Mutations in the DJ-1 gene cause early-onset autosomal recessive Parkinson's disease (PD), although the role of DJ-1 in the degeneration of dopaminergic neurons is unresolved. Here we show that the major interacting-proteins with DJ-1 in dopaminergic neuronal cells are the nuclear proteins p54nrb and pyrimidine tract-binding protein-associated splicing factor (PSF), two multifunctional regulators of transcription and RNA metabolism. PD-associated DJ-1 mutants exhibit decreased nuclear distribution and increased mitochondrial localization, resulting in diminished co-localization with co-activator p54nrb and repressor PSF. Unlike pathogenic DJ-1 mutants, wild-type DJ-1 acts to inhibit the transcriptional silencing activity of the PSF. In addition, the transcriptional silencer PSF induces neuronal apoptosis, which can be reversed by wild-type DJ-1 but to a lesser extent by PD-associated DJ-1 mutants. DJ-1-specific small interfering RNA sensitizes cells to PSF-induced apoptosis. Both DJ-1 and p54nrb block oxidative stress and mutant α-synuclein-induced cell death. Thus, DJ-1 is a neuroprotective transcriptional co-activator that may act in concert with p54nrb and PSF to regulate the expression of a neuroprotective genetic program. Mutations that impair the transcriptional co-activator function of DJ-1 render dopaminergic neurons vulnerable to apoptosis and may contribute to the pathogenesis of PD.
The highly conserved DJ-1 gene has been identified as one of the most frequently mutated genes in familial Parkinson's disease (PD) (1–3). Homozygous deletion or point mutations in DJ-1 cause early-onset autosomal recessive Parkinsonism (1,2). DJ-1 was originally cloned as an oncogene that cooperatively transforms cells together with H-ras (4). In addition to a potential role in oncogenesis, DJ-1 acts as a regulatory subunit of a 400 kDa RNA-binding protein complex (5). DJ-1 also modulates androgen receptor-dependent transcriptional activity by binding and blocking transcriptional repressors such as PIASxa and DJBP (6,7).
The responsiveness of DJ-1 to oxidative stress has provided a potential functional link to the pathogenesis of PD. When exposed to oxidative insults including paraquat and H2O2, DJ-1 shifts its isoelectric point from 6.2 to a more acidic 5.8 (8,9). A recent report further indicates that DJ-1 is an anti-oxidant capable of self-oxidation (9). Consistent with these observations, gene deletion or down regulation of DJ-1 by small interfering RNA (siRNA) sensitizes cells to oxidative and ER stress (10,11). In addition, DJ-1 serves as a redox sensitive molecular chaperon capable of preventing the aggregation of α-synuclein (12). However, the role of these various functions of DJ-1 in the pathogenesis of PD in vivo remains to be confirmed.
To identify the molecular functions of DJ-1-specific to dopaminergic cells, we adopted an unbiased approach to identify the major interacting proteins using affinity purification and mass spectrometry, followed by protein interaction assays and co-localization studies. Our results indicate that DJ-1 forms a nuclear complex with p54nrb and pyrimidine tract-binding protein-associated splicing factor (PSF). p54nrb and PSF are multifunctional nuclear proteins (13). PSF was originally identified as a protein interacting with polyprimidine tract (14), an intronic region important for splicing. In addition, PSF is a part of the spliceosome C complex (15) and is required for in vitro splicing of pre-mRNA (14). Both p54nrb and PSF contain homologous RNA recognition motifs and form heterodimeric complex capable of binding RNA (13). p54nrb and PSF heterodimers also bind DNA and regulate gene transcription (13,16,17). We have found that DJ-1 cooperates with p54nrb to activate transcription by inhibiting the transcriptional silencing activity of PSF. Moreover, DJ-1 and p54nrb prevent apoptosis induced by PSF, mutant α-synuclein or oxidative stress. In addition, cells pre-treated with siRNA specific for DJ-1 are more sensitive to PSF-induced apoptosis. PD-associated DJ-1 mutants show decreased nuclear localization and significantly reduced transcriptional activation and protection against apoptosis. Therefore, these findings provide a novel potential mechanism for the pathogenic effects of a gene associated with recessive PD.
DJ-1 forms complex with p54nrb and PSF
To explore the function of DJ-1 in dopaminergic cells, we attempted to identify DJ-1 interacting proteins. Cell lysates from SH-SY5Y cells expressing an empty plasmid vector (control) or Myc-His-tagged human wild-type DJ-1 were used in affinity purification with Ni-NTA agarose beads. After extensive washing, bound proteins were eluted and resolved by SDS–PAGE. Three major bands specifically appeared in DJ-1-expressing cells. Analysis by mass spectrometry identified these proteins as DJ-1, nuclear RNA binding protein p54nrb and pyrimidine tract-binding protein-associated splicing factor (Supplementary Material, Table S1, Fig. 1A). Absence of these three proteins in the eluted proteins from the control sample was confirmed by parallel mass spectrometry analysis, suggesting that p54nrb and PSF bound specifically to tagged DJ-1. To eliminate the possibility of false positive interactions between Ni-NTA beads or the Myc-His tag on DJ-1 and p54nrb or PSF, we examined the interaction between endogenous DJ-1 and p54nrb or PSF. An antibody specifically recognizing the C-terminus of DJ-1 co-immunoprecipitated endogenous p54nrb and PSF in native SH-SY5Y cells (Fig. 1B). In addition, antibodies recognizing endogenous PSF or p54nrb specifically co-precipitated endogenous DJ-1 (data not shown). These results indicate that DJ-1 forms protein complexes with p54nrb and PSF and that these are the major DJ-1 interacting proteins in dopaminergic SH-SY5Y cells.
p54nrb and PSF have been frequently shown to co-exist in multifunctional protein complexes that regulate nuclear functions including transcription, RNA splicing and editing (13). On the other hand, DJ-1 appears to regulate anti-oxidative stress response, protein aggregation and transcription in both cytoplasm and nucleus. Double labeling of endogenous DJ-1 and p54nrb or PSF confirmed the nuclear co-localization of the three proteins (Supplementary Material, Fig. S1A). To examine whether mutations associated with PD affect the distribution of DJ-1 and its co-localization with p54nrb or PSF, we labeled cells stably expressing Myc-His-tagged wild-type DJ-1 or mutant DJ-1 with anti-His antibody and anti-p54nrb (Fig. 2A) or anti-PSF antibodies (Supplementary Material, Fig. S1B). Like wild-type DJ-1, a non-pathogenic heterozygous polymorphism of DJ-1, R98Q (18), was localized diffusely in cytoplasm and nucleus. However, pathogenic DJ-1 mutants L166P, M26I and D149A exhibited reduced nuclear distribution and decreased co-localization with p54nrb or PSF (Fig. 2A and Supplementary Material, Fig. S1B, middle and right panels). In agreement with a prior report showing increased mitochondrial localization of the L166P DJ-1 mutant (1), we have observed similar distribution patterns for the M26I and D149A mutants using multiple mitochondrial markers including cytochrome c, MnSOD and HSP60 (Fig. 2B, MnSOD and HSP60; data not shown). In addition, biochemical analysis confirmed the increased mitochondrial distribution of the M26I and D149A DJ-1 mutants (Fig. 2C). Therefore, decreased nuclear distribution and increased mitochondrial co-localization of pathogenic DJ-1 mutants likely result in their diminished interactions with nuclear PSF and p54nrb.
L166P DJ-1 has been shown to be unstable when compared with the wild-type DJ-1 (19–21). In cells either transiently or stably expressing the wild-type or pathogenic mutant DJ-1 (M26I, L166P and D149A), we have found that the steady-state levels of the homozygous L166P and M26I DJ-1 mutants were consistently lower than those of others (data not shown). To evaluate the effects of various DJ-1 mutations on the protein stability, we performed pulse-chase experiments using cells stably expressing Myc-His-tagged wild-type or mutant DJ-1 (Fig. 3A). Although not as unstable as L166P DJ-1, the stability of another mutant found in homozygous state in PD, M26I DJ-1, was significantly (P<0.05) lower than that of wild-type or R98Q DJ-1 after 4 h of chase (Fig. 3B). The stabilities of D149A and wild-type DJ-1 did not differ significantly during the 4 h of chase time we examined. In short, pathogenic DJ-1 mutations may result in accelerated degradation and/or decreased nuclear localization of the protein, and likely attenuate the normal nuclear functions of DJ-1.
DJ-1 blocks transcriptional silencing and apoptosis induced by PSF
Although PSF has been demonstrated to act as a transcriptional repressor (13), we explored the possibility that DJ-1 might modulate this activity. To answer this question, we assayed the transcriptional activity of the p54nrb/PSF complex using a Gal4–p54nrb fusion protein that activates basal transcription of a luciferase reporter gene directed by a promoter containing 5 Gal4-binding sites (G5–Luc) (Fig. 4A, column 2). When PSF is co-expressed with Gal4–p54nrb, it binds p54nrb and represses transcription (Fig. 4A, column 3). PSF did not affect the expression of G5–Luc in the absence of Gal4–p54nrb, indicating that PSF did not non-specifically bind the G5 promoter (Supplementary Material, Fig. S2). PSF also had no effect on the expression of a β-galactosidase reporter gene driven by a CMV promoter (Supplementary Material, Fig. S2) and did not repress G5–Luc in the presence of unrelated Gal4 fusion transcriptional activators, such as Gal4–VP16 and Gal4–c-Jun (data not shown), suggesting that PSF does not bind Gal4 directly or repress transcription indiscriminately. To evaluate the effects of wild-type and mutant forms of DJ-1 on transcription, we co-transfected equal amounts of wild-type or mutant DJ-1 constructs together with Gal4–p54nrb and PSF. Wild-type and the non-pathogenic R98Q DJ-1 mutant, but not the pathogenic DJ-1 mutants, reversed the repression of transcription induced by PSF (Fig. 4A, columns 4–8). In addition, wild-type DJ-1 reversed PSF-mediated transcriptional repression in a dose-dependant manner in both HeLa cells (Fig. 4B, column 3–5) and SH-SY5Y cells (data not shown). To test whether DJ-1 modulates transcription directly when it is bound to a promoter, we co-transfected Gal4-DJ-1 with the G5–Luc reporter plasmid. Consistent with a prior report (4), DJ-1 did not activate reporter gene expression independently (data not shown). To evaluate the cooperative activity between DJ-1 and p54nrb, and to rule out the possibility that the transcriptional regulation by DJ-1 was due to non-specific binding to the Gal4-binding site rather than its interaction with p54nrb, we tested whether DJ-1 affected transcription in the absence and presence of Gal4–p54nrb. DJ-1 activated transcription of the G5–Luc reporter gene only when it was co-expressed with Gal4–p54nrb (Fig. 4C), suggesting that DJ-1 is a transcriptional co-activator that regulates transcription without directly binding any promoter. In addition, DJ-1 cooperated with p54nrb to activate transcription in the absence of transfected PSF (Fig. 4C), indicating that over-expression of DJ-1 reversed the silencing effects of endogenous transcriptional repressors affecting p54nrb, including PSF. Taken together, these results confirm the functional interactions of DJ-1, p54nrb and PSF in the nucleus, and suggest a role of DJ-1 in gene regulation by antagonizing p54nrb-binding transcriptional repressors, particularly PSF.
Loss-of-function mutations of DJ-1 associated with PD would be predicted to increase transcriptional repression by PSF. To determine whether this could predispose cells to neurodegeneration, we transfected SH-SY5Y cells with a PSF expression vector, and then exposed the cells to dopamine, a source of oxidative stress implicated in PD. Expression of PSF, but not a control green fluorescent protein (GFP) vector, markedly increased neuronal apoptosis (Fig. 5A and B, columns 1–2). After being treated with dopamine, >70% of PSF-expressing cells were apoptotic (Fig. 5A and B). Co-expression of the wild-type DJ-1 or the non-pathogenic R98Q DJ-1 mutant abolished PSF-induced apoptosis (Fig. 5B, columns 5 and 9). PD-associated DJ-1 mutants were less effective in blocking PSF-mediated toxicity (Fig. 5B, columns 6–8), consistent with their reduced transcriptional activity. Thus, PSF not only represses gene expression, but can also induce neuronal apoptosis and sensitizes SH-SY5Y cells to dopamine-induced cell death. Wild-type and the non-pathogenic R98Q DJ-1, but not the pathogenic DJ-1 mutants, effectively block PSF-mediated gene silencing and apoptosis. To further explore the effects of the loss of DJ-1 functions, we pre-transfected SH-SY5Y cells with siRNA constructs specifically targeting the human DJ-1 or with non-specific control siRNA constructs, and then re-transfected these cells with the control GFP or PSF plasmids. Reduction of endogenous DJ-1 by DJ-1 siRNA significantly (P<0.01) sensitized cells to PSF-induced apoptosis (Fig. 5C). These results are consistent with the observations from other groups showing increased sensitivity to oxidative stress in cells with null or reduced DJ-1 expression (10,11), and suggest that the balance between PSF-mediated transcriptional repression and DJ-1-regulated transcriptional activation may affect neurodegeneration in dopaminergic cells.
DJ-1 and p54nrb protects against apoptosis induced by α-synuclein and oxidative stress
Accumulation of α-synuclein causes dopaminergic neuronal cell death and directly leads to PD (22–24), likely because of increased oxidative stress signals (22,25). To examine whether DJ-1 is neuroprotective against α-synuclein, we co-transfected dopaminergic SH-SY5Y cells with GFP (control), wild-type human DJ-1 or the indicated DJ-1 mutants together with A30P α-synuclein (Fig. 6A). Consistent with our previous observations (22), over-expression of mutant A30P α-synuclein in native SH-SY5Y cells induced neuronal cell death (Fig. 6A, column 1). Wild-type and R98Q DJ-1 significantly inhibited A30P α-synuclein-induced apoptosis (Fig. 6A, columns 2 and 6). Three pathologic DJ-1 mutants, including M26I, L166P and D149A, showed reduced neuroprotective activity against α-synuclein. The amount of transfected A30P α-synuclein was not affected by the various co-transfected DJ-1 constructs (Supplementary Material, Fig. S3). In addition, wild-type DJ-1 significantly reduced apoptosis in a human primary dopaminergic neuronal culture treated with H2O2 or transfected with α-synuclein (data not shown). These results correlate with the ability of DJ-1 to relieve PSF-mediated transcriptional silencing and apoptosis (Figs 4A and 5). As p54nrb cooperates with DJ-1 to activate transcription (Fig. 4C), we then tested the effects of p54nrb over-expression on oxidative stress or mutant α-synuclein-induced apoptosis. Like wild-type DJ-1, p54nrb significantly prevented apoptosis in cells exposed to H2O2 and A30P α-synuclein accumulation (Fig. 6B). These results strongly suggested that DJ-1 modulates cellular response to oxidative stress via transcriptional regulation in cooperation with p54nrb, and that pathogenic mutations attenuating the nuclear functions of DJ-1 predispose dopaminergic neurons to neurodegeneration.
Our experiments have demonstrated that nuclear proteins p54nrb and PSF are major binding partners of DJ-1 in dopaminergic neuronal cells. DJ-1 reverses the transcriptional silencing activity of the PSF nuclear complex, potentially by modulating the interaction between DJ-1-binding transcriptional co-activator p54nrb and repressor PSF. In addition, DJ-1 prevents apoptosis that is induced by PSF over-expression. The ability of DJ-1 to reverse transcriptional silencing and protect against apoptosis induced by the transcriptional repressor is significantly reduced by DJ-1 mutations that cause PD. These PD-associated mutations result in decreased protein stability and/or nuclear distribution, and diminish the normal nuclear functions of DJ-1. Our results are consistent with the idea that transcriptional repression induces apoptosis by reducing the expression of neuroprotective proteins or anti-apoptotic genes. This conclusion is strengthened by the observation that both DJ-1 and p54 prevent apoptosis induced by PD-causing α-synuclein accumulation or oxidative stress. However, we cannot, at this point, exclude the possibility that PSF may induce apoptosis through one of its other functions involving RNA splicing and editing, which could possibly be modulated by DJ-1. We further showed that reducing intracellular DJ-1 levels lead to increased sensitivity to PSF-mediated apoptosis. Consistent with this observation, cells deficient in DJ-1 or transfected with DJ-1-specific siRNA are susceptible to oxidative and ER stress (10,11). Because of the generation of reactive oxygen species during the metabolism of dopamine (26), dopaminergic neurons may be particularly vulnerable to oxidative stress. Hence, the neuroprotective effects of DJ-1 may be critical to the maintenance and survival of dopaminergic neurons.
All the pathogenic DJ-1 mutants examined (M26I, L166P, D149A) exhibited reduced neuroprotective and transcriptional co-activator activity. However, the molecular mechanisms underlying their loss-of-function are not entirely the same. The L166P and M26I DJ-1 mutants are unstable. These homozygous mutations result in reduced cellular DJ-1 levels, and render dopaminergic neurons vulnerable to various insults, such as oxidative stress (10,11). The stability of M26I DJ-1 has been reported with mixed results (21,27). However, the current study is the first to use quantitative analysis on the basis of multiple pulse-chase experiments. On the other hand, D149A DJ-1 is relatively stable, suggesting a disease-causing mechanism other than loss-of-function caused by protein instability. Although this heterozygous mutation is identified in only one early-onset PD patient (2), our results showed that like the two homozygous mutations in DJ-1, the D149A mutation caused decreased nuclear localization, and increased mitochondrial distribution of DJ-1 protein. These observations agreed with the reduced co-localization of pathogenic DJ-1 mutants with nuclear proteins p54nrb and PSF, and their decreased abilities to regulate transcription and PSF-induced apoptosis. It is unclear, however, whether the increased mitochondrial localization is coincidental or disease-causing. The increased mitochondrial localization of the L166P DJ-1 has been previously reported (1). Owing to the instability of the L166P and M26I mutant DJ-1, an apparent increased mitochondrial localization could be an artifact of accelerated protein turnover by the proteasome in the cytoplasm. However, a similar distribution pattern for the relatively stable D149A raises an interesting possibility that the pathogenic mutants may disrupt the normal mitochondrial functions. Defects in the mitochondrial functions have been linked to pathogenesis of PD (28); and multiple PD-associated proteins may affect mitochondrial functions (29). A previous study using an artificial C106A DJ-1 mutant that is unable to translocate to mitochondria suggested that mitochondrial localization of DJ-1 is required for its neuroprotective effects (30). However, DJ-1 appears to be a multifunctional protein expressed in various cellular compartments with different functions. In addition to the nuclear functions of DJ-1 demonstrated in current and previous studies (6,7), a recent investigation indicates that DJ-1 is a cytosolic redox-dependent molecular chaperone capable of preventing protein aggregation (12). It is unclear whether the C106A mutation also affects the nuclear and cytoplasmic functions of DJ-1, in particular its interaction with the PSF–p54nrb complex and the ability to prevent protein aggregation, and whether these activities could be the basis of reduced neuroprotective activity. As pathogenic DJ-1 mutants show increased mitochondrial localization, it is also of interest to determine whether wild-type and the pathogenic DJ-1 mutants function differently in the mitochondria. Therefore, mutations in DJ-1 may affect multiple pathways in various cellular compartments. It will be of great interest to determine which function is more relevant to PD pathogenesis and whether these pathways are functionally linked. By using knock-in strategies, recently established DJ-1 deficient cell (11) and animal models (31) will be useful tools to test the biological effects of the pathogenic DJ-1 mutations.
The crystal structure (32–34) and biochemical studies (19,21,27,35) of DJ-1 indicate that DJ-1 forms a homodimer. The L166P mutation in DJ-1 not only renders the protein unstable, but also disrupts the dimer formation (19,21,27). However, studies on the ability of other DJ-1 mutants to dimerize generate conflicting results (27,35). Hence, whether the dimer formation is essential for the stability or the functions of DJ-1 remains unclear. A recent study using tagged wild-type and mutant DJ-1 constructs in immuno-precipitation assays suggests reduced homodimerization of pathogenic DJ-1 mutants, including heterozygous A104T and D149A (35). Given the distinct distribution patterns and the reduced nuclear activities of pathogenic DJ-1 mutants we demonstrate here, it will be of interest to examine whether dimerization also affects the subcellular transport and transcriptional activity of DJ-1.
DJ-1 has previously been reported to regulate androgen receptor (AR)-mediated transcription through interactions with PIASxa and DJBP (6,7). However, these DJ-1-interacting proteins are predominantly expressed in testis, and were not among the major DJ-1-binding proteins that we resolved in dopaminergic neuronal cells. A more recent report suggests that DJ-1 regulates p53 transcriptional activity by binding Topors/p53BP3 (36). Like its regulation of p54nrb–PSF-modulated gene transcription that we describe here, DJ-1 restores transcription repressed by PIASxa, DJBP and Topors/p53BP3 either by preventing the recruitment of the histone deacetylase complex (17,37) or by modulating sumoylation (36). Therefore, DJ-1 functions as a transcriptional co-activator and regulates gene expression in various tissues by cooperating with different transcription factors such as AR, p54nrb and p53 and antagonizing repressor activities. The link between DJ-1 and p53 is intriguing. p53 is a tumor suppressor protein governing cell growth, apoptosis as well as DNA repair and stress response (38). Therefore, because of the neuroprotective effect and the potential involvement of DJ-1 in the oxidative stress response, it will be of interest to determine whether DJ-1 selectively activates p53-mediated stress response, but not the apoptotic pathway.
MATERIALS AND METHODS
Plasmids, chemicals and cells
Human wild-type DJ-1 cDNA was amplified by RT–PCR using RNA from SH-SY5Y cells and inserted in frame into pcDNA3(−) Myc-His vector between the XhoI and HindIII restriction sites. PCR mutagenesis was used to generate all the DJ-1 mutants. Mutations were confirmed by sequencing. Human PSF expression plasmid was kindly provided by Dr Randall Urban (University of Texas Medical Branch, Galveston, TX, USA). GST–p54nrb used as a template for PCR cloning were gifts from Dr James Patton (Vanderbilt University, Nashville, TN, USA). Gal4–p54nrb and Flag-tagged pCMVp54nrb were generated by PCR subcloning into pM vector (Clontech) or pCDNA3.1. Sequences containing 5 Gal4-binding sites were inserted into pGL3basic (Promega) to generate the G5–Luc reporter plasmid. pCMV-β-Gal plasmid encodes β-galactosidase. pEGFP plasmid was from Clontech. Dopamine and hydrogen peroxidase were from Sigma. Human HeLa and neuroblastoma SH-SY5Y cells were used in various transfection experiments. Cells stably expressing Myc-His-tagged wild-type or mutant DJ-1 were established by transiently transfection followed by G418 (800 µg/ml) selection. Clones were picked, expanded and tested for transgene expression.
Identification of DJ-1-associated proteins and mass spectrometry analysis
At least 6×108 SH-SY5Y cells stably transfected with an empty vector or Myc-His-tagged human wild-type DJ-1 were harvested and lysed in non-denaturing buffer containing 0.5% NP-40, 50 mm NaH2PO4, 300 mm NaCl, 10 mm imidazole and 1× protease inhibitors (Roche), and incubated with 300 µl of Ni-NTA agarose beads (Qiagen) at 4°C with rocking overnight. Beads were washed 6 times with washing buffer containing 40 mm imidazole. Unbound supernatant and washing buffers were collected for later analysis. Bound proteins were eluted twice with 300 µl of elution buffer containing 250 mm imidazole. Eluted proteins were pooled and resolved by 4–20% Tris–glycine gel (Invitrogen). The gel was fixed and stained with Coomassie blue to reveal proteins purified by Ni-NTA beads. The three bands specifically present in DJ-1 stable cells were excised and subjected to trypsin proteolysis. The gel slices from control lane at the positions corresponding to these three proteins were excised and analyzed simultaneously. MS/MS spectra of resulting peptides were acquired by nanoscale microcapillary LC-MS/MS (39) on a DECA XP mass spectrometer (ThermoElectron). Spectra were searched with the SEQUEST algorithm against human entries from the non-redundant protein sequence database (NRP) available from the NCBI.
Immunoprecipitation and immunoblotting
To immunoprecipitate endogenous DJ-1, a rabbit polyclonal antibody against the C-terminus of DJ-1 was used (I. Woldman and C. Pifl). Co-precipitated endogenous p54nrb and PSF were detected with monoclonal antibodies against p54 (1:1000, BD Bioscience) and PSF (1:1000, Sigma).
Co-localization of DJ-1 with p54, PSF and mitochondria
Native SH-SY5Y cells were labeled with a rabbit polyclonal anti-DJ-1, and monoclonal antibodies against p54nrb (1:1000, BD Bioscience) and PSF (1:1000, Sigma) for endogenous proteins. To study the co-localization of wild-type or mutant DJ-1 with p54 or PSF, cells stably expressing various Myc-His-tagged DJ-1 constructs were fixed and labeled with anti-His and anti-p54nrb or PSF. Antibodies against mitochondrial markers (cytochrome c, BD Bioscience; MnSOD, Stressgen; and HSP60, Santa Cruz) were used with anti-His antibodies to double label mitochondria and DJ-1, followed by labeling with appropriate Cy2 or Cy3-conjugated secondary antibodies (Jackson ImmunoResearch Lab).
Mitochondrial DJ-1 content
SH-SY5Y cells stably expressing the wild-type, M26I, D149A or R98Q DJ-1 were harvested in 1× cold PBS, and mechanically disrupted with Dounce homogenizers in buffer A [0.25 m sucrose, 5 mm HEPES, pH 7.5, 5 mm KCl, 1 mm MgCl, fresh protease inhibitor (1×, Roche) and DTT (1 mm final)]. Disruption of the cells (>70%) was confirmed by microscopy examination. Unlysed cells, debris and nuclei were removed by centrifugation at 1200g twice for 10 min each. The resulting supernatant was centrifuged again at 10 000g for 30 min to precipitate mitochondria. