Parkinson’s disease (PD) is a neurodegenerative disorder characterized by loss of dopaminergic neurons. The I93M mutation in ubiquitin C-terminal hydrolase L1 (UCH-L1) is associated with familial PD, and we have previously shown that the I93M UCH-L1-transgenic mice exhibit dopaminergic cell loss. Over 90% of neurodegenerative diseases, including PD, occur sporadically. However, the molecular mechanisms underlying sporadic PD as well as PD associated with I93M UCH-L1 are largely unknown. UCH-L1 is abundant (1–5% of total soluble protein) in the brain and is a major target of oxidative/carbonyl damage associated with sporadic PD. As well, abnormal microtubule dynamics and tubulin polymerization are associated with several neurodegenerative diseases including frontotemporal dementia and parkinsonism linked to chromosome 17. Here we show that familial PD-associated mutant UCH-L1 and carbonyl-modified UCH-L1 display shared aberrant properties: compared with wild-type UCH-L1, they exhibit increased insolubility and elevated interactions with multiple proteins, which are characteristics of several neurodegenerative diseases-linked mutants. Circular dichroism analyses suggest similar structural changes in both UCH-L1 variants. We further report that one of the proteins interacting with UCH-L1 is tubulin, and that aberrant interaction of mutant or carbonyl-modified UCH-L1 with tubulin modulates tubulin polymerization. These findings may underlie the toxic gain of function by mutant UCH-L1 in familial PD. Our results also suggest that the carbonyl modification of UCH-L1 and subsequent abnormal interactions of carbonyl-modified UCH-L1 with multiple proteins, including tubulin, constitute one of the causes of sporadic PD.
Parkinson’s disease (PD) is the most common neurodegenerative movement disorder and is characterized by progressive cell loss confined mostly to dopaminergic neurons in the substantia nigra pars compacta. The I93M mutation in ubiquitin C-terminal hydrolase L1 (UCH-L1) was reported in a German family with dominantly inherited PD (1). To assess the correlation of the I93M mutation and pathogenesis of PD, we have previously generated UCH-L1I93M-transgenic mice. These mice exhibited progressive dopaminergic cell loss in the substantia nigra (2), suggesting that the I93M mutation in UCH-L1 is a causative mutation for PD. The S18Y polymorphism in UCH-L1 has been reported to be associated with decreased risk of PD (3). However, it has also been reported that S18Y is not associated with risk of PD (4).
UCH-L1 is abundant (1–5% of total soluble protein) in the brain (5) and is thought to hydrolyse polymeric ubiquitin and ubiquitin conjugates to monoubiquitin (6). UCH-L1 has also been reported to act as a ubiquitin ligase in vitro (7). In addition to these enzymatic activities, we have found that UCH-L1 binds to and stabilizes monoubiquitin in neurons (8). Our previous studies using circular dichroism (CD) and small-angle neutron scattering strongly suggested that the I93M mutation in UCH-L1 alters the conformation of UCH-L1 (9,10). We have previously shown that mice deficient in UCH-L1 do not exhibit obvious dopaminergic cell loss, in contrast to UCH-L1I93M-transgenic mice (2,8,11), suggesting that a loss or decrease in the level of UCH-L1 is not the main cause of PD, and that UCH-L1I93M-associated PD is caused by an acquired toxicity. Thus, although the hydrolase activity of UCH-L1I93M is decreased (1,9), this decreased activity may not be a major cause of PD.
Increased oxidative stress is associated with neurodegenerative diseases (12,13). In sporadic PD brains, UCH-L1 is a major target of carbonyl formation (12), which is the most widely used marker for oxidative damage to proteins. UCH-L1 has also been identified as a component of several inclusion bodies characteristic of neurodegenerative diseases, including Lewy bodies (14). These findings suggest that UCH-L1 and its modification by carbonyl formation are involved in the cause of sporadic PD. Despite the fact that the majority of PD cases occur sporadically, the molecular mechanisms underlying the causes of sporadic PD, as well as UCH-L1I93M-associated PD, are largely unknown. Moreover, the biochemical properties of UCH-L1I93M and carbonyl-modified UCH-L1 in mammalian cells, such as their protein interactions or detergent insolubility (i.e. the amount of a protein in the insoluble fraction), are poorly understood.
In this study, we analyzed the molecular properties of carbonyl-modified UCH-L1 and UCH-L1I93M and elucidated novel properties of UCH-L1 variants, including protein interactions. We show that carbonyl-modified UCH-L1 and UCH-L1I93M share common properties. Our findings provide novel insights into understanding the mechanisms underlying the toxic gain of function by mutant UCH-L1 and suggest that oxidative stress and subsequent protein interactions of carbonyl-modified UCH-L1 constitute one of the causes of sporadic PD. We also discuss the possible involvement of oxidative modifications of UCH-L1 in other neurodegenerative diseases.
