Parkinson's disease (PD) is a common neurodegenerative disease, but its pathogenesis remains elusive. A mutation in ubiquitin C-terminal hydrolase L1 (UCH-L1) is responsible for a form of genetic PD which strongly resembles the idiopathic PD. We previously showed that 1-(3′,4′-dihydroxybenzyl)-1,2,3,4-tetrahydroisoquinoline (3′,4′DHBnTIQ) is an endogenous parkinsonism-inducing dopamine derivative. Here, we investigated the interaction between 3′,4′DHBnTIQ and UCH-L1 and its possible role in the pathogenesis of idiopathic PD. Our results indicate that 3′,4′DHBnTIQ binds to UCH-L1 specifically at Cys152 in vitro. In addition, 3′,4′DHBnTIQ treatment increased the amount of UCH-L1 in the insoluble fraction of SH-SY5Y cells and inhibited its hydrolase activity to 60%, reducing the level of ubiquitin in the soluble fraction of SH-SY5Y cells. Catechol-modified UCH-L1 as well as insoluble UCH-L1 were detected in the midbrain of 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine-treated PD model mice. Structurally as well as functionally altered UCH-L1 have been detected in the brains of patients with idiopathic PD. We suggest that conjugation of UCH-L1 by neurotoxic endogenous compounds such as 3′,4′DHBnTIQ might play a key role in onset and progression of idiopathic PD.
We investigated the interaction between ubiquitin C-terminal hydrolase L1 (UCH-L1) and the brain endogenous parkinsonism inducer 1-(3′,4′-dihydroxybenzyl)-1,2,3,4-tetrahydroisoquinoline (3′,4′DHBnTIQ). Our results indicate that 3′,4′DHBnTIQ binds to UCH-L1 specifically at cysteine 152 and induces its aggregation. 3′,4′DHBnTIQ also inhibits the hydrolase activity of UCH-L1. Catechol-modified as well as insoluble UCH-L1 were detected in the midbrains of MPTP-treated Parkinson's disease (PD) model mice. Conjugation of UCH-L1 by neurotoxic endogenous compounds like 3′,4′DHBnTIQ might play a key role in onset and progression of PD.
matrix-assisted laser desorption ionization time-of-flight mass spectrometry
polyacrylamide gel electrophoresis
m-amino-phenylboronic acid agarose
serine 18 to tyrosine polymorphism
S18Y mutant of UCH-L1
sodium dodecyl sulfate
TBS supplemented with 1% polysorbate 20
ubiquitin C-terminal hydrolase L1
2-(4-iodophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H-tetrazolium monosodium salt
Parkinson's disease (PD) is one of the most common incurable neurodegenerative diseases at this moment, pathologically characterized by the selective degeneration of nigrostriatal dopaminergic neurons and the appearance of cytoplasmic proteinaceous aggregates named Lewy bodies (LB) in the surviving ones. Its main symptoms are resting tremor, rigidity, bradykinesia, and postural instability and they usually appear after the age of 50. Although the concrete pathogenesis is still elusive, the hypothesis that the onset of PD stands in the interaction between genetic and environmental factors is the most plausible. Most cases of PD are sporadic. Only about a tenth of the patients suffer from the genetic form. There have been identified several genes responsible for familial PD. Among them, a point mutation in UCH-L1 gene which leads to an isoleucine to methionine (I93M) substitution in UCH-L1 protein (I93MUCH-L1) was found to cause a late-onset autosomal dominant form of genetic PD which closely resembles the idiopathic form. While most forms of genetic PD have an early or juvenile onset and symptoms vary, the symptoms of I93MUCH-L1-caused PD appear around the age of 50 starting with resting tremor, and as the disease advances, the other PD-specific symptoms show (Leroy et al. 1998). Wada group generated transgenic mice expressing I93MUCH-L1 and found that these mice exhibited a significant decrease of the dopaminergic neurons in substantia nigra and of dopamine content in the striatum. Surviving dopaminergic neurons exhibited reduced levels of ubiquitin and midbrains of I93M transgenic mice contained increased amounts of insoluble UCH-L1 (Setsuie et al. 2007). On the other hand, a polymorphism of serine to tyrosine gene at codon 18 (S18Y) in UCH-L1 (S18YUCH-L1) is associated with a reduced risk of idiopathic PD in Caucasian, German, as well as in Japanese population, underlying the close relation between UCH-L1 and PD (Maraganore et al. 1999; Wintermeyer et al. 2000; Zhang et al. 2000; Satoh and Kuroda 2001).
UCH-L1 is a 223 amino acids protein (25 kDa), mainly expressed in neurons, which constitutes up to 5% of the total soluble protein in the brain, best known for its function as deubiquitinating enzyme (Wilkinson et al. 1989; Liu et al. 2002). Down-regulated as well as oxidatively modified UCH-L1 have been detected in the brains of patients with idiopathic PD (Choi et al. 2004). Moreover, UCH-L1 together with parkin and α-synuclein are the major components of LB (Schlossmacher et al. 2002; Eriksen et al. 2003; Ardley et al. 2004). Also, mutations in the genes encoding these proteins are known to cause different forms of genetic PD. Although LB's relevance to the onset of PD is not yet well understood, LB and the constituting proteins undoubtedly represent an important feature when approaching PD. Thus, elucidating the relationship between UCH-L1 and PD could be a promising strategy to properly understand the onset of this disease.
We have reported several tetrahydroisoquinoline (TIQ) derivatives to be factors, endogenous or exogenous, related to PD (Kohno et al. 1986; Ohta et al. 1987; Tasaki et al. 1991; Kohta et al. 2010). Among them, 1-benzyl-1,2,3,4-tetrahydroisoquinoline (1BnTIQ) as well as its analogue 1-(3′,4′-dihydroxybenzyl)-1,2,3,4-tetrahydroisoquinoline (3′,4′DHBnTIQ) were detected in the brains of non-treated mice as endogenous TIQ derivatives (Kotake et al. 1995; Kawai et al. 1998). Moreover, on one hand, 1BnTIQ concentration was found to be three times higher in the CSF of patients with PD than in that of patients with other neurological diseases (Kotake et al. 1995) and repeated administration of 1BnTIQ induced parkinsonism in primates (Kotake et al. 1996). On the other hand, successive intraperitoneally administration of 3′,4′DHBnTIQ to mice induced PD-like symptoms, which were attenuated with L-DOPA treatment (Kawai et al. 1998). Unlike 1BnTIQ, 3′,4′DHBnTIQ contains a dopamine (DA) moiety, making it easy to be uptaken by dopaminergic neurons through DA transporters and exert dopaminergic neurons-specific toxicity (Kawai et al. 2000a,b). The catechol skeleton from the structure of 3′,4′DHBnTIQ can undergo autoxidation, leading to highly reactive quinone species which are expected to covalently bind proteins through Michael addition reaction and irreversibly affect their structure or function, as shown in Fig. 1.
Because of the features mentioned above, 3′,4′DHBnTIQ together with UCH-L1 seem to be promising research targets when studying the onset and progression of PD. In this study, we investigated whether 3′,4′DHBnTIQ interacts with UCH-L1 and the effects that this interaction may have on its structure or function.
Materials and methods
Chemicals and reagents
Human UCH-L1-pET15b vector and Escherichia coli BL21 cells were purchased from Novagen (Madison, WI, USA). Dulbecco's modified eagle medium was purchased from Nissui Pharmaceutical (Tokyo, Japan). l-glutamine, sodium fluoride, glycerol, avidin agarose, dopamine hydrochloride, m-aminophenylboronic acid agarose (PBA), protease inhibitor cocktail, anti-rabbit IgG peroxidase conjugate, anti-mouse IgG peroxidase antibody, anti-β-actin mouse monoclonal antibody (ab6276), and 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Penicillin G potassium and streptomycin sulfate were purchased from Meiji Seika Pharma (Tokyo, Japan). Sodium hydrogen carbonate was purchased from Kanto Chemical (Tokyo, Japan). Fetal bovine serum was purchased from PAA Laboratories (Pasching, Austria). Horse serum was purchased from Life Technologies (Carlsbad, CA, USA). Sodium orthovanadate, phenylmethylsulfonylfluoride, sodium chloride (NaCl), sodium dodecyl sulfate (SDS), hydrochloric acid (HCl), glycine, and dimethyl sulfoxide were purchased from Wako Pure Chemical Industries (Tokyo, Japan). Bromophenol blue was purchased from Katayama Chemical Industries (Osaka, Japan). N-6-(biotinylamino)hexa-noyl-N'-[2-(N-maleimido)ethyl]piperazone (biotin-PEAC5-ma-leimide), 2-(4-iodophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H-tetrazolium, monosodium salt (WST-1), and 1-methoxy- 5-methylphenazinium methyl sulfate were purchased from Dojindo (Kumamoto, Japan). 2-amino-2-hydroxymethylpropane-1,3-diol hydrochloride (Tris-HCl), nonidet P-40, sodium deoxycholate, 2-mercaptoethanol, bovine albumin, tetramethylethylenediamine, Chemi-Lumi One L, Chemi-Lumi One Super, and protease inhibitor cocktail were purchased from Nacalai Tesque (Kyoto, Japan). Anti-PGP9.5 (UCH-L1) rabbit polyclonal antibody (PA1-38446) was purchased from Thermo Fisher Scientific (Waltham, MA, USA). Anti-ubiquitin (P4D1) mouse monoclonal antibody (sc-8017) was purchased from Biomol International (Plymouth Meeting, PA, USA). LDN-57444 was purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Ubiquitin-7-amino-4-methylcoumarin (Ub-AMC) was purchased from Life Sensors (Malvern, PA, USA). Probenecid was purchased from Setareh Biotech (Eugene, OR, USA). 3′,4′DHBnTIQ was synthesized as described previously (Kawai et al. 1998).
Purification of recombinant human UCH-L1
E. coli BL21 cells were transformed with human UCH-L1-pET15b vector, then the cells were grown in LB medium at 37°C with shaking at 120 rpm (Taitec, Saitama, Japan). UCH-L1 expression was induced by incubation for 24 h at 30°C in the presence of 1 mM isopropyl-1-thio-β-d-galactopyranoside. The collected BL21 cells were suspended in lysis buffer (50 mM Tris-HCl pH 7.5, 0.1 mM sodium chloride, 10 mM 2-mercaptoethanol, 5% glycerol). The cells were sonicated and then centrifuged at 100 000 g for 1 h. The supernatant was applied to a ProBond Nickel-Chelating Resin column (40 × 10 mm inner diameter), purchased from Invitrogen (Carlsbad, CA, USA), which had previously been equilibrated with a solution containing 50 mM Tris-HCl (pH 7.5), 0.1 mM sodium chloride, and 10 mM 2-mercaptoethanol. The column was extensively washed with the equilibration buffer, and UCH-L1 was eluted by linear gradient with this buffer containing 0–150 mM imidazole. Purified UCH-L1 was incubated with 2 mM dithiothreitol for 1 h and the dithiothreitol was removed using an Econo-Pac 10 DG column purchased from Bio-Rad Laboratories (Hercules, CA, USA). The purified protein was stored at −80°C until further use.
MALDI-TOF MS analysis
Recombinant UCH-L1 protein (10 μM) was incubated with or without 3′,4′DHBnTIQ for 1 h at 37°C in buffer containing 20 mM Tris-HCl (pH 7.5). Next, the samples were digested by incubation with trypsin at 37°C for 16 h. The peptides were mixed with α-cyano-4-hydroxy-cinnamic acid (2.5 mg/mL) containing 50% acetonitrile and 0.1% trifluoroacetic acid and dried on stainless steel targets at 23°C. Analyses were performed using AXIMA-TOF2 (Shimadzu, Kyoto, Japan). All analyses were performed in the positive ion mode, and the instrument was calibrated prior to each series of studies.
Cell culture and treatments
Human neuroblastoma SH-SY5Y cells, purchased from American Type Culture Collection, were cultivated at 37°C under humidified 5% CO2 atmosphere, in Dulbecco's modified eagle medium supplemented with 0.584 g/L glutamine, 100 U/mL penicillin G, 0.1 mg/mL streptomycin, 0.2% sodium hydrogen carbonate, 5% fetal bovine serum, and 5% horse serum. The medium was changed every 2–3 days and cell passage was performed every time the cells became 80% confluent. For experiments, cells were seeded at a concentration of 3 × 105 cells/mL medium (1 × 105 cells/500 μL medium for WST-1 assay) and incubated for 24 h, then the medium was changed and treatment was performed.
3′,4′DHBnTIQ and LDN-57444 were prepared in dimethyl sulfoxide at a stock of 100 mM and 25 mM, respectively. Dopamine was prepared in ultrapure water at a stock of 100 mM. All chemicals were stored at −30°C after dissolution. Prior to addition, the chemicals were further diluted in medium or assay buffer for the desired final concentrations.
After treatment, SH-SY5Y cells were washed three times with phosphate-buffered saline, then lysed with 50 mM Tris buffer pH = 7.8, 150 mM NaCl, 1% nonidet P-40 (TNE buffer) supplemented with 1 mM sodium orthovanadate, 10 mM sodium fluoride, 1 mM phenylmethylsulfonylfluoride, and 1% (v/v) protease inhibitor cocktail, for 20 min, on ice. Then, cell lysate was centrifuged at 22 000 g for 20 min at 4°C. For sample, preparation from midbrains of MPTP-treated mice, radioimmunoprecipitation assay (RIPA) buffer was used instead of TNE buffer. Samples were homogenized on ice, then the same procedures as in the case of cells were applied. The supernatant stood for the soluble fraction of cells and the precipitate was resuspended in 2% SDS solution supplemented with 1% protease inhibitor cocktail and stood for the insoluble fraction of cells. For biotin-PEAC5-maleimide (BPM) precipitation assay, RIPA buffer (50 mM Tris buffer pH = 8, 0.1% SDS, 1% nonidet P-40, 0.5% sodium deoxycholate, 150 mM NaCl) supplemented with 1% protease inhibitor cocktail was used instead of TNE buffer and only the soluble fraction of cells was considered for assays. Protein concentration of both soluble and insoluble fractions was determined with Pierce® BCA Protein Assay Kit (Thermo Scientific, Rockford, IL, USA). Cell lysates (soluble and insoluble fractions) were heated with Laemmli's sample buffer (Laemmli 1970) at 95°C for 5 min and then stored at −30°C until analysis.
Cell viability assay
Viability of SH-SY5Y cells was measured with WST-1 assay. Cells were seeded in a 24-well plate (Becton Dickinson Labware, Franklin Lakes, NJ, USA) for this assay. After treatment, cells were incubated for 1 h with 300-μL medium supplemented with 0.42 mM WST-1 and 16.7 μM 1-methoxy-5-methylphenazinium methylsulfate per well. Next, 200 μL of the medium was transferred to a 96-well plate (Santa Cruz Biotechnology, Santa Cruz, CA, USA) and its absorbance at 415 nm was measured with Perkin Elmer EnSpire® Multimode Plate Reader (Waltham, MA, USA).
PBA pull-down assay
PBA pull-down assay was performed as described in the literature, with several modifications (LaVoie et al. 2005). A volume of the soluble fraction containing 200-μg protein (500-μg protein for midbrain samples) was diluted with 100 mM Tris buffer (pH = 8.6) and the final volume was adjusted to 1 mL. When using recombinant UCH-L1, 1-μg protein was incubated with 3′,4′DHBnTIQ or DA for 30 min at 37°C in 50 mM phosphate buffer, then the final volume was adjusted to 1 mL with 100 mM Tris buffer (pH = 8.6). Next, previously washed PBA beads were added and the mixture was incubated overnight at 4°C, with rotation. Next day, PBA beads were separated with a microcolumn, and next washed with 1 mM Tris buffer (pH = 8.6) three times. Then, bound proteins were released by elution with 20 μL of 50 mM glycine (pH = 2), and the pH was adjusted with 6 μL of 1 M Tris buffer (pH = 8). 26 μL of Laemmli's sample buffer was added and samples were heated at 95°C for 5 min, then stored at −30°C until analysis.
BPM precipitation assay
In the case of treatment to cells or of midbrain samples, BPM precipitation assay was performed as described in the literature (Toyama et al. 2013) with slight modifications. In the case of incubation with cell lysate, a volume of the soluble fraction of untreated cells, containing 100-μg protein, was further diluted with RIPA buffer and the final volume was adjusted to 990 μL. When using recombinant UCH-L1, 1-μg of protein was used instead of the soluble fraction and RIPA buffer was replaced with 50 mM phosphate buffer. Then, 1 μL of 100 mM 3′,4′DHBnTIQ or DA was added and the mixture was incubated at 37°C for 30 min. Next the same procedures as in the case of treatment to cells were performed.
Equal amount of lysate protein was loaded and separated on SDS–polyacrylamide gel electrophoresis gel, then transferred to a polyvinylidene difluoride membrane, which was next blocked for 1 h with 5% skim milk (2% for pull-down or precipitation assays) in TBS supplemented with 1% polysorbate 20 (TBS-T). Next, the membranes were incubated with primary antibodies at 4°C overnight, washed three times with TBS-T, and then incubated with secondary antibodies at 23°C for 1 h. After incubation, the membranes were washed three times with TBS-T and then desired proteins were detected with enhanced chemiluminescence (Chemi-Lumi One L or Super), using GE Healthcare Life Sciences ImageQuant LAS 4000 (Waukesha, WI, USA). Band intensity was quantified with the software Image J (National Institutes of Health, Bethesda, MD, USA).
UCH-L1 hydrolase activity measurement
This experiment was performed at 23°C. Recombinant UCH-L1 (100 nM) was incubated with different concentrations of 3′,4′DHBnTIQ, DA, or LDN-57444 for 1 h, then the mixture was transferred to a 96-well black assay plate, purchased from Perkin Elmer. Equal volume of 1.6 μM Ub-AMC was added and after 15 min, the fluorescence emission of AMC at 460 nm (excitation at 380 nm) was measured with Perkin Elmer EnSpire® Multimode Plate Reader.
Docking simulation and graphical presentation
Molecular docking was performed using the software AutoDock Vina (Trott and Olson 2010). The structures of human UCH-L1 as well as of human S18YUCH-L1 were obtained from RCSB Protein Data Bank (Das et al. 2006; Boudreaux et al. 2010) and were further optimized for docking simulation using AutoDockTools software (Sanner 1999; Morris et al. 2009). Water molecules as well as unnecessary ligands or ions were removed and missing hydrogen atoms were added. Monomeric UCH-L1 (wild type as well as S18Y variant) was used for the study and the side chains of the amino acids comprising loop L8 were set flexible. The grid comprised the entire structure of the protein and the conformation with the highest binding affinity was considered as model of interaction. The structures of ligands were constructed with the software ChemBioDraw (CambridgeSoft, Cambridge, MA, USA) and geometry optimization was performed. All rotatable bonds of the ligands were permitted free rotation. Figure 8a and b were generated with the software UCSF Chimera (Pettersen et al. 2004).
MPTP-treated PD model mice
This study was approved by Hiroshima University's Animal Ethics Committee. MPTP-treated PD model mice were produced according to the methods described in the previous reports (Lau et al. 1990; Meredith et al. 2008). Male C57BL/6 mice (8 weeks old) were injected with MPTP hydrochloride (25 mg/kg in saline, i.p.) as well as with probenecid (250 mg/kg in Tris-HCl buffer, i.p.) over 4 weeks at intervals of 3.5 days. Control mice were injected only with probenecid under the same schedule. The whole brains were removed 4 days after the last injection, then midbrains were excised and used for experiments immediately.
All data are expressed as mean + S.D. of at least three independent experiments. Statistical analysis was performed by anova followed by Tukey's test. p < 0.05 was considered significant.
3′,4′DHBnTIQ conjugates UCH-L1 at Cys152 in vitro
If 3′,4′DHBnTIQ undergoes readily autooxidation and yields its quinone form under physiological conditions (see Fig. 1), this electrophilic form would modify UCH-L1 through its cysteine residues (Kumagai et al. 2012). To address this possibility, we incubated recombinant UCH-L1 in the absence and presence of 3′,4′DHBnTIQ at 37°C, then digested the protein with trypsin and subjected the mixture to matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) for analysis (Fig. 2a). Compared with the calculated mass of the unmodified peptides, modified peptide P-1 showed a 250 Da increase in mass, corresponding to the addition of a single equivalent of the oxidized form of 3′,4′DHBnTIQ. The corresponding peptide sequences of arrowed peaks are shown in Fig. 2b. These data suggest that 3′,4′DHBnTIQ is easily oxidized to its quinoid structure, which is able to covalently conjugate UCH-L1 through Cys152.
Effect of 3′,4′DHBnTIQ on the survival rate of SH-SY5Y cells
Before investigating whether the interaction between 3′,4′DHBnTIQ and UCH-L1 also occurs in cell system, we first measured the survival rate of SH-SY5Y cells under 3′,4′DHBnTIQ short-term treatment and determined its proper concentration to be used for further experiments. 3′,4′DHBnTIQ treatment induced SH-SY5Y cell death in a time- and concentration-dependent manner (Fig. 3a and b). However, 6-h treatment of 100 μM 3′,4′DHBnTIQ had no effect on the survival rate of SH-SY5Y cells. We used these conditions for the following experiments.
3′,4′DHBnTIQ covalently modifies UCH-L1 in cell system
It is still problematic to directly detect cysteine adducts of proteins from cells or from in vivo samples even with the latest techniques. For this reason, to detect whether 3′,4′DHBnTIQ conjugates UCH-L1 in cell system, we used two indirect pull-down assays, PBA pull-down assay, and BPM-precipitation assay. With PBA pull-down assay, proteins modified by catechol derivatives can be detected, while with BPM precipitation assay, proteins which have free cysteine residues, in other words which are not modified by electrophiles, can be identified. Before applying these pull-down assays to cell lysates, we first confirmed their usefulness in vitro. A proteomic study suggested UCH-L1 to be conjugated by DA quinone (Van Laar et al. 2009). Therefore, we also investigated the effect of DA as comparative control, together with 3′,4′DHBnTIQ. Recombinant UCH-L1 was incubated with 3′,4′DHBnTIQ or DA, then the mixtures were subjected to the two assays. Next, the eluted proteins were separated by SDS–polyacrylamide gel electrophoresis and then analyzed for UCH-L1 immunoreactivity. Incubation of UCH-L1 with 3′,4′DHBnTIQ led to an increase in the amount of UCH-L1 eluted with PBA beads and to a decrease in the amount of UCH-L1 eluted with avidin agarose, supporting the MALDI-TOF MS data that 3′,4′DHBnTIQ covalently modifies UCH-L1. Similar results were obtained in the case of DA, as well (Fig. 4a). In addition, we have also investigated the binding between recombinant UCH-L1 and 3′,4′DHBnTIQ in vitro at lower concentrations, under the same conditions. UCH-L1 modification by 3′,4′DHBnTIQ could be detected at 1 μM as well, suggesting that lower concentrations of 3′,4′DHBnTIQ also conjugate UCH-L1 (data not shown).
Next, we treated SH-SY5Y cells with 3′,4′DHBnTIQ or DA, then we subjected cell lysates to the pull-down assays. We have confirmed that 6-h treatment of 100 μM DA does not decrease the survival rate of SH-SY5Y cells. In the case of PBA pull-down assay, treatment with 3′,4′DHBnTIQ led to a significant increase in the amount of detected UCH-L1, suggesting that 3′,4′DHBnTIQ covalently modifies UCH-L1 (Fig 4b). In the case of BPM precipitation assay, treatment with 3′,4′DHBnTIQ as well as with DA led to a significant decrease in the amount of UCH-L1 protein containing unbound cysteine residues (Fig. 4c). We also incubated untreated cell lysate with 3′,4′DHBnTIQ or DA, and next, subjected the mixture to BPM precipitation assay. Direct incubation also led to a significant decrease in the amount of eluted UCH-L1 (Fig. 4d). Together, these data suggest that 3′,4′DHBnTIQ covalently modifies UCH-L1 not only in vitro but also in cell system and that DA may also conjugate UCH-L1.
3′,4′DHBnTIQ induces aggregation of UCH-L1
Insoluble UCH-L1 has been detected in the post-mortem brains of patients with Parkinson's disease (Schlossmacher et al. 2002). However, the factors which cause the insolubility of UCH-L1 are still not clear. Electrophile-protein reactions are associated with various cellular processes (Rudolph and Freeman 2009). We thought that electrophilic conjugation may be a factor and therefore we investigated the effect of 3′,4′DHBnTIQ as well as of DA and UCH-L1 inhibitor LDN-57444 on the cellular level of UCH-L1. Treated cell lysates were subjected to western blotting and immunoreactivity for UCH-L1 was analyzed separately for the soluble and insoluble fractions of SH-SY5Y cells. In the soluble fraction, no significant change in the level of UCH-L1 was observed (Fig. 5a). However, in the insoluble fraction, 3′,4′DHBnTIQ and DA but not LDN-57444 treatment led to a significant increase in the amount of UCH-L1, suggesting that 3′,4′DHBnTIQ and DA induce the aggregation of UCH-L1 in cell system (Fig. 5b).
3′,4′DHBnTIQ reduces the hydrolase activity of UCH-L1
The enzymatic activity of UCH-L1 is necessary to maintain the well functioning of neuronal cells. Therefore, we investigated the effect of 3′,4′DHBnTIQ on the hydrolase activity of UCH-L1 in vitro. We incubated recombinant UCH-L1 with 3′,4′DHBnTIQ, DA or LDN-57444, then we measured its enzymatic activity using the fluorogenic substrate Ub-AMC. UCH-L1 cleaves the bond between the N-terminal glycine of ubiquitin and AMC and the fluorescence emission of free AMC reflects the enzymatic activity of UCH-L1. 1 μM 3′,4′DHBnTIQ had no effect on the hydrolase activity of UCH-L1, whereas 10 or 100 μM of 3′,4′DHBnTIQ significantly reduced its activity to approximately 60%. Incubation of UCH-L1 with 100 μM DA led to a 60% decrease in its hydrolase activity, whereas incubation with 10 μM LDN-57444, a reversible active site-directed inhibitor of UCH-L1, reduced its hydrolase activity to approximately the same level as 3′,4′DHBnTIQ (Fig. 6).
3′,4′DHBnTIQ reduces the level of ubiquitin in cell system
UCH-L1 is thought to be crucial for the homeostasis of ubiquitin in cells. Chemical inhibition of UCH-L1 by LDN-57444 was reported to decrease the level of ubiquitin and alter synaptic structure of neurons (Cartier et al. 2009). Post-translational modifications by ubiquitin are known to regulate various cellular processes like protein degradation or apoptosis. Because inhibition of UCH-L1 was detected in vitro, we next examined the effect of 3′,4′DHBnTIQ, DA or LDN-57444 on the level of ubiquitin in SH-SY5Y cells. As a result, 3′,4′DHBnTIQ as well as LDN-57444 treatment reduced to half the level of free ubiquitin in cell system, while DA treatment led to just a 35% decrease of ubiquitin (Fig. 7). DA inhibited UCH-L1 stronger than 3′,4′DHBnTIQ in vitro but the effect was opposite in the case of monoubiquitin (Fig. 7).
3′,4′DHBnTIQ directly interacts with UCH-L1 in silico
Incubation of recombinant UCH-L1 as well as SH-SY5Y cells treatment with 3′,4′DHBnTIQ or DA led to a decrease in the hydrolase activity of UCH-L1. This may occur because of Cys152-conjugation of UCH-L1 by 3′,4′DHBnTIQ. However, 3′,4′DHBnTIQ may directly inhibit the hydrolase activity of UCH-L1, as well. To analyze this possibility, we applied molecular docking simulation to UCH-L1 and 3′,4′DHBnTIQ, DA, or LDN-57444. Interestingly, the most stable conformation, with an affinity energy of −7.7 kcal/mol for (S)-3′,4′DHBnTIQ and of -7.9 kcal/mol for (R)-3′,4′DHBnTIQ, shows 3′,4′DHBnTIQ in close contact with the amino acids Leu52, Gln84, Val158, Phe160, and Arg178, which are all around Cys90, the active site of UCH-L1. 3′,4′DHBnTIQ acts like a lid, which covers Cys90 (Fig. 8c). LDN-57444 also interacted with UCH-L1 (Fig. 8d). The affinity energy was −7.1 kcal/mol and the amino acids in close contact were Gln84, Gly87, Asn88, and Val158. Active site-directed inhibition could be visualized (Fig. 8d). DA was in close contact with the amino acids Val75, Pro126, and Phe181 and formed hydrogen bonds with the amino acids Phe81 and Arg129. The highest affinity energy was −5.3 kcal/mol (Fig. 8e). S18YUCH-L1 is associated with a reduced risk of developing PD. We also investigated its interaction with 3′,4′DHBnTIQ. To our surprise, the interaction site of S18YUCH-L1 differed from the interaction site of wild-type UCH-L1. 3′,4′DHBnTIQ was in close contact with the amino acids Val31, Val212, Phe214, Ser215, and Ala216, which are apart from the active site of UCH-L1. The affinity energy was −6.4 kcal/mol for (R)-3′,4′DHBnTIQ and -6.2 kcal/mol for (S)-3′,4′DHBnTIQ (Fig. 8f).
Catechol-modified UCH-L1 detected in midbrains of MPTP-treated PD model mice
We have showed that 3′,4′DHBnTIQ conjugates UCH-L1 at position Cys152 and alters its structure and function in SH-SY5Y cells as well as in vitro. 3′,4′DHBnTIQ seems to be one endogenous dopamine derivative candidate which conjugates and alters UCH-L1. To investigate whether UCH-L1 is covalently modified in vivo as well, we investigated the presence of electrophilic-modified UCH-L1 as well as the expression level of UCH-L1 in the midbrain of MPTP-treated PD model mice. RIPA buffer-soluble fraction of midbrains was subjected to PBA pull-down assay and to BPM precipitation assay, then immunoreactivity for UCH-L1 was detected. In the midbrains of MPTP-treated mice, an increase in the catechol-modified UCH-L1 and a decrease in unmodified UCH-L1 was observed (Fig. 9a), suggesting that UCH-L1 altered by cysteine modification exists in the brains of MPTP-treated PD model mice. Furthermore, increased levels of UCH-L1 were detected in the RIPA buffer-insoluble fraction of midbrains from MPTP-treated mice (Fig. 9c) while no change in the level of soluble UCH-L1 was observed (Fig. 9b). These data are consistent with our findings in vitro and suggest that UCH-L1 modification by catechol derivatives may occur in vivo, too.
UCH-L1 is one of the highly expressed proteins in the brain, mostly known for its involvement in ubiquitin turnover in cells. While UCH-L1 stabilizes ubiquitin and avoids its degradation (Osaka et al. 2003), reversible monoubiquitination negatively regulates the activity of UCH-L1 (Meray and Lansbury 2007). Thus, the balance between UCH-L1 and ubiquitin is believed to be crucial for the normal functioning of neuronal cells. A glutamate to alanine (E7A) mutation, which almost abolishes the hydrolase activity of UCH-L1 and reduces its binding affinity for ubiquitin, leads to early-onset autosomal recessive progressive neurodegeneration (Bilguvar et al. 2013). Moreover, I93M mutation halves the hydrolase activity of UCH-L1, reduces its solubility and its affinity for ubiquitin, finally leading to late-onset autosomal dominant PD. In an effort to find a connection between genetic and idiopathic PD through UCH-L1, we report here an interesting interaction between this protein and the brain endogenous parkinsonism inducer 3′,4′DHBnTIQ.
It is generally recognized that when a nucleophilic attack of protein thiols to quinones occurs, the reaction products bound to the protein are converted into hydroquinones according to Michael addition reaction. However, our MALDI-TOF MS analysis revealed that the molecular mass of an adduct bound to UCH-L1 during incubation of this protein with 3′,4′DHBnTIQ was indeed 250 Da, suggesting that 3′,4′DHBnTIQ moiety on UCH-L1 further undergoes autooxidation (Fig. 2). The chemical characteristic of quinone form of 3′,4′DHBnTIQ was in agreement with that of 1,2-naphthoquinone or 2-tert-butyl-1,4-benzoquinone (Miura et al. 2011a,b; Abiko and Kumagai 2013). While there are six cysteine residues in human UCH-L1, oxidized 3′,4′DHBnTIQ specifically modified UCH-L1 through Cys152. This cysteine, unlike the other cysteine residues, is located on the exterior surface of the protein, in loop L8, a region of the protein which can be easily accessed by electrophiles (Fig. 8a). Loop L8 comprises the amino acids Gly150 to Lys157 and is a flexible region in UCH-L1 (Fig. 8a) which seems to be essential for maintaining the tertiary structure of the protein and which connects its two lobes, like a bridge (Das et al. 2006). Thus, because this covalent modification may affect the 3D structure or function of UCH-L1, we next investigated whether 3′,4′DHBnTIQ covalently modifies UCH-L1 in cell system, as well (Fig. 4). Together with 3′,4′DHBnTIQ, we also examined the effect that DA may have on UCH-L1. To detect modified UCH-L1, we used one direct (PBA pull-down assay) and one indirect (BPM precipitation assay) assay. 3′,4′DHBnTIQ as well as DA treatment led to an increase in catechol derivative-bound UCH-L1 (Fig. 4b) and to a decrease in electrophile-unbound UCH-L1 (Fig. 4c). These data suggest that 3′,4′DHBnTIQ conjugates UCH-L1 both in vitro and in cell system with a higher affinity than DA. This conjugation may be important when studying neurotoxicity against dopaminergic neurons.
Our data also show that treatment of SH-SY5Y cells with 3′,4′DHBnTIQ leads to a significant increase in the amount of UCH-L1 in the insoluble fraction of the cells (Fig. 5b), while no major changes occur in the soluble fraction (Fig. 5a). Conjugation of 3′,4′DHBnTIQ to Cys152 could disrupt the hydrophobic interactions inside the protein and destabilize the protein, inducing unfolding and finally aggregation (Fig. 8b). It has been previously reported that 15deoxy-Δ12,14-prostaglandin J2 covalently modifies UCH-L1 in vitro only at Cys152 and it destroys its native structure, resulting in unfolding and aggregation (Koharudin et al. 2010). It seems that Cys152 is crucial for the well folding of UCH-L1 and Cys152-electrophilic modifications severely compromise the tertiary structure of UCH-L1. Insoluble UCH-L1 has been detected in Lewy bodies (Schlossmacher et al. 2002) in patients with PD and I93MUCH-L1, the cause of a form of genetic PD, has low solubility (Nishikawa et al. 2003). It seems that the insolubility of UCH-L1 is related to PD and Cys152-electrophilic modifications could be one mechanism through which it occurs. DA treatment also increased the level of UCH-L1 in the insoluble fraction of SH-SY5Y cells. However, further studies are needed to make clear the mechanism through which DA-induced aggregation occurs.
A closer look at the tertiary structure of UCH-L1 reveals that loop L8 is not only important for the well folding of the protein but it also crosses over the tunnel that leads to Cys90, the active site of UCH-L1 (Fig. 8a). Conjugation of Cys152 with 3′,4′DHBnTIQ could disrupt the interactions between the amino acids around the opening to the active site of UCH-L1, thus hindering the access of the substrate and, as a result, lead to an apparent decrease in the hydrolase activity of UCH-L1. Also, 3′,4′DHBnTIQ could directly inhibit the hydrolase activity of UCH-L1, reason why we applied molecular docking to UCH-L1 and 3′,4′DHBnTIQ. Interestingly, the most stable conformation shows 3′,4′DHBnTIQ to act like a lid which covers Cys90, the active site of UCH-L1, suggesting that not only the Cys152-bound 3′,4′DHBnTIQ but also free 3′,4′DHBnTIQ might inhibit the enzymatic activity of UCH-L1 (Fig. 8c). We also applied molecular docking to S18YUCH-L1 and 3′,4′DHBnTIQ (Fig. 8f). To our surprise, 3′,4′DHBnTIQ interacted with a different site of the protein, apparently not related with the enzymatic activity site, and the affinity energy was lower. Similar results were obtained in the case of DA and UCH-L1. We also applied molecular docking to UCH-L1 and its direct site inhibitor, LDN-57444 and direct inhibition could be visualized (Fig. 8d), indicating the accuracy of the simulation. It seems that interaction of small molecules with the pocket under loop L8 from the structure of UCH-L1 might be somehow related with the alteration of UCH-L1. Deeper investigation about the effect of small molecules which interact with the pocket under loop L8 on UCH-L1 might be interesting. Incubation of 3′,4′DHBnTIQ with recombinant UCH-L1 led to a decrease in the hydrolase activity of UCH-L1 only to approximately 60% (Fig. 6). This result is consistent with our computer-generated data (Fig. 8) which suggests indirect inhibition of UCH-L1 by 3′,4′DHBnTIQ. Unlike 3′,4′DHBnTIQ, LDN-57444 treatment did not change the solubility of UCH-L1 (Fig. 5b) although both compounds interact with the amino acids surrounding the active site of UCH-L1 (Fig. 8c and d). These data suggest that just protein–ligand interaction or enzymatic inhibition does not induce aggregation of UCH-L1 and provide supportive data that it is through Cys152-covalent binding that UCH-L1's solubility decreases. 3′,4′DHBnTIQ treatment reduced to half the level of monoubiquitin in SH-SY5Y cells. 100 μM 3′,4′DHBnTIQ and 10 μM LDN-57444, which inhibited to the same extent the enzymatic activity of UCH-L1 in vitro, reduced to the same extent the level of monoubiquitin in cells, suggesting that the inhibition in the enzymatic activity of UCH-L1 which we observed in vitro, also occurs in cell system (Fig. 7). DA inhibited UCH-L1 stronger than 3′,4′DHBnTIQ in vitro (Fig. 6). However, the inhibition was markedly reduced in SH-SY5Y cells (Fig. 7). Similarly, according to the results of the pull-down assays, 3′,4′DHBnTIQ exhibited a higher affinity for UCH-L1 than DA (Fig. 4). DA is a representative catechol in the brain and its concentration is substantially higher than the concentration of neuronal metabolites like TIQs. However, as soon as it is synthesized in neurons, DA is further stored in monoaminergic synaptic vesicles by vesicular monoaminergic transporter-2, a DA transporter localized in the membrane of these vesicles (Miller et al. 1999). DA uptake into monoaminergic vesicles prevents both the accumulation of free dopamine in the cytosol and the oxidation of DA to highly reactive quinones (Guillot and Miller 2009; Segura-Aguilar et al. 2014). However, this protection mechanism might not apply for neuronal metabolites like 3′,4′DHBnTIQ. We have identified no such reports up to the present time. Therefore, compounds like 3′,4′DHBnTIQ seem more likely to exert neurotoxicity than dopamine. It would be interesting in further studies to make clear the mechanism through which 3′,4′DHBnTIQ-induced cell death occurs and the implications that our present study have in it.
We were interested in finding out whether UCH-L1 is covalently modified in vivo as well, and therefore investigated the presence of electrophilic modified as well as the expression level of UCH-L1 in the midbrain of MPTP-treated PD model mice. As a result, an increase in the catechol-modified UCH-L1 and a decrease in unmodified UCH-L1 was observed (Fig. 9a), suggesting that UCH-L1 altered by cysteine modifications exists in the brains of MPTP-treated PD model mice. Furthermore, increased levels of UCH-L1 were detected in the insoluble fraction of midbrains from MPTP-treated mice (Fig. 9c), whereas no change in the level of soluble UCH-L1 was observed (Fig. 9b). These data are consistent with our finding in this study and suggest that UCH-L1 modification by catechols may occur in vivo too, in PD model mice.
Comparing to the wild-type UCH-L1, the enzymatic activity of I93MUCH-L1 is decreased to 55% and I93MUCH-L1 also exhibits a lower interaction level with ubiquitin and a lower solubility (Nishikawa et al. 2003). If we have a look at 3′,4′DHBnTIQ-conjugated UCH-L1 and at I93MUCH-L1, it seems that 3′,4′DHBnTIQ-conjugated UCH-L1 have common characteristics with the mutant form responsible for genetic PD. It is believed that I93M mutation causes parkinsonism not only because of a loss-of-function but also because of a gain-of-function of the toxic UCH-L1, which is yet to be clarified (Setsuie and Wada 2007). Structurally as well as functionally altered UCH-L1 have been detected in the brains of patients with idiopathic PD (Choi et al. 2004). Conjugation of UCH-L1 by neurotoxic dopamine derivatives like 3′,4′DHBnTIQ may be a new start, a possible clue for the clarification of the mechanism through which PD develops.
Acknowledgments and conflict of interest disclosure
We thank the Analysis Center of Life Science, Hiroshima University for the use of their facilities. This work was mainly supported by JSPS Grant-in-Aid for Scientific Research (B) No. 24406004, 24651061 (to Yaichiro Kotake) and Grant-in-Aid for Challenging Exploratory Research.
All experiments were conducted in compliance with the ARRIVE guidelines. The authors declare that there are no conflicts of interest.