergoing cell death were counted after their nuclei had stained with DAPI. Neurons overexpressing p35 showed a reduced propor- tion of dead neurons, 30% compared with 50% in the control virus-infected neurons (Fig. 8B).
We next examined the effect of down-regulation of p35 on neuronal cell death using lentiviral shRNA p35 knockdown vector. The expression of p35 was significantly reduced in neurons infected with a knockdown vector compared with those infected with control luciferase knockdown vector (Fig. 8C). The knockdown of p35 increased dead neurons; the ratio was 35% compared with 20% of control vector-infected neurons (Fig. 8D, left three bars). MPP ϩ treatment further increased the cell death of neurons infected with p35 shRNA vector (Fig. 8D,  right three bars). Down-regulation of Cdk5 with Cdk5 shRNA vector also induced neuronal cell death (data not shown). These results indicate strongly that Cdk5 activity is essential for neuronal survival and that it protects neurons from cell death induced by MPP ϩ toxicity.

DISCUSSION
Here, we studied the effects of MPP ϩ on the metabolism of p35, which is a major determinant of Cdk5 activity, using mouse brain cortical neurons. High concentrations of MPP ϩ (500 -1000 M) induced the cleavage of p35 to p25 when neuronal cell death occurred. Conversely, MPP ϩ at concentrations even lower than 100 M induced the proteasomal degradation of p35 before the neurons underwent cell death. We found that this brief treatment with MPP ϩ caused the impairment of mitochondrial protein import, resulting in increased proteasomal activity, which led to the degradation of p35 and then the inac-tivation of Cdk5 activity. The short term MPP ϩ treatment was also sufficient to commit neuronal cell death. The expression of p35 suppressed the cell death progression of MPP ϩ -treated neurons. In contrast, knockdown of p35 increased the cell death. These results indicate that Cdk5-p35 functions as a survival factor during MPP ϩ -induced neuronal commitment to cell death.
Two different types of p35 proteolysis were observed in primary neurons treated with MPP ϩ in both a time-and concentration-dependent manner. The cleavage of p35 to p25 was shown to result in the higher Cdk5 activity accompanied by its mislocalization, leading to cell death. This type of MPP ϩ action has already been reported in cultured neurons and MPTPtreated mice as a model of PD (24,26). Cdk5-p25 thus generated translocates into the nucleus to inactivate the survival factor MEF2 by phosphorylation. However, the degradation of p35 induced by a brief MPP ϩ treatment is attributable to a distinct effect of MPP ϩ on p35 metabolism. In this case, p35 was rapidly degraded without the generation of p25 by the proteasome, and Cdk5 activity was down-regulated as a result of this p35 degradation. This down-regulation might have been missed in previous studies because the major focus was on the generation of p25 and the consequent cell death.
A number of studies have measured the proteasomal activity in MPP ϩ -treated neurons, but their results are controversial, with some reports describing inactivation or no effect and others activation. MPP ϩ itself does not affect the proteolytic activity of isolated 20S proteasome (39). Therefore, MPP ϩ should affect, if any, proteasome activity indirectly through intracellular environmental changes or regulatory factors, most likely mitochondrial dysfunction as discussed below. The down-regulation of proteasome activity is detected mainly after longer MPP ϩ treatments, e.g. for ϳ24 h both in cultured cells (40) and mouse brain (41). In these cases, long term treatment with MPP ϩ disrupts mitochondrial complex I activity, resulting in the depletion of ATP. Because the proteasome requires ATP for the degradation of its target proteins, the depletion of ATP may lead to the down-regulation of proteasome activity. This impaired ubiquitin-proteasome pathway is thought to be relevant to the neuronal degeneration of PD. It has been reported that MPP ϩ does not alter the proteasome activity in TSM-1 and SH-SY5Y neuronal cell lines (39). In this report, proteasome activity was examined by the degradation of ␣-synuclein 8 h after MPP ϩ or 6-hydroxydopamine (6-OHDA) treatment. MPP ϩ did not affect the proteasomal degradation of ␣synuclein, whereas 6-OHDA inactivated 20S proteasome both in vitro and in vivo through oxidative stress. In contrast, several studies have reported increased proteasome activity as observed here. MPP ϩ induces the rapid proteasomal degradation of short-lived proteins such as Ret protein in PC12 cells (42) and cyclin D1 in MG63 osteosarcoma cells and HeLa cells (43). Proteasome activity, when measured with a peptide substrates, is elevated by exposure to 10 or 100 M MPP ϩ for 2 h (44). Taking these results together, the above discrepancies could be derived from differences in experimental conditions such as treatment time, types of cells or proteins analyzed. Although systematic studies will be required, MPP ϩ may affect proteasome activity biphasically, with activation followed by inactivation. MPP ϩ apparently had no effect after 8 h of treatment (39) when activation turned to inactivation.
Because rotenone, a more specific complex I inhibitor, showed an effect similar to MPP ϩ on the degradation of p35, we assumed that mitochondrial dysfunction occurs upstream of p35 degradation. Complex I inhibition would disrupt the mitochondrial membrane potential, which is required for mitochondrial protein import (45). In fact, it has been shown that MPP ϩ reduces the mitochondrial membrane potential, which ultimately causes cell death (46,47). Although we could not demonstrate this conclusively, several indirect lines of evidence suggest that the mitochondrial import system is an intermediate between complex I inhibition and proteasome activation. MPP ϩ inhibited the import of GFP-mito and Su9-EGFP into the mitochondria. Mitochondrial proteins would similarly fail to be imported into the mitochondria. If mitochondrial proteins are not imported into the mitochondria properly, they are degraded rapidly by proteasomes (48). For example, preornithine transcarbamylase is degraded by proteasomes in the cytosol when its import is inhibited by oxidative stress (38). A slight increase in mitochondrial proteins in the cytoplasm may enhance the degradation activity of the proteasome. This may be a kind of quality control system imposed by mitochondrial dysfunction. Although p35 is not a mitochondrial protein, the increased proteasomal activity may also target short-lived proteins such as p35.
The degradation of p35 in cortical neurons is stimulated by glutamate (35). However, the MPP ϩ -induced degradation of p35 appears to be distinct from that induced by glutamate. Whereas the glutamatedependent degradation of p35 is transient and physiological, its MPP ϩ -induced degradation appears long lasting and pathological, as demonstrated by the fact that p35 did not recover even after the  [15][16][17][18][19][20] and 26S peak (fractions 10 -15). The total activity of the 20S proteasome was increased to 127% in MPP ϩ -treated neurons, whereas that of the 26S proteasome was not changed. Typical data from one of four independent experiments with similar results are shown (n ϭ 4). G, effect of MPP ϩ treatment on the expression of proteasome subunits. Neurons were treated with 100 M MPP ϩ for 3 h, and the expression of proteasome were detected by Western blotting with anti-Rpt6 antibody for the 19S regulator, ATPase subunit Rpt6, anti-20S proteasome ␣-subunit antibody MCP231, and anti-20S proteasome ␤-subunit antibody MCP421. Actin was used as the loading control.
MPP ϩ was washed out. Whereas the autophosphorylation of p35 by Cdk5 is required for its glutamate-induced degradation, this was not observed for its MPP ϩ -induced degradation. Ubiquitination of p35 was also decreased in MPP ϩ -treated neurons. Some short-lived proteins are degraded by the ubiquitin-independent proteasome pathway (49). It is well known that ornithine decarboxylase is degraded independently of ubiquitin by the 20S (50) and 26S (51) proteasomes. p53 and cyclin D1 are also degraded by the ubiquitin-independent proteasome pathway, in addition to the ubiquitin-dependent one (52,53). Ubiquitin-independent proteasomal degradation has been demonstrated for several mitochondrial proteins. For example, steroidogenic acute regulatory protein (StAR) is proteolysed independently of ubiquitin in the cytoplasm by proteasomes (48). Preornithine transcarbamylase is degraded in the cytoplasm by proteasomes when mitochondrial import is inhibited by Paraquat (38), but its ubiquitination was not detected despite extensive experimental efforts. These results implicate MPP ϩ as an inducer of the ubiquitin-independent degradation of p35 downstream of mitochondrial import impairment.
The brief treatment with MPP ϩ down-regulated Cdk5 activity by degrading p35, leading to the commitment to cell death. We found that the neuronal cell death was counteracted by the overexpression of p35. These results suggested the role of Cdk5-p35 as a survival factor. This was supported by the experiment showing that depletion of p35 in neurons with shRNA increased cell death, consistent with the results that knockdown of Cdk5 induces neuronal cell death (31). Treatment with MPP ϩ may trigger cell death commitment by down-regulating Cdk5 activity through the degradation of p35. Several recent reports show a role for Cdk5-p35 as a survival factor in neurons. For example, Cdk5 activates the phosphatidylinositol 3/Akt pathway by phosphorylating ErB2 and ErB3 to down-regulate proapoptotic signals (29). Cdk5 also stimulates ERK to up-regulate the anti-apoptotic factor Bcl-2 (30) or to directly phosphorylate Bcl-2 at Ser-70 to maintain neuronal survival activity. Restored Cdk5 expression in Cdk5 null mice rescues their phenotype and perinatal mortality (54). A question for future research is how Cdk5 activity counters MPP ϩ -mediated cell death signals.
MPP ϩ is a drug that is used frequently to induce the cell death of dopaminergic neurons in models of PD. Previous studies have argued that MPP ϩ causes neuronal cell death through several different pathways, including the generation of reactive oxygen species, the secretion of cytotoxic glutamate, and mitochondrial dysfunction (6). It is important to understand which pathway is critical to the induction of cell death. Our results raise another possibility, that the down-regulation of Cdk5 activity triggers the commitment to cell death much earlier than previously thought. The ectopic overactivation of proteasome systems followed by Cdk5 down-regulation may also be involved in the commitment to cell death in neurodegenerative disorders such as PD.