Disease-associated mutants including UCH-L1I93M display aberrant insolubility
Aberrantly increased insolubility compared with wild-type protein is a common biochemical feature of several mutant proteins associated with neurodegenerative diseases: for example, mutant α-synuclein associated with familial PD (15), mutant SOD1 associated with familial amyotrophic lateral sclerosis (ALS) (16,17) and mutant tau associated with frontotemporal dementia and parkinsonism linked to chromosome 17 (18). Although we have previously shown that the insolubility of UCH-L1 in the UCH-L1I93M-transgenic mouse brain is increased compared with that in wild-type mouse (2), the insolubility of UCH-L1I93M itself has been unclear. We observed that pathogenic α-synuclein and SOD1 mutant proteins exhibit increased detergent insolubility in mammalian cells compared with wild-type proteins (Fig. 1A and B). The insolubility of UCH-L1I93M was examined under the same experimental conditions, in which the causative mutants are distinguishable from wild-type proteins. We found that, in dopaminergic SH-SY5Y cells, the protein level of UCH-L1I93M in the insoluble fraction was markedly higher than the levels of UCH-L1WT, UCH-L1S18Y, UCH-L1D30K, which lacks hydrolase activity and binding affinity for ubiquitin (8), and UCH-L1C90S, which lacks hydrolase activity but maintains binding affinity for ubiquitin (8) (Fig. 1C). There was no notable difference among the soluble protein levels (Fig. 1C). The formation of high molecular weight aggregates, which is also a common feature of several mutants, was observed almost exclusively in the insoluble fraction with UCH-L1I93M (Fig. 1C), consistent with the report that UCH-L1I93M produced more aggresomes than UCH-L1WT (19). Increased insolubility of UCH-L1I93M and UCH-L1S18Y/I93M and an increase in the amounts of aggregates specific for these proteins were observed in COS-7 cells (Fig. 1D; Supplementary Material, Fig. S1A), which express very low levels of endogenous UCH-L1. These results demonstrate that UCH-L1I93M shares common features with several mutant proteins linked to neurodegenerative diseases, thus, further supporting the idea that the I93M mutation in UCH-L1 is a causative mutation for PD. Our results also suggest that the insolubility of UCH-L1 is independent of monoubiquitin-binding.
UCH-L1I93M abnormally interacts with multiple proteins
Although increased insolubility is a common characteristic of several mutant proteins associated with neurodegenerative diseases, and this may play a role in the neurotoxicity of the mutant proteins, accumulating evidence suggests that a soluble mutant is the main cause of neurodegeneration (20,21). Studies of dominantly inherited neurodegenerative disease-linked mutants strongly suggest that abnormal physical interactions of the mutant proteins with other proteins constitute a cause of disease (22–26). Hence, we next examined the effect of the I93M mutation on the protein interactions of soluble UCH-L1 using a co-immunoprecipitation (coIP) assay. Silver staining of immunoprecipitant revealed that UCH-L1WT interacts with multiple proteins over 30 kDa (Fig. 1E). We found that the amount of each protein interacting with UCH-L1I93M is mostly higher than the amount interacting with UCH-L1WT or other UCH-L1 variants (Fig. 1F; Supplementary Material, Fig. S1B). Monoubiquitin binding of UCH-L1I93M was decreased compared with that of UCH-L1WT (Fig. 1G), consistent with the decreased hydrolase activity of UCH-L1I93M (1,9). However, the cellular monoubiquitin level in cells expressing UCH-L1I93M was not changed compared with that in cells expressing UCH-L1WT (Fig. 1G). Since UCH-L1I93M-associated PD is presumably caused by an acquired toxicity, the toxic function of UCH-L1I93M may not be mainly mediated by a decreased interaction with monoubiquitin, but rather by aberrantly elevated interactions with multiple other proteins.
Carbonyl-modified UCH-L1 exhibits aberrant properties common to UCH-L1I93M
In the brains of sporadic PD patients, UCH-L1 is a major target of carbonyl formation (12). Carbonyl groups can be introduced into proteins in vivo mainly by reactions with 2-alkenals, 4-hydroxy-2-alkenals (HAE) or ketoaldehydes, which are endogenous aldehydic products formed by lipid peroxidation or glycooxidation (27,28). Protein carbonyls can also be produced by metal-catalyzed reactions with H2O2in vitro (28,29). To analyze the biochemical properties of carbonyl-modified UCH-L1, we used several carbonyl compounds or H2O2 to modify UCH-L1. We have previously reported that UCH-L1 is modified by 4-hydroxy-2-nonenal (HNE) in vitro (9). In COS-7 cells transfected with UCH-L1WT, UCH-L1 was modified by physiological concentrations of HNE (10–100 µm) (9) or 4-hydroxy-2-hexenal (HHE) in a dose-dependent manner (Fig. 2A and B; Supplementary Material, Fig. S1C). Carbonyl modification of UCH-L1 was also detected when cells were treated with 100 µm 2-propenal (Fig. 2A), but not with 100 or 500 µm methylglyoxal, 100 or 500 µm malondialdehyde, both of which are ketoaldehydes, or 0.1 or 1 mm H2O2 (data not shown). Thus, carbonyl-modified UCH-L1 can be produced by reactions with HAE or 2-alkenals in mammalian cells.
Interestingly, carbonyl-modified UCH-L1 and UCH-L1I93M exhibit common biochemical properties: ubiquitin binding of HNE-modified UCH-L1 was decreased (Fig. 2B), and both the insolubility of HNE-modified UCH-L1 and the interactions of HNE-modified UCH-L1 with proteins over 30 kDa were increased, compared with those of UCH-L1WT (Fig. 2C–E). HHE and 2-propenal had similar effects to HNE (Fig. 2F–I). Treatment of cells with 100 µm H2O2, methylglyoxal or malondialdehyde had no effect on the insolubility of UCH-L1 or the interactions of UCH-L1 with other proteins (data not shown). Consistent with the report that UCH-L1 is a major target of carbonyl formation in the brains of sporadic PD patients (12), UCH-L1 is a major target of carbonyl modification in cells treated with HNE (Fig. 3A). We used the EF1 promoter to yield abundant expression of UCH-L1 in this experiment, since the amount of UCH-L1 is 1–5% of soluble protein in the brain (5). These results suggest that the carbonyl-modified UCH-L1 in sporadic PD brains functions as a causative factor for disease in a similar manner to UCH-L1I93M.
Cys-90 and Cys-152 of UCH-L1 are targets for HAE modification
The appearance of HNE-modified proteins in nigral neurons has been shown to be associated with sporadic PD (30,31). Therefore, we next determined the HNE-modified amino acid residues of UCH-L1 that regulate its insolubility and protein interactions. HNE can form covalent cross-links with cysteine, lysine and histidine residues in proteins (28). To test the specificity of HNE modification in mammalian cells, we used cells transfected with α-synuclein, which contains no cysteine residues. HNE modification of α-synuclein was not detected when cells were treated with 100 µm HNE (Fig. 3B). These results suggest that among the amino acid residues of UCH-L1, cysteine residues are the primary target for HAE. We speculated that Cys-90 is accessible to HAE, since it is accessible to ubiquitin. Using the three-dimensional structure of human UCH-L1 (32), we observed that not only Cys-90 but also Cys-132 and Cys-152 are located on the surface of the protein (Fig. 3C and D). Thus, we tested the insolubility and protein interactions using C90S, C132S and C152S UCH-L1 mutant proteins. We also used C220S UCH-L1 as a control. We found that the C152S mutant bound to monoubiquitin in both HNE-treated cells and untreated cells (Fig. 4A). UCH-L1C90S did not exhibit notably increased insolubility upon HNE-treatment compared with UCH-L1WT (1.3-fold increase in UCH-L1C90S, 2.5-fold increase in UCH-L1WT) (Fig. 4B). The amount of proteins over 30 kDa interacting with UCH-L1C90S was markedly lower than that interacting with UCH-L1WT when cells were treated with HNE (Fig. 4C). Similar results were obtained when cells were treated with HHE (Fig. 4E and F; Supplementary Material, Fig. S1D). Mutations at Cys-132 and Cys-220 had no effect on protein insolubility or interactions (Fig. 4A–C). Consistent with these results, HNE modification of C90S and C152S mutants was decreased compared with that of UCH-L1WT when cells were treated with HNE (∼40 and 60% decrease, respectively) (Fig. 4D). These results indicate that HAE modification of UCH-L1 at Cys-90 increases the insolubility and interactions of UCH-L1, and modification of Cys-152 reduces monoubiquitin binding. The level of HNE modification of UCH-L1I93M upon HNE-treatment was markedly lower than that of UCH-L1WT (Fig. 4G). Since the location of Cys-90 is close to Ile-93 (Supplementary Material, Fig. S2), it is possible that the I93M mutation and HAE modification at Cys-90 cause similar structural changes in UCH-L1.
HNE modification causes structural changes in UCH-L1
To address the structural changes in carbonyl-modified UCH-L1, we used CD spectroscopy to estimate the secondary structure. We have previously shown that, compared with UCH-L1WT, the I93M mutant displays lower ellipticity around 195 nm, suggesting a decreased α-helix content, and an increase in the content of β-sheet (9,10). Relative to wild-type protein, HNE-modified UCH-L1 also displayed a lower peak around 190–195 nm (Fig. 3E and F). The relative proportions of α-helix, β-sheet and other secondary structural features in these proteins were estimated from mean residue ellipticity data. HNE-modified UCH-L1 also exhibited decreased α-helix content, and an increase in the content of β-sheet compared with UCH-L1WT (42.9% α-helix, 20.9% β-sheet, 20.6% β-turn and 15.7% random for UCH-L1WT, and 34.0% α-helix, 27.3% β-sheet, 22.3% β-turn and 16.4% random for HNE-modified UCH-L1). These results suggest that UCH-L1I93M and carbonyl-modified UCH-L1 adopt a similar aberrant structure.
The ALS-linked mutation in SOD1 increases its hydrophobicity, which may promote aberrant interactions of SOD1 with other cellular constituents (33). However, the interactions of UCH-L1I93M or HNE-modified UCH-L1 with hydrophobic beads were not altered relative to those of UCH-L1WT (data not shown), indicating that the I93M mutation and HNE modification of UCH-L1 do not increase its hydrophobicity. Considering the fact that unnatural β-sheet proteins readily become insoluble or form further β-hydrogen-bonding with other β-strands they encounter (34), our results suggest that the increased insolubility and protein interactions of abnormal UCH-L1 are due to the increased β-sheet content of UCH-L1.
UCH-L1 physically interacts with tubulin
To understand the molecular mechanism underlying toxic gain of function by UCH-L1, we attempted to identify UCH-L1I93M-interacting proteins by coIP assay and subsequent LC-MS/MS analysis (Fig. 5A). A database search of the peptide sequences obtained identified α-tubulin as a UCH-L1I93M-interacting protein (Supplementary Material, Table S1). The interaction between UCH-L1 and endogenous α-tubulin was confirmed with transiently expressed UCH-L1 (Fig. 5B and C). The interaction of UCH-L1I93M with α-tubulin was increased compared with that of UCH-L1WT (Fig. 5B). We detected the interaction of endogenous α-tubulin with endogenous UCH-L1 using Neuro2a cells (Fig. 5D). Tubulin is composed of a heterodimer of α- and β-tubulin, and we confirmed, using native-PAGE, that tubulin exists as a heterodimer in cell lysates in coIP experimental conditions (data not shown), indicating that UCH-L1 interacts with tubulin. Indeed, β-tubulin was also precipitated with UCH-L1 (Supplementary Material, Fig. S3). In contrast to tubulin, interaction of β-actin with UCH-L1 was not detected (Fig. 5C). To test whether UCH-L1 directly interacts with tubulin, we performed pull-down assay using recombinant UCH-L1 and purified tubulin. Direct interaction of UCH-L1 with tubulin was observed (Fig. 5E).
Since the interactions between UCH-L1 and proteins over 30 kDa are increased by carbonyl modification or I93M mutation of UCH-L1, we tested the effects of HAE on the interaction of UCH-L1 with tubulin. We found that HAE modification of UCH-L1 promotes interactions between UCH-L1 and tubulin (Fig. 5F, G and I). In addition, a coIP assay using C90S, C132S and C152S UCH-L1 mutants showed less binding of UCH-L1C90S to tubulin than UCH-L1WT did, when cells were treated with HNE or HHE (Fig. 5G–I), indicating that the increased interaction of UCH-L1 with tubulin is caused by the HAE modification of Cys-90 of UCH-L1. These results are consistent with the results showing that the HAE modification of Cys-90 of UCH-L1 promotes the interaction of UCH-L1 with multiple proteins. The I93M mutation and HNE modification of UCH-L1 also promote direct interactions between UCH-L1 and tubulin (data not shown). Thus, UCH-L1I93M and HNE-UCH-L1 also exhibit common biochemical properties with respect to the interactions with tubulin.
Both UCH-L1I93M and carbonyl-modified UCH-L1 aberrantly promote tubulin polymerization
Microtubules are dynamic polymers composed of tubulin that continuously grow and shorten through tubulin addition and loss at the microtubule ends. Microtubule-stabilizing agents such as paclitaxel, which promote tubulin polymerization and suppress microtubule dynamics, are effective chemotherapeutic agents for the treatment of many cancers. However, neuropathy is a major adverse effect of microtubule-stabilizing agents-based chemotherapy (35). Paclitaxel induces apoptosis in cortical neurons by a mechanism independent of its cell cycle effects, because postnatal cortical neurons are postmitotic (36). These findings indicate that tubulin polymerization must be tightly regulated for neurons to function and remain viable. Furthermore, abnormal microtubule dynamics and tubulin polymerization are associated with several neurodegenerative diseases including frontotemporal dementia and parkinsonism linked to chromosome 17 (37,38). Therefore, we examined the effects of UCH-L1WT, UCH-L1I93M and HNE-UCH-L1 on tubulin polymerization using an in vitro assay. Interestingly, both UCH-L1I93M and HNE-UCH-L1 promote tubulin polymerization, although UCH-L1WT had almost no effect on it (Fig. 6A and B). Promotion of tubulin polymerization may result in a stabilization of microtubules because of the dynamic instability of microtubules. To test whether abnormal UCH-L1 also promotes tubulin polymerization in mammalian cells, we analyzed the amounts of soluble, polymeric and total tubulin in cells expressing UCH-L1I93M. Although transient expression of UCH-L1I93M had no effect on the amount of total tubulin (Fig. 5B), cells stably expressing UCH-L1I93M contained increased amount of total tubulin compared with control cells or cells expressing other UCH-L1 variants (Fig. 6C). Consistent with the in vitro polymerization assay, the amount of polymeric tubulin was increased in cells expressing UCH-L1I93M, whereas the amount of soluble tubulin was not (∼1.4 and 1.0-fold increase, respectively, compared with the amount of tubulin in cells expressing UCH-L1WT) (Fig. 6D). The amount of β-actin was not affected by the expression of UCH-L1 variants (Fig. 6C and D), also consistent with the results showing that UCH-L1 does not interact with β-actin. We did not detect specific interaction of UCH-L1 with polymerized tubulin (Fig. 6E), indicating that UCH-L1 may not interact with microtubules, although the possibility is not excluded that they can interact under certain conditions or at a limited number of sites such as the microtubule ends.
Since D30K and C90S mutations had no effect on the interaction of UCH-L1 and tubulin (Fig. 5B), we speculated that the tubulin-binding region of UCH-L1 is different from ubiquitin-binding region. To elucidate the amino acid residues of UCH-L1 involved in the interaction with tubulin and to show that modulation of tubulin polymerization is caused by the increased interaction of UCH-L1 with tubulin, we made a series of alanine substitutions of basic and acidic residues located on the surface of UCH-L1 and performed coIP assays using these mutants (Fig. 7A; Supplementary Material, Fig. S3). The R63A and H185A mutants displayed increased interactions with tubulin (Fig. 7A), indicating that Arg-63 and His-185, which are distinct from the ubiquitin-binding region (Fig. 7B), are involved in this interaction. The increased interactions of R63A and H185A UCH-L1 with tubulin may be caused by altered ionic interactions. In contrast to the I93M mutant or HNE-UCH-L1, the R63A mutant caused a decrease in tubulin polymerization (Fig. 7C). Although UCH-L1R63A has opposite effects to the I93M mutant or HNE-UCH-L1, it also modulated tubulin polymerization. Thus, modulation of tubulin polymerization by UCH-L1 variants is caused by the abnormally increased interaction of UCH-L1 with tubulin.
From our results, we hypothesized that UCH-L1I93M-associated neurodegeneration or PD is at least partly mediated by aberrant tubulin polymerization. Therefore, we tested the effects of UCH-L1I93M and paclitaxel on neuronal cell death using differentiated Neuro2a cells, which have been used to assess the toxicity of mutant proteins linked to neurodegenerative diseases (17,39,40). We confirmed that paclitaxel does not interfere with the interaction between UCH-L1 and tubulin (data not shown). Treatment of cells with 5 µm paclitaxel slightly but significantly elevated cell death in cells expressing UCH-L1I93M, but had no effect in cells expressing UCH-L1WT (Fig. 6F). This indicated that the toxicity of UCH-L1I93M may be at least in part mediated by aberrant microtubule dynamics or tubulin polymerization. Given that tightly regulated tubulin polymerization is necessary for neurons to be viable, our findings strongly suggest that aberrant tubulin polymerization caused by UCH-L1I93M partly underlies the toxic gain of function of mutant UCH-L1, and that carbonyl-modified UCH-L1 also functions as a toxic protein in neurons. We propose that interactions of mutant or carbonyl-modified UCH-L1 with other proteins, including tubulin, constitute one of the causes of not only familial PD, but also sporadic PD (Fig. 7D).
Our previous study using CD suggests that the I93M mutation increases the β-sheet content, but reduces the α-helix content of UCH-L1 (9). We have also shown, using small-angle neutron scattering, that UCH-L1WT has an ellipsoidal shape, whereas UCH-L1I93M has a more globular shape in an aqueous solution (10). However, the biochemical and molecular properties of UCH-L1I93M in mammalian cells, as well as the molecular mechanisms that underlie UCH-L1I93M-associated PD, have not been elucidated. In this study, we have shown that, compared with UCH-L1WT, UCH-L1I93M displays increased insolubility, which is characteristic of several neurodegenerative disease-linked mutants, aberrantly elevated interactions with multiple proteins over 30 kDa and decreased interaction with monoubiquitin (Fig. 1). Taken together, our new and previous findings indicate that the I93M mutation in UCH-L1 alters its conformation, resulting in changes in the biochemical properties of UCH-L1.
Similar to UCH-L1I93M, other dominantly inherited neurodegenerative disease-linked mutants, such as mutant SOD1 and mutant α-synuclein, cause neurodegeneration, presumably via an acquired toxicity. Studies of the mutants strongly suggest that abnormally increased interactions of these mutant proteins with other proteins constitute a cause of disease (22–25). Therefore we screened for UCH-L1-interacting proteins using a coIP assay and subsequent LC-MS/MS analysis. We found that tubulin is a novel UCH-L1-interacting protein, and that the interactions of UCH-L1I93M with these proteins are increased compared with those of UCH-L1WT (Fig. 5B). We have also shown that UCH-L1I93M promotes tubulin polymerization and stabilizes microtubules (Fig. 6B–D). UCH-L1I93M and paclitaxel coordinately induced neuronal cell death (Fig. 6F). Together with the fact that tightly regulated tubulin polymerization is essential for neurons to function and remain viable, and that abnormal microtubule dynamics and tubulin polymerization are associated with several neurodegenerative diseases (37,38), our results strongly suggest that aberrant tubulin polymerization caused by mutant UCH-L1 at least partly constitutes a toxic function of mutant UCH-L1. Other than tubulin, mutant UCH-L1 interacts with multiple proteins (Figs 1F and 5A). These other interactors may also be involved in the mechanism of UCH-L1-mediated neurodegeneration (Fig. 7D). We have identified some of these interactors (T.K. and K.W., unpublished data), and these proteins are currently under investigation.
It is known that the majority of PD cases occur sporadically, and that oxidative/carbonyl stresses are elevated in PD brains (12,13). However, the molecular mechanisms underlying the causes of sporadic PD have remained largely unknown. Choi et al. (12) have shown that UCH-L1 is a major target of carbonyl damage associated with sporadic PD, implying that carbonyl-modified UCH-L1 is involved in the cause of sporadic PD. In the present study, we found that carbonyl-modified UCH-L1 and UCH-L1I93M share molecular and functional properties. Importantly, both UCH-L1s display shared properties in all of the experiments we performed (Supplementary Material, Table S2). These results strongly suggest that carbonyl-modified UCH-L1 is also toxic to neurons and constitutes one of the causes of sporadic PD. Considering that UCH-L1 is abundant in the brain (5), and that UCH-L1 is a major target of carbonyl damage in PD brains (12), it is possible that carbonyl-modified UCH-L1 is the major cause of the disease.
It has been reported that UCH-L1 mRNA is expressed abundantly in dopaminergic neurons in the human brain (41). Abundant expression of UCH-L1 protein in dopaminergic neurons was also observed in mouse brains (Supplementary Material, Fig. S4 and S5). Dopaminergic neurons are particularly exposed to oxidative and carbonyl stresses because dopamine can auto-oxidize into toxic dopamine quinone, superoxide radicals and hydrogen peroxide (42). In addition, it has been reported that oxidative stresses in dopaminergic neurons in sporadic PD brains are higher than the stresses in control brains (30). Thus, in PD, UCH-L1I93M or oxidative/carbonyl-modified UCH-L1 is possibly overproduced in dopaminergic neurons, leading to the selective loss of dopaminergic neurons (Fig. 7D).
Oxidatively modified UCH-L1 has also been found in the brains of both familial and sporadic Alzheimer’s disease (AD) patients (12,43,44). AD is characterized pathologically by deposition of the amyloid β-protein in the form of amyloid plaques in the brain, and the deposition of the amyloid β is thought to be a major cause of both familial and sporadic AD (20). Thus, although it is possible that toxicity of carbonyl-modified UCH-L1 is involved in amyloid β-mediated neurodegeneration in AD, carbonyl-modified UCH-L1 may not be the primary cause of AD. A recent report has shown that brains from patients with sporadic PD and AD contain decreased levels of UCH-L1 (30 and 50% decrease, respectively) (12). Gong et al. (45) showed that the introduction of exogenous UCH-L1 rescued the synaptic and cognitive functions of AD model mice, which exhibit decreased levels of UCH-L1 in their hippocampi. We have also shown that mice deficient in UCH-L1 exhibit memory dysfunction (46). These findings indicate that a reduction in the levels of functional UCH-L1 may contribute to the pathogenesis of AD. Oxidative modification of several proteins, including antioxidant proteins, is found in mice deficient in UCH-L1 (47), suggesting involvement of these proteins in AD. Since diminution of the proteasome activity may lead to neurodegeneration (48), it is also possible that decreased UCH-L1 function leads to dysfunction of the ubiquitin-proteasome system and this dysfunction contributes to neurodegeneration in AD. On the contrary, mice deficient in UCH-L1 do not exhibit obvious dopaminergic cell loss, indicating that a loss or decrease in the level of UCH-L1 is not the main cause of PD. Investigation of the relationship between the specificity of brain areas that is affected by oxidative stress and genetic or environmental factors should generate further insights into the mechanism of oxidative stress in the pathogenesis of sporadic PD and AD.
In conclusion, familial PD-associated UCH-L1I93M and carbonyl-modified UCH-L1, which is associated with sporadic PD, display common aberrant properties. Thus, UCH-L1I93M would be a useful tool for studying the molecular mechanism underlying sporadic PD. We propose that the abnormal interactions of UCH-L1 variants with other proteins including tubulin constitute one of the causes of not only familial PD associated with UCH-L1I93M, but also sporadic PD, and can be therapeutic targets for these diseases and possibly for other neurodegenerative diseases.
MATERIALS AND METHODS
pCI-neo-hUCH-L1 plasmids containing human WT UCH-L1 and UCH-L1 variants with or without FLAG tag were prepared as described previously (49) or generated using a QuikChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA). The expression plasmid pCR3-hα-synuclein containing FLAG-tagged human α-synuclein was kindly donated by Ryosuke Takahashi (Kyoto University, Kyoto, Japan) and Yuzuru Imai (Tohoku University, Miyagi, Japan) (50). The pcDNA3-hSOD1 expression plasmids containing WT, A4V, G85R or G93A mutant SOD1, and pCI-hα-synuclein expression plasmids containing WT, A30P or A53T mutant α-synuclein were prepared as described previously (17). The expression plasmid pEF-hUCH-L1 containing WT UCH-L1 was constructed by ligating the cDNA encoding UCH-L1 into pEF-BOS vector (51). The bacterial expression plasmid pPROTetE-hUCH-L1 containing 6HN-tagged UCH-L1 was prepared as described previously (9). pGEX-hUCH-L1 bacterial expression plasmids containing WT, I93M or R63A UCH-L1 with a GST-tag were constructed by ligating the cDNA encoding each UCH-L1 into pGEX-6P-1 vector (GE Healthcare UK Ltd, Buckinghamshire HP7 9NA, UK).
Cell culture and transfection
Neuro2a, SH-SY5Y, COS-7 and HeLa cells were maintained in Dulbecco’s modified Eagle’s medium (Sigma, St Louis, MO, USA) supplemented with 10% fetal bovine serum (JRH Biosciences, Lenexa, KS, USA). NIH-3T3 cells stably expressing human UCH-L1 with a FLAG-HA double-tag at the N terminus were cultured as described previously (49). Transient transfection of Neuro2a, SH-SY5Y and COS-7 cells with each vector was performed using the FuGENE 6 Transfection Reagent (Roche Diagnostics, Indianapolis, IN, USA), TransFectin Lipid Reagent (Bio-Rad, Hercules, CA, USA) and Lipofectamine Reagent (Invitrogen, Carlsbad, CA, USA), respectively. For the experiments investigating the carbonyl modification of UCH-L1, cells were incubated at 37°C for 90 min with each carbonyl compound or H2O2 in PBS containing 5 mm glucose, 0.3 mm CaCl2 and 0.62 mm MgCl2.
SDS–PAGE was performed under reducing conditions. Immunoblotting was performed according to standard procedures. The preparation of detergent (1% Triton X-100)-soluble and -insoluble fractions was performed as described previously (17). Mouse anti-α-tubulin and anti-β-tubulin antibodies were purchased from Sigma. Rabbit anti-α-tubulin and anti-β-tubulin antibodies were from Cell Signaling (Danvers, MA, USA). Mouse anti-HNE and rabbit anti-HNE antibodies were from Oxis (Portland, OR, USA) and Alpha Diagnostic (San Antonio, TX, USA), respectively. Antibodies against SOD1, UCH-L1 and reduced-HNE were purchased from Stressgen Bioreagents (Victoria, BC, Canada), UltraClone (England, UK) and Calbiochem (Darmstadt, Germany), respectively. Anti-β-actin, ubiquitin and FLAG antibodies were from Sigma. The antibody against α-synuclein was from Chemicon (Temecula, CA, USA). For immunoblotting with anti-reduced HNE antibody, the proteins on a PVDF membrane were reduced with 10 mm NaBH4 in Tris-buffered saline for 30 min at room temperature before being reacted with anti-reduced HNE antibody. Carbonyl modification of proteins was detected using an OxyBlot Protein Oxidation Detection Kit (Chemicon) containing an anti-DNP antibody.
Immunoprecipitation was performed as previously described (52). Cells were harvested by cold immunoprecipitation buffer (15 mm Tris pH 7.5, 120 mm NaCl, 25 mm KCl, 2 mm EGTA, 2 mm EDTA, 0.5% Triton X-100 and protease inhibitors). The lysates were centrifuged at 20 000g for 10 min at 4°C. The supernatant was subjected to immunoprecipitation. Lysates (1 mg protein in immunoprecipitation buffer) were incubated with 5 μg of antibody for 12 h. Twenty microliters of protein G Sepharose (GE Healthcare) was then added, and incubation was continued for 1 h. For the immunoprecipitation of FLAG-tagged proteins, lysates (1–2 mg protein in immunoprecipitation buffer) were incubated with 30 µl anti-FLAG M2 affinity gel (Sigma) for 2 h. After the beads were washed three times with immunoprecipitation buffer, proteins were eluted with SDS sample buffer (10 mm Tris, pH 7.8, 3% SDS, 5% glycerol and 0.02% bromophenol blue). In some experiments, proteins were eluted with SDS sample buffer containing 2% 2-mercaptoethanol. For the immunoprecipitation of endogenous UCH-L1 (Fig. 5D), 100 µg anti-UCH-L1 antibody (53) or 100 µg normal rabbit IgG (Santa Cruz Biotechnology, Santa Cruz, CA, USA) was immobilized to 100 µl of protein G beads using a Seize X Protein G Immnoprecipitation Kit (Pierce, Rockford, IL, USA). Cell lysates (1 mg protein in 50 mm Tris, pH 7.5, 150 mm NaCl, 5 mm EDTA, 0.25% Triton X-100 and protease inhibitors) were incubated with 25 µl of beads for 12 h. Protein G beads without antibody and protein G beads cross-linked with normal rabbit IgG were used as controls.
Mass spectrometry analysis
Protein bands were sliced from the gel and subjected to in-gel trypsin digestion, and LC-MS/MS analysis was performed at APRO Life Science Institute, Inc. (Naruto, Japan) as a custom service.
CD measurements of 0.1 mg/ml (4 µm) of recombinant human UCH-L1 without a tag (Boston Biochem, Cambridge, MA, USA) in 20 mm sodium phosphate buffer (pH 8.0) were performed as described previously (9,10). Since two cysteine residues in UCH-L1, Cys-90 and Cys-152, are major targets of HNE modification (Fig. 4), 4 µm UCH-L1 was reacted with 8 µm HNE. Far UV CD spectra (190–250 nm) were recorded in a 1 mm quartz cuvette on a Jasco J-820 spectropolarimeter (Jasco, Tokyo, Japan) equipped with a temperature controller by scanning at a rate of 50 nm/min at 25°C. For all spectra, 12 scans were averaged. All CD spectra were corrected by background subtraction of the spectrum obtained with buffer alone and smoothed. Spectra were analyzed for the percentage of secondary structural elements by a computer program, based on an algorithm that compares experimental spectra with those of known proteins (54).
Preparation of recombinant proteins
6HN-tagged human UCH-L1 proteins were prepared as described previously (9). For purification of UCH-L1 without a tag, the pGEX UCH-L1 vectors were transformed into Escherichia coli BL21. Production of fusion proteins was induced by the addition of isopropyl-β-d-thiogalactopyranoside to a final concentration of 0.5 mm. After a 4 h induction at 37°C, the cells were harvested and lysed by sonication in PBS containing 1% Triton X-100 and protease inhibitors. Purification of GST-tagged UCH-L1 was performed using glutathione Sepharose 4B (GE Healthcare), and UCH-L1 was released from GST by digestion using PreScission Protease (GE Healthcare). Purified proteins were resolved by SDS–PAGE under reducing conditions and visualized by Coomassie brilliant blue R-250 to confirm purity (Supplementary Material, Fig. S6).
TALON resin (Clontech, Palo Alto, CA, USA) was blocked with 3% BSA for 1 h in order to prevent non-specific binding of tubulin (data not shown) and washed three times with PBS containing 0.05% Triton X-100. Five micrograms of recombinant UCH-L1 with an HN tag and 5 µg of purified tubulin (>99% pure tubulin, Cytoskeleton, Denver, CO, USA) were mixed and incubated for 4 h in PBS containing 0.05% Triton X-100. As a control, vehicle was mixed instead of UCH-L1. Twenty microliters of TALON resin blocked with BSA was then added, and incubation was continued for 1 h. After beads were washed three times with PBS containing 0.05% Triton X-100, proteins were eluted with SDS sample buffer.
Tubulin polymerization assay
An in vitro tubulin polymerization assay was performed using a tubulin polymerization assay kit, OD based, >99% pure tubulin (Cytoskeleton), according to the manufacturer’s protocol. Briefly, recombinant UCH-L1 without a tag and tubulin were mixed to give a final concentration of 0.05 mg/ml UCH-L1 and 3 mg/ml tubulin in tubulin polymerization buffer (80 mm PIPES, pH 6.9, 2 mm MgCl2, 0.5 mm EGTA, 1 mm GTP, 5% glycerol) and subjected to a tubulin polymerization assay. As a control, vehicle was mixed instead of UCH-L1. Since two cysteine residues in UCH-L1 are major targets of HNE modification (Fig. 4), 40 µm UCH-L1 was reacted with 80 µm HNE to prepare the HNE-modified UCH-L1. To analyze the interaction between UCH-L1 and polymerized tubulin, the polymerized tubulin was pelleted by centrifugation after a tubulin polymerization assay. The supernatant (100 µl) was mixed with 50 µl of 3× SDS sample buffer (30 mm Tris, pH 7.8, 9% SDS, 15% glycerol, 0.06% bromophenol blue). The pellet was washed twice with tubulin polymerization buffer and then dissolved in 150 µl of SDS sample buffer.
Preparation of cell extracts containing soluble and polymeric tubulin
Preparation of soluble and polymeric fractions of tubulin was performed as described (55) with slight modification. Briefly, cells were washed very gently with a microtubule stabilizing buffer (0.1 M N-morpholinoethanesulfonic acid, pH 6.75, 1 mm MgSO4, 2 mm EGTA, 0.1 mm EDTA, 4 M glycerol). Soluble proteins were extracted at 37°C for 5 min in microtubule stabilizing buffer containing 0.04% saponin. The remaining cytoskeletal fraction in the culture dish was washed with microtubule stabilizing buffer containing 0.4% saponin and dissolved in SDS sample buffer.
Quantitative assessment of cell death
Neuro2a cells were transfected with plasmids. Four hours after transfection, neuronal cell differentiation was induced by addition of 5 mm dibutyryl cAMP as described in the literature (40), and cells were incubated for 24 h. Cells were then incubated with or without 5 µm paclitaxel for another 24 h. Cell death was assessed by a lactate dehydrogenase release assay, as described previously (17).
For comparison of two groups, the statistical difference was determined by Student’s t-test.
This work was supported by Grants-in-Aid for Scientific Research of Japan Society for the Promotion of Science; Research Grant in Priority Area Research of the Ministry of Education, Culture, Sports, Science and Technology, Japan; Grants-in-Aid for Scientific Research of the Ministry of Health, Labour and Welfare, Japan; Program for Promotion of Fundamental Studies in Health Sciences of the National Institute of Biomedical Innovation (NIBIO), Japan; New Energy and Industrial Technology Development Organization (NEDO), Japan.
We thank Dr Ryosuke Takahashi (Kyoto University) and Dr Yuzuru Imai (Tohoku University) for the gift of pCR3-hα-synuclein plasmid, Dr Yasuyuki Suzuki (National Institute of Neuroscience) for valuable discussion; Naoki Takagaki (National Institute of Neuroscience) for support with English.
Conflict of Interest statement. None declared.