N-Terminal Truncated UCH-L1 Prevents Parkinson's Disease Associated Damage

Ubiquitin C-terminal hydrolase-L1 (UCH-L1) has been proposed as one of the Parkinson's disease (PD) related genes, but the possible molecular connection between UCH-L1 and PD is not well understood. In this study, we discovered an N-terminal 11 amino acid truncated variant UCH-L1 that we called NT-UCH-L1, in mouse brain tissue as well as in NCI-H157 lung cancer and SH-SY5Y neuroblastoma cell lines. In vivo experiments and hydrogen-deuterium exchange (HDX) with tandem mass spectrometry (MS) studies showed that NT-UCH-L1 is readily aggregated and degraded, and has more flexible structure than UCH-L1. Post-translational modifications including monoubiquitination and disulfide crosslinking regulate the stability and cellular localization of NT-UCH-L1, as confirmed by mutational and proteomic studies. Stable expression of NT-UCH-L1 decreases cellular ROS levels and protects cells from H2O2, rotenone and CCCP-induced cell death. NT-UCH-L1-expressing transgenic mice are less susceptible to degeneration of nigrostriatal dopaminergic neurons seen in the MPTP mouse model of PD, in comparison to control animals. These results suggest that NT-UCH-L1 may have the potential to prevent neural damage in diseases like PD.


Introduction
Ubiquitin C-terminal hydrolase-L1 (UCH-L1) catalyzes the hydrolysis of C-terminal ubiquitin esters and amides. UCH-L1 is highly expressed in metastatic lung cancer [1] and is abundant in brain, comprising 1-2% of total brain protein [2], and is a major component of the protein aggregates called Lewy bodies found in the brains of PD patients [3]. Also, a mutant of UCH-L1, I93M (Ile93 to Met), was shown to cause a type of autosomal dominant PD in one German family [4]. These disparate observations have led to a suggestion that UCH-L1 may be a PD related gene. However, the molecular connection between UCH-L1 and PD was not fully established.
Mutations, environmental stresses, and aging cause protein denaturation, rendering them aggregation-prone forms [5]. Chaperones play roles in refolding the denatured proteins or for clearing these in ubiquitin-proteasome system and in autophagy [6][7][8]. However, when these defense systems fail to repair, protein aggregates accumulate and induce cell death [5]. Parkinson's disease (PD) is known to associated with formation of protein aggregates and Lewy bodies as hallmarks of PD [9,10]. Although there is evidence that protein aggregates are toxic to cells [11,12], it is not a necessary and sufficient condition to develop PD in human patients and animal models [13][14][15]. In fact, Lewy bodies have been suggested to have neuroprotective effect [16][17][18]. For example, a-synuclein, another PD causing protein, is known to form various oligomeric structures, which show both toxicities and protective effects on cells [19,20].
Post-translational modifications of proteins and alternative splicing can change biochemical properties including solubility of a protein. Phosphorylation, ubiquitination, and truncation affect aggregation behavior of a-synuclein [21]. In the case of UCH-L1, carbonylation and modification by cyclopentenone prostaglandin, decrease its solubility supporting its possible relationship with PD [22,23]. Monoubiquitination of UCH-L1 was also reported to restrict its hydrolase activity [24], but its effect on the solubility of UCH-L1 has not been studied. In general, polyubiquitination is required for proteasomal degradation of a protein while monoubiquitination enables its participation in DNA repair, histone regulation, gene expression, and receptor endocytosis [25].
Oxidative stress has been shown to be the cause of nigrostriatal dopaminergic neuron loss in PD patients and in the MPTP mice model of PD [34]. Toxins such as MPTP, rotenone, 1,19-dimethyl-4,49-bipyridinium dichloride (paraquat), and 6-hydroxydopamine (6-OHDA), which induce PD like symptoms in mice are all oxidative stress inducers [35][36][37][38]. Cysteine thiol group (-SH) of proteins are susceptible to oxidative stress and are readily oxidized to disulfide, sulfenic acid, sulfinic acid and sulfonic acid. Disulfide crosslinking resulting in protein aggregation, has been demonstrated in the development of various diseases [39,40].
In this study we identified a variant of UCH-L1 lacking Nterminal 11 amino acids designated as NT-UCH-L1, and compared it to UCH-L1 by physical, chemical, and proteomic approaches with the goal of understanding their possible relation to PD. We found that NT-UCH-L1 is aggregation prone and is localized in mitochondria, which are regulated by monoubiquitination. Furthermore, NT-UCH-L1 was found to have a protective role in the PD model in vitro and in vivo.

Cell culture
Human NCI-H157, SH-SY5Y, and HeLa cells were maintained respectively in RPMI 1640, dulbecco's modified eagle medium and minimum essential medium supplemented with 10% fetal bovine serum, 100 mM/mL of streptomycin and 100 units/ mL of penicillin G (all from Invitrogen).

Antibodies
The sources of antibodies used in this study were as follows: Monoclonal anti-myc antibody from Millipore, polyclonal anti-UCH-L1 and anti-ubiquitin antibody from Chemicon, anti-flag antibody (M2) from Sigma, anti-tubulin antibody from Santa Cruz Biotechnology, Alexa Fluor 488 goat anti-mouse antibody from Molecular Probes. Polyclonal antibodies against N-terminal peptide of UCH-L1 were generated in rabbits and characterized by Abclon (Korea).

2D gel electrophoresis
This was performed as described previously [41].

Establishment of UCH-L1 expressing stable cell lines
The lentiviral vector for expressing wild type UCH-L1 and NT-UCH-L1 was constructed by inserting PCR gene fragment into the XbaI-EcoRV site of LentiM1.4 vector. Pseudotyped lentiviruses were produced and used for stable expression of UCH-L1s as described earlier [42].

Measurement of ubiquitin C-terminal hydrolase activity
Ubiquitin C-terminal hydrolase activity was measured using the Ub-AMC (AG scientific) as substrate [1]. Five nM of GST-UCH-L1 or GST-NT-UCH-L1 were incubated with 0-1000 nM of Ub-AMC and the release of free AMC monitored at 460 nm.
Hydrogen/Deuterium exchange (HDX) mass spectrometry UCH-L1 and NT-UCH-L1 (about 1 mg/mL) were diluted 10fold with D 2 O and maintained at 25uC for several time scales. The labeling reaction was quenched by 5 mM tris(2-carboxyethyl)phosphine (TCEP), pH 2.3. For peptic digestion, porcine pepsin (1 mg/mL) was added to each quenched protein sample and incubated at 0uC for 3 min before injection [43,44]. Peptic peptides were desalted and separated as previously described [41]. The autosampler chamber was set at 5uC. The trap, analytical column and all tubing were immersed in an ice bath to minimize deuterium back-exchange. Both mobile phase bottles containing 0.1% formic acid were placed on ice. Gradient chromatography was performed at a flow rate 0.6 mL/min and was sprayed on line to nanoAcquity TM /ESI/MS (SYNAPT TM HDMS TM , Waters). All mass spectral measurements were taken at: capillary voltage 2.5 kV, cone voltage 35 V, extraction cone voltage 4.0 V, and source temperature 80uC. TOF mode scan was performed in the range of m/z 300-1500 with scan time of 1 s.

Mass spectrometry
The gel spots of proteins were destained and digested with trypsin and the resulting peptides were analyzed by nanoAcquity UPLC/ESI/MS (SYNAPT HDMS, Waters) as described previously [41,45,46]. The peptides were separated using a C18 reversed-phase 75 mm i.d. x 150 mm analytical column (1.7 mm particle size, BEH130 C18, Waters) with an integrated electrospray ionization PicoTip (610 mm, New Objective). Seven mL of peptide mixtures were dissolved in buffer A (Water/formic acid; 100:0.1, v/v), injected on the column and eluted by a linear gradient of 5-80% buffer B (acetonitril/formic acid; 100:0.1, v/v) over 120 min. Samples were desalted on line prior to separation using a trap column cartridge (ID 180 mm x 20 mm, Symmetry C18, Waters). Initially, the flow rate was set to 300 nL/min and a capillary voltage of 2.5 keV was applied to the LC mobile phase before spray. Chromatography was performed on line to mass spectrometer. MS parameters for efficient data-dependent acquisition were: intensity of .10 and 3 components to be switched from MS to MS/MS analysis.
The individual MS/MS spectra acquired from each of the precursors within a single LC run were combined, smoothed, deisotoped and centroided using the Micromass ProteinLynx Global Server (PLGSTM) 2.1 data processing software and output, as a single MASCOT-searchable peak list (.pkl) file. The peak list files were used to query the SwissProt database using the MASCOT (global search engine), with the following parameters: peptide mass tolerance, 0.2 Da; MS/MS ion mass tolerance, 0.2 Da; allowing up to 1 missed trypsin cleavage site, considering variable modifications, such as acetylation, deamidation, pyro-glu (N-term E, Q), oxidation, formylation, phosphorylation, carbamidomethyl and cysteine propionamide but not fixed modifications; enzyme limited to trypsin; and toxonomy limited to mouse. All reported assignments were verified by automatic and manual interpretations of spectra from MASCOT.

Preparation of soluble and insoluble fractions of cell lysates
Soluble and insoluble fractions of cell lysates were prepared as previously described [47]. Briefly, cells from 100 mm cell culture plates were lysed in 500 mL RIPA buffer (50 mM Tris-Cl, pH 8.0, 150 mM NaCl, 1% (v/v) NP-40, 0.5% sodium deoxycholate and 0.1% SDS, protease inhibitor cocktail (Sigma), 10 mM NEM) by passing 10 times through a 26.5-gauge needle. After centrifugation at 12,000 x g for 20 min, pellets were resuspended in 200 mL of gel sample buffer (62.5 mM Tris-Cl, pH 6.8, 10% (w/v) glycerol, 5% b-mercaptoethanol, 2.3% SDS). The supernatant was mixed with same volume of gel sample buffer and 20 mL of each fraction were resolved in 10% SDS-PAGE.

Immunoprecipitation
Cells or mouse brain tissues were lysed with immunoprecipitation buffer (50 mM Tris-Cl, pH 7.4, 50 mM NaCl, 0.5 mM EDTA, 1 mM PMSF, 5 mg/mL aprotinin, 10 mg/mL leupeptin, 10 mg/mL pepstatin A, 1 mM Na 3 VO 4 , 0.5% NP-40). Lysates were centrifuged at 14,000 rpm for 10 min and the supernatant was incubated for 3 h at 4uC with anti-flag or anti-myc antibody cross-linked to protein G sepharose beads. Beads were washed 3 times with immunoprecipitation buffer, bound proteins were eluted by gel sample buffer and separated on SDS-PAGE.

Confocal microscopy
Cells were grown on the SecureslipTM cell culture cover slip (Gracebiolab). Cells were then incubated with 250 nM of MitoTracker Red 580 in HBSS at 37uC for 30 min and fixed with 4% paraformaldehyde for 10 min at room temperature and permeabilization was done with 0.1% Triton X-100 in HBSS for 15 min at room temperature. The cells were treated with blocking solution (3% BSA, 0.2% Tween 20 and 0.2% gelatin in PBS) for 1 h at room temperature and incubated with monoclonal anti-myc antibody, diluted 1:250 in HBSS containing 1% BSA and 1% sucrose for 1 h at 37uC. After three washes with PBS, the cells were incubated with Alexa Fluor 488 conjugated goat anti mouse IgG diluted 1:50 for 1 h at 37uC. After three washes, the cells were analyzed under Zeiss LSM 510 META confocal microscope.

Preparation of mitochondrial fractions
Mitochondrial fractions of the cells were prepared as described previously [48]. Cells were suspended in 200 mL of 10 mM Tris-Cl, pH 7.5, supplemented with protease inhibitor cocktail and 10 mM NEM and passed through a 26.5-gauge needle 20 times and 40 mL of ice-cold 1.5 M sucrose solution was added and centrifuged at 600 x g for 10 min, the supernatants were centrifuged at 14,000 x g for 10 min and mixed with the same volume of gel sample buffer. The pellets containing mitochondrial fraction were washed with 50 mL of 10 mM Tris-Cl, pH 7.5 twice and lysed in 30 mL of gel sample buffer. Same volumes of the supernatant and gel sample buffer mixture and mitochondrial fraction in gel sample buffer, were resolved in 10% SDS-PAGE.

Measurement of ROS
Cellular ROS were measured using CM-H 2 DCFDA (Invitrogen) following manufacturer's instructions. Briefly, trypsinized cells were washed and resuspended in 0.5 mL of HBSS; 3 mL of 0.5 mM CM-H 2 DCFDA was added and incubated at 37uC for 15 min. After brief wash with HBSS, the cells were analyzed with FACS Calibur flow cytometer (BD Biosciences).

Real-time cell analysis
Five thousand HeLa cells were plated in a 96 well E-plate for the real-time cell analyzer, xCELLigence (Roche Applied Science). For H 2 O 2 treatment, cells were treated with H 2 O 2 for 1 h before plating in the E-plate. After 24 h, the cells were treated with rotenone or CCCP and cell growth was monitored by measuring electrical impedance every 30 min.

Generation of hNT-UCH-L1 expressing transgenic mice
We generated transgenes by cloning the hNT-UCH-L1-myc DNA under the control of CAG promoter in pCAGEN vector (Addgene). The plasmid was linearized by digestion with ScaI and BamHI and microinjected into the pronuclei of newly fertilized C57BL/6N mouse eggs (Macrogen). Germline transmission of NT-UCH-L1 was obtained in five independent NT-Tg lines and the levels of transgenic mRNA were assessed by quantitative RT-PCR. cDNAs from mouse whole brain were prepared using RNA extraction kit (Qiagen) and SuperScript III (Life Technologies) reverse transcriptase. For quantitative RT-PCR, specific primers for the transgene (forward: 59-TGCTGAACAAAGTGCTGTCC-39; reverse: 59-AGCCCAGAGACTCCTCTTCC-39) were used. The offsprings were screened with specific primers (

MPTP treatment and analysis of hNT-UCH-L1 transgenic mice
All experiments were conducted with eight-week-old male C57BL/6N mice (23-25 g) maintained in a room at 20-22uC on 12 h light/dark cycle with food and water available ad libitum. As described previously [49][50][51], mice were administered four i.p. injections of MPTP (20 mg/kg, free base; Sigma-Aldrich) dissolved in PBS at 2 h intervals, and were transcardially perfused and fixed with 4% paraformaldehyde dissolved in 0.1 M phosphate buffer (PB). Brain tissues were cut into 30-mm-thick coronal sections and processed for immunostaining, as described [49,50]. In brief, the brain sections were rinsed and incubated overnight at room temperature with primary antibodies, anti-TH, at 1:2000 dilution; Pel-Freez Biologicals) for DA neurons and anti-OX-42 (CD11b, at 1:400 dilution; Serotec) for microglia. On the following day, the sections were rinsed, incubated with the appropriate biotinylated secondary antibody and processed with an avidin-biotin complex kit (Vectastain ABC kit; Vector Laboratories). Bound antibody was visualized by treatment with diaminobenzidine-HCl (Sigma-Aldrich) and hydrogen peroxide (Sigma-Aldrich). The diaminobenzidine reaction was terminated by rinsing the tissues in 0.1 M PB. Labeled tissue sections were mounted on gelatin-coated slides and analyzed under a bright-field microscope (Nikon).

Stereological cell counts
As described previously [49,50], the total number of THpositive neurons was counted in the various animal groups at 7 days post injection (MPTP or PBS) using the optical fractionators method performed on an Olympus Computer Assisted Stereological Toolbox system version 2.1.4 (Olympus Denmark). The counting was performed using a 100X oil objective. The total number of neurons was estimated using the optical fractionator equation [52].

Densitometric analysis
As described previously [49,50], the OD of TH-positive fiber in STR was examined at 35 original magnification using the IMAGE PRO PLUS system (Version 4.0; Media Cybernetics) on a computer attached to a light microscope (Zeiss Axioskop) interfaced with a CCD video camera (Kodak Mega Plus model 1.4 I; Kodak). To control variations in background illumination, the average of background density readings from the corpus callosum was subtracted from that of density readings of the STR for each section. For each animal, the average of all sections was calculated separately before data were statistically processed.

Discovery of N-terminal truncated UCH-L1 (NT-UCH-L1)
We screened for variants of UCH-L1, several of which are shown in Fig. S1A in human lung cancer cell line, NCI-H157, and neuroblastoma cell line, SH-SY5Y. We detected four UCH-L1 spots by immunoblotted 2D-PAGE (Fig. 1A, left top panel). The spots from the corresponding silver stained gels (Fig. 1A, left middle panel) were characterized using mass spectrometry (MS). We found that spot 1 was missing the peptide 1 MQLKPMEIN-PEMLNK 15 (1815.9144 Da) of UCH-L1, but this was replaced by 12 MLNK 15 (505.2905 Da) (Fig. S1B). The peptide, missing in spot 1, was present in spot 3, and both spots had similar molecular weights about 25 kDa, suggesting that spot 3 was full length UCH-L1 while spot 1 is a truncated UCH-L1 missing an N-terminal peptide (NT-UCH-L1). To confirm this, we raised polyclonal antibodies (anti-N-terminal peptide) against the peptide, 1 MQLKPMEINPE 11 in rabbit, and examined their reactivity with the four spots. The antibodies did not react with spot 1 which lacks the peptide, but reacted with spots 2, 3, and 4 ( Fig. 1A left bottom panel). NT-UCH-L1 was also detected in human neuroblastoma cell line, SH-SY5Y (Fig. 1A, right panels) and mouse whole brain tissue (Fig. 1B). The deleted peptide of NT-UCH-L1, corresponds to exon 1 of the UCH-L1 gene (PARK5) (Fig. 1C). The exon 1 deleted form is an mRNA registered in NCBI nucleotide database under the name, protein gene product 9.5 (PGP9.5, accession number X04741). Given the structure of its
In order to understand the structural basis of these differences, we employed mass spectrometry with hydrogen/deuterium exchange (HDX) in which residues exposed to surface more readily exchange deuterium allowing their mass increases to be detected. HDX occurred slowly in UCH-L1, taking up to 16 h to complete, while in NT-UCH-L1, this exchange was completed in just about 2 min ( Fig. 2A and B). This indicates that NT-UCH-L1 has a more flexible structure than UCH-L1. The flexible regions were identified by analyzing HDX in pepsin-digested peptides (Fig. 2C). More deuterium exchange occurred in the N-and Ctermini of NT-UCH-L1 than of UCH-L1, and the peptide   NT-UCH-L1 is monoubiquitinated at either Lys15 or Lys157 UCH-L1 recruited free ubiquitin resulting in high cellular free ubiquitin levels [53]. We therefore examined possible differences between NT-UCH-L1 and UCH-L1, in binding free ubiquitin. NCI-H157 cells transiently expressing UCH-L1-myc or NT-UCH-L1-myc were immunoprecipitated and the resulting precipitates were analyzed (Fig. 3A). We detected several bands including UCH-L1-myc and NT-UCH-L1-myc in silver stained gel. The unknown bands were identified by peptide sequencing with MS/MS. One band of about 7 kDa, co-precipitating with UCH-L1, was identified as previously shown, as ubiquitin noncovalently bound to UCH-L1 (Fig. 3A arrow, 3B right panel arrow), in agreement with previous studies [53]. Two bands coprecipitating with NT-UCH-L1 (Fig. 3A arrow heads) were identified as mixtures of NT-UCH-L1 and ubiquitin. We found that the molecular weight difference between NT-UCH-L1-myc and the two bands was close to the molecular weight of one ubiquitin. We performed Western blot analysis with anti-myc ( Fig. 3B left panel) and anti-ubiquitin (Fig. 3B right panel) antibodies and confirmed the upper band as monoubiquitinated NT-UCH-L1 (Ub-NT-UCH-L1). When NCI-H157 cells were analyzed in 2D-PAGE and blotted with anti-UCH-L1 or anti-Nterminal peptide antibody, we detected a shifted NT-UCH-L1 spot possibly endogenous ubiquitinated NT-UCH-L1 (Fig. S2  arrow).
We identified the sites of monoubiquitinated lysine residue(s) in NT-UCH-L1, employing both peptide sequencing with MS/MS and mutation studies. We analyzed peptides of immunoprecipitated and trypsinized NT-UCH-L1 by MS/MS and detected Lys15 having a mass increase 114 Da which is equivalent to Cterminal Gly-Gly of attached ubiquitin. This indicates that the candidate site of ubiquitination is Lys15 (Fig. 3C). We constructed two mutants of NT-UCH-L1, K157R (a previously known monoubiquitination site of UCH-L1) [24] and K15/157R, and examined monoubiquitination of these mutants (Fig. 3D). Monoubiquitination of NT-UCH-L1 was reduced in K157R NT-UCH-L1 and was mostly absent in K15/157R NT-UCH-L1 mutants. These results suggest that there are two populations of Ub-NT-UCH-L1, one monoubiquitinated at K15 and the other at K157, not both.

Ub-NT-UCH-L1 localizes in cytosol while unubiquitinated-NT-UCH-L1 localizes in mitochondria
The focus of PD research has been on mitochondrial functions and its respiratory activities. The presence of PD-related gene products in autopsied PD brain tissues and in mitochondria, underscores their effect on mitochondrial functions [26][27][28][29][30][31][32][33]. We therefore compared the subcellular localizations of UCH-L1 and NT-UCH-L1. We used HeLa cells because the expression of endogenous UCH-L1 in these cells was blocked by methylation of the UCH-L1 gene [54]. In confocal microscopy, the majority of UCH-L1-myc localizedin the cytosol, while NT-UCH-L1-myc localized in both cytosol and mitochondria (Fig. 4A). Moreover, ubiquitination-free K15/157R NT-UCH-L1 localized primarily in the mitochondria. This suggests that monoubiquitination governs subcellular localization of NT-UCH-L1. These localizations were corroborated by Western blot studies of crude fractionated cytoplasm and mitochondria. Whereas full length UCH-L1 and the ubiquitinated species of NT-UCH-L1 were found predominantly in the cytosol, the non-ubiquitinated NT-UCH-L1 and K15/157R NT-UCH-L1 were found in both the cytosol and mitochondrial fractions, but more predominantly in mitochondria ( Fig. 4B and 4C). It is possible that the binding of K15/157R NT-UCH-L1 with mitochondrial outer membrane might be weak and results in its partial cytosolic localization in mitochondrial fractionation assay. Although some part of UCH-L1 was monoubiquitinated as shown in Fig. 3B, most of UCH-L1 and its monoubiquitination free form localized in cytosol (Fig.  S3A). Monoubiquitination on UCH-L1 did not seem to determine its subcellular localization.
To further dissect the location of NT-UCH-L1 in mitochondria, we incubated the mitochondrial fractions with varying concentrations of proteinase K and performed Western blot analysis with antibodies specific for Tom20, cytochrome C, and COX-4, markers of the outer membrane, intermembrane space, and inner membrane of mitochondria, respectively (Fig. 4D). The proteolytic degradation pattern of NT-UCH-L1 was similar to that of Tom20 suggesting that NT-UCH-L1 localized in the outer membrane of mitochondria.
Since Parkin, another PD related protein, is recruited to mitochondria and degraded via mitotophagy in response to mitochondrial damage [55], we investigated whether mitochondrial localization of NT-UCH-L1 is related to m-chlorophenylhydrazone (CCCP) induced mitochondrial damage. We did not detect any increase of NT-UCH-L1 in mitochondria after CCCP treatment (Fig. 4E). Rather, NT-UCH-L1 in mitochondrial fraction decreased. When HeLa cells were treated with CCCP in combination with MG132, NH 4 Cl, and 3MA, inhibitors of proteasomal, lysosomal, and autophagic degradation, respectively, both NT-UCH-L1 and its K15/157R mutant were protected from degradation by MG132 treatment suggesting proteasomal degradation of NT-UCH-L1 (Fig. 4F). That means NT-UCH-L1's degradation pathway is clearly distinguishable from parkin involved mitophagy.
To understand how NT-UCH-L1 is degraded in normal condition, we compared NT-UCH-L1 levels in cells treated with MG132, NH 4 Cl, and 3MA ( Fig. 5C and 5D). We found that only MG132 induced the accumulation of NT-UCH-L1 and its K15/ 157R mutant in HeLa cells suggesting that in steady state as well as CCCP treated condition, NT-UCH-L1 is readily degraded by ubiquitin-proteasome system.

NT-UCH-L1 is aggregation prone
UCH-L1 is a major component of the protein aggregates called Lewy bodies found in the brains of PD patients [2]. We examined the aggregation characteristics of NT-UCH-L1 accumulating in cells. Cells were treated with MG132 and separated into detergent soluble (S) and insoluble (I) fractions ( Fig. 5E and 5F). NT-UCH-L1 and its K15/157R mutant significantly accumulated in the insoluble fraction, while negligible amounts of UCH-L1 were detected in the insoluble fraction after MG132 treatment. Also, only ubiquitination free NT-UCH-L1 was present in the insoluble fraction after MG132 treatment ( Fig. 5E and 5F, middle panel). When we examined the insoluble fraction in a dopaminergic neuronal progenitor cell line SN4741, we found more than 20% of the transiently expressed NT-UCH-L1 and K15/157R NT-UCH-L1 in insoluble fractions, even without MG132 treatment, while most of the UCH-L1 was found in the soluble fraction (Fig. 5G). With increasing expression of NT-UCH-L1, its insoluble fraction increased (Fig. 5H, top panel). When 0.5 and 1.5 mg of NT-UCH-L1-myc expressing DNA were used for transient transfection, most of NT-UCH-L1 was in the soluble fraction. However, a part of NT-UCH-L1 expressed from 3 mg of DNA was in the insoluble fraction. This suggests that there is a threshold for the level of soluble NT-UCH-L1, and above this threshold, NT-UCH-L1 becomes insoluble in cells. We confirmed the tendency of NT-UCH-L1 to aggregate and become insoluble in cells, by examining the cells under confocal microscopy. NT-UCH-L1, not UCH-L1, was found aggregated in response to MG132 treatment (Fig. 5I). These results demonstrate that NT-UCH-L1 is more aggregationprone than UCH-L1.

Cys90 and Cys132 of NT-UCH-L1 form disulfide bonds
To investigate the nature of the insoluble aggregates of NT-UCH-L1, the soluble and insoluble fractions were analyzed in reducing ((+) 2-ME) and non-reducing ((2) 2-ME) conditions (Fig. 5H). Some of the insoluble NT-UCH-L1 appeared to be crosslinked through disulfide bonds, as it was detected at a high molecular weight complex on non-reducing gel (Fig. 5H, lower panel, S-S NT), but not UCH-L1 (Fig. S3C, lower panel). To confirm that NT-UCH-L1 is able to form disulfide crosslinking by themselves, recombinant UCH-L1 and NT-UCH-L1 were separated in gel filtration FPLC based on molecular weight. Fig. 6A shows that UCH-L1 mostly eluted as monomers (of about 29 kDa), while NT-UCH-L1 eluted as multimers larger than 66 kDa (Fig. 6A, upper panel). These high molecular weight NT-UCH-L1s dissociate into monomers with similar molecular weight with UCH-L1 under reducing condition ((+) 2-ME) (Fig. 6A, lower  panel). This indicates that insoluble aggregates of NT-UCH-L1 contain intermolecular disulfide crosslinking.
The cysteine residues involved in the formation of insoluble aggregates, were identified by mutational and proteomic approaches. Under the non-reducing condition, NT-UCH-L1 having 6 Cys residues including Cys90 at its active site, formed multimers (Fig. 6B, lane 1). We performed mutagenesis of each 6 Cys to serine and examined their disulfide crosslinking formation. The C90S NT-UCH-L1 contained fewer multimers and more monomers of size around 25 kDa than other mutants (Fig. 6B,  arrow). Disulfide crosslinking sites in NT-UCH-L1 multimer were also identified by peptide sequencing with nonoUPLC ESI-q-TOF MS/MS combined with the searching algorithm of disulfide site, DBond [54]. We detected many disulfide containing peptides (Table 1). Of the two surface cysteine residues, Cys132 and Cys152 [56], Cys132 formed disulfide bonds with all 6 Cys including itself. Based on these results, we generated C90S, C132S, and C90/132S mutants of NT-UCH-L1 and investigated whether they form less disulfide crosslinkings in response to oxidative stress. Cells were treated with H 2 O 2 and the proteins were separated under both reducing and non-reducing conditions (Fig. 6C). UCH-L1 did not form disulfide bonded multimers, but NT-UCH-L1 did, in cells treated with 0.5 and 1 mM H 2 O 2 . C90S NT-UCH-L1 formed fewer multimers than NT-UCH-L1 or C132S NT-UCH-L1. This indicates that NT-UCH-L1 forms disulfide bonds mainly through Cys90 in response to oxidative stress.

NT-UCH-L1 reduces cellular ROS levels
The mitochondrial localization of NT-UCH-L1 prompted us to speculate on its possible function in mitochondria. Since mitochondria are part of the major intracellular ROS generation machinery and NT-UCH-L1 is sensitive to oxidative stress, we examined the possible role of NT-UCH-L1 in controlling ROS levels. We assessed cellular ROS levels by measuring fluorescence generated by ROS after loading the cells with CM-H 2 DCFDA. Expression of NT-UCH-L1 and K15/157R NT-UCH-L1 resulted in decreased cellular ROS levels by approximately 2-fold, compared to the expression of mock and UCH-L1 (Fig. 6D). However, C90/132S NT-UCH-L1 did not lower ROS levels. These results suggest a relationship between disulfide formation in NT-UCH-L1 and cellular ROS levels. NT-UCH-L1-myc were examined. (A, B) Cells were treated with 10 mg/mL cycloheximide for the indicated times and immunoblotted using anti-myc antibody (A) and the bands were quantified (B). (C, D) Cells were treated with DMSO, 10 mM MG132, 10 mM NH 4 Cl or 10 mM 3MA 16 h and immunoblotted using anti-myc antibody (C) and the bands were quantified (D). (E, F) Cells were treated with DMSO or 10 mM MG132 for 9 h and divided into soluble and insoluble fractions. Samples were immunoblotted using anti-myc antibody (E) and the bands were quantified (F). Quantitative analysis was done and the mean 6 s.d. of three independent experiments is shown. (G) SN4741 cells were transiently transfected with pcDNA3.1 UCH-L1-myc, pcDNA3.1 NT-UCH-L1-myc, and pcDNA3.1 K15/157R NT-UCH-L1-myc and fractionated into soluble and insoluble fractions. Fractions were immunoblotted using anti-myc and anti-a-tubulin. Due to the higher expression level of UCH-L1 than NT-UCH-L1s, immunoblot was done in separate membrane and the border lines were drawn. (H) SN4741 cells were transiently transfected with various amounts of pcDNA3.1 NT-UCH-L1-myc expressing plasmid. Cells were divided into soluble and insoluble fractions and analyzed in non-reducing ((-) 2-ME) and reducing ((+) 2-ME) gels. Proteins were immunoblotted using anti-myc and anti-a-tubulin antibodies. (I) HeLa stable cells expressing UCH-L1-myc and NT-UCH-L1myc were treated with DMSO or 10 mM MG132 for 16 h. Myc tagged proteins were probed with anti-myc and Alexa Flour 488 secondary antibody (green), mitochondria were stained with Mitotracker (red) and nucleus were stained with DAPI (blue). Cells were examined under confocal microscope. doi:10.1371/journal.pone.0099654.g005

NT-UCH-L1 protects cells from H 2 O 2 , rotenone, and CCCP induced damage
Oxidative damage such as the one produced with H 2 O 2 induces cell death which is related to the onset of PD [57,58]. We tested if NT-UCH-L1 functions in H 2 O 2 induced cell death. Cells were treated with 2 mM H 2 O 2 for 1 h and monitored cell growth using a real-time cell analyzer (Fig. 7A). We found that cells expressing NT-UCH-L1 but not UCH-L1 were resistant to H 2 O 2 -induced cell death. Accumulating evidences suggests that mitochondrial dysfunction is another contributor to the pathogenesis of PD. Reduced activity of the mitochondrial electron transfer chain in PD has been reported [59,60]. Rotenone and CCCP are known to induce mitochondrial damage and cell death [31,61], we therefore examined whether NT-UCH-L1 plays a role in rotenone and CCCP induced cell death. Cells were treated with rotenone and CCCP and monitored cell growth using a real-time cell analyzer ( Fig. 7B and 7C). We found that cells expressing NT-UCH-L1, but not UCH-L1 were resistant to rotenone-and CCCP-induced cell death.

Generation of human NT-UCH-L1-myc expressing transgenic mice
Finally, we addressed the question of whether the protective effects of NT-UCH-L1 against toxins seen in cell culture (Fig. 7) might also be manifested in vivo. We generated transgenic mice expressing human NT-UCH-L1 with myc tag (NT-Tg). We used a CAG promoter to drive expression of NT-UCH-L1 in transgenic mice (Fig. 8A). Germline transmission of NT-UCH-L1 was obtained in five independent NT-Tg lines and the levels of transgenic mRNA were assessed by quantitative RT-PCR (Fig. 8B). We chose the line #1 in which expression of NT-UCH-L1-myc mRNA was the highest for further analysis. We detected NT-UCH-L1-myc protein in mouse whole brain by immunoprecipitation and Western blot analysis (Fig. 8C, lane 2) co-migrating with the NT-UCH-L1 transiently expressed in HEK293 cells (Fig. 8C, lane 3). We confirmed that by silver staining and MS/MS to identify the immuprecipitated NT-UCH-L1 (Fig. S4A). We detected a faint silver stained band and identified a peptide of UCH-L1 only in the immunoprecipitant of NT-Tg but not in non-Tg sample.
NT-UCH-L1 protects nigrostriatal dopaminergic neurons from the MPTP neurotoxicity in transgenic mice expressing human NT-UCH-L1 We then tested the effect of NT-UCH-L1 on the survival of DA neurons in the MPTP mouse model of PD. The brains of PBS or MPTP injected mice were removed and sectioned. The sections were immunostained for tyrosine hydroxylase (TH) to specifically detect DA neurons. As we reported previously [49][50][51], immunohistochemical analysis, stereological counts, and densitometric analyses revealed a 60% loss of TH-positive cell bodies in the substantia nigra pars compacta (SNpc) (Fig. 8D-c and E), and a 31% loss in TH-positive fibers in the striatum (STR) (Fig. 8D-d and F) in MPTP-injected non-Tg mice, compared with those in PBS control of non-Tg mice ( Fig. 8D-a, b, E, and F). In contrast, the number of TH-positive cell bodies in the SNpc (Fig. 8D-g and E) and the density of TH-positive fibers in the STR (Fig. 8D-h and F) were significantly higher in the MPTP-treated NT-Tg mice, compared with MPTP-treated non-Tg mice (Fig. 8D-c, d, E, and F). When we stained cells with Cresyl violet stain and Cresyl violet/TH double staining, we also observed higher survival rates of SNpc cells in MPTP treated NT-Tg mice than in non-Tg mice ( Fig. S4B and S4C). These findings suggest that NT-UCH-L1 promotes the survival of nigrostriatal DA neruons in the MPTP model of PD.

Disulfide bonds
Peptides with disulfide bonds

Discussion
In this study, we found that a variant of UCH-L1 lacking 11 Nterminal amino acids, that we designated NT-UCH-L1, exists in lung cancer and neuronal cells, and in brain tissue. NT-UCH-L1 differs from UCH-L1 in its tertiary structure, in its lack of deubiquitinating activity, lack of ability to increase cell migration, in its greater tendency to be monoubiquitinated and readily aggregated.
Protein isoforms from one gene are produced by alternative splicing, alternative promoters, or alternative translation initiation sites. The 11 N-terminal amino acids missing in NT-UCH-L1, corresponds to exon 1 of the UCH-L1 gene (PARK5). The exon 1deleted mRNA has been reported to be expressed originally in a protein given a different name, PGP9.5 (NCBI accession number X04741) [62]. Thus, NT-UCH-L1 appears to be formed by translation from an alternatively spliced mRNA. Compared to alternative splicing and alternative promoters, alternative translation initiation was less studied. Recently, alternative translation start sites within a single transcript have been paid more attention [63]. As NT-UCH-L1 has a starting codon and an additional downstream in-frame AUG codon, alternative translation initiation from the 12 th methionine may also possible. Several examples including human insulin-degrading enzyme [64], human neuropeptide Y [65], AtLIG1 gene in Arabidopsis thaliana [66], rat ornithine decarboxylase-enzyme [67], are reported as having alternative start codons and the isoforms showed distinguished subcellular localizations [68].
To compare the amounts of UCH-L1 and NT-UCH-L1 in mouse brain tissue and cell lines, we separated them in 2D-PAGE and probed with two different kinds of antibodies; one for both UCH-L1 and NT-UCH-L1 and the other only for UCH-L1. Significant amounts of NT-UCH-L1 exist in mouse brain tissue, neuronal cells, and lung cancer cells (Fig. 1A and 1B).
We further compared UCH-L1 and NT-UCH-L1 with regard to their structures, stability, modifications, localizations, and biological functions. NT-UCH-L1 did not exhibit the ubiquitin hydrolase activity of UCH-L1 ( Fig. 1D and S1C). In HDX data ( Figure 2C), the peptide containing active site Cys90 of NT-UCH-L1 was less deuterium exchanged than UCH-L1. It might explain different enzyme activities of UCH-L1 and NT-UCH-L1. In addition, UCH-L3 sharing considerable homology with UCH-L1 (52% amino acid identity) was postulated that it contacts basic residues located on ubiquitin with its acidic residues (Glu10, Glu14, Asp33, Glu219) [69]. All of the acidic residues are conserved in UCH-L1 and among them, Glu10 and Glu14 are missing in NT-UCH-L1. This suggests that N-terminal 11-peptide is required for catalytic activity in addition to active site. Employing HDX mass spectrometry, we found that NT-UCH-L1 has a more flexible structure than UCH-L1. During the preparation of recombinant NT-UCH-L1, it forms visible protein aggregates more easily than UCH-L1. This indicates that NT-UCH-L1 has unfolded structure similar to another PD causing gene product a-synuclein [70]. Alpha-synuclein is typically unfolded and can form various kinds of oligomeric structures such as pore forming structure, fibrils, and amorphous aggregates [19,20]. As shown in Fig. 5H lower panel, not all of the NT-UCH-L1 in the insoluble fraction forms disulfide crosslinkings and there still exists disulfide linkage free NT-UCH-L1 (arrow head in the lane 8). It is possible that NT-UCH-L1 also has diverse kinds of muntimer forms and studying oligomeric structures of NT-UCH-L1 in detail would be interesting.
Protein monoubiquitination plays roles in DNA repair, histone regulation, gene expression, and receptor endocytosis [25]. In this study, we showed another function of monoubiquitination: it also changes subcellular localization and solubility of a protein. We showed that Ub-NT-UCH-L1 localized in the cytosol and ubiquitination free NT-UCH-L1 mainly localized in the mitochondria. In addition, we showed monoubiquitination free form of NT-UCH-L1 but not Ub-NT-UCH-L1 became insoluble when accumulated in cells. As monoubiquitination free form of NT-UCH-L1 localized in mitochondria and it became more insoluble in MG132 treated cells, it is possible that NT-UCH-L1 forms insoluble particle in mitochondria.
NT-UCH-L1 was easily oxidized by H 2 O 2 treatments, forming inter-molecular disulfide bonds and cells expressing NT-UCH-L1 resist H 2 O 2 induced cell death. This is similar to the action of DJ-1 under oxidizing environment. Cysteines in DJ-1 are easily oxidized in response to ROS and play antioxidant roles [71][72][73]. Further studies of the relationships among disulfide bond formations, lowering cellular ROS levels, and protection of cells from oxidative stress by NT-UCH-L1, would be informative.
Based on our results, we suggest a molecular mechanism involving ubiqitination for the role of NT-UCH-L1 in protecting cells against stresses. This suggested mechanism is summarized as follows (Fig. 9). NT-UCH-L1 present in the mitochondria is in a state of equilibrium with Ub-NT-UCH-L1 present in the cytosol. When the amount of NT-UCH-L1 is increased above saturation level or when the cells are stressed, monoubiquitination free NT-UCH-L1 aggregates via intermolecular disulfide bonds mostly involving Cys90. Once NT-UCH-L1 becomes insoluble, it does not seem to go back to soluble fraction. The aggregates thus formed in the mitochondria are cleared by proteasome. NT-UCH-L1 is degraded also by proteasome in steady state.
This study suggests that NT-UCH-L1 has the potential for preventing neurotoxicity in PD. However, approaches to increasing NT-UCH-L1 levels in neurons to protect against mitochondrial or other damage might be hampered by thresholds for NT-UCH-L1 solubility, because NT-UCH-L1 forms insoluble protein aggregates once the concentration threshold is exceeded. This suggests that a fine control of the synthesis and degradation of NT-UCH-L1, is necessary for the maintenance of optimum levels of NT-UCH-L1. Currently, we do not know which factor(s) regulates the expression of NT-UCH-L1 and which E3 ubiquitin ligase promotes its monoubiquitination and degradation. Further studies are needed to shed light on qualitative and quantitative aspects of NT-UCH-L1's function in PD.  Figure S3 (A) HeLa cells transiently expressing UCH-L1-myc, K15/157R UCH-L1-myc, and NT-UCH-L1-myc were stained with anti-myc and Alexa Fluor 488 secondary antibody (green). Mitochondria and nucleus were stained with Mitotracker (red) and DAPI (blue), respectively. Cells were visualized by confocal microscopy. Lower pannel is enlarged figure of the dotted rectangle region of each upper pannel. (B) HeLa cells transiently expressing UCH-L1-myc, K15/157R UCH-L1-myc, and NT-UCH-L1-myc were treated with 10 mg/mL cycloheximide for the indicated times and immunoblotted using anti-myc antibody. (c) SN4741 cells were transiently transfected with pcDNA3.1 UCH-L1-myc or NT-UCH-L1-myc expressing plasmid. Cells were divided into soluble and insoluble fractions and analyzed in nonreducing ((-) 2-ME) and reducing ((+) 2-ME) gels. Proteins were immunoblotted using anti-myc antibody. (TIF) Figure S4 (A) Immunoprecipitation analysis of hNT-UCH-L1myc in control and NT-Tg mouse brain using anti-myc antibody. Immunoprecipitants were analyzed by silver staining. The band only detected in NT-Tg sample (arrow head) was cut and identified by peptide finger printing and MS spectrometry. We observed one peptide peak with M.W. 742.9290 which matches with the UCH-L1 sequence, 66 QIEELKGQEVSPK 78 . (B, C) Mice in each group (non-Tg control (a, b), NT-Tg (c, d)) were sacrificed 7 d after the last MPTP injection (b, d) or PBS as controls (a, c). Brain tissues were processed for Nissl staining (blue) (B) and Nissl (blue) and TH (brown) double staining (C). Dotted lines indicate substantia nigra pars compacta. Insets, pictures with higher magnifications. Scale bars, 100 mm. (TIF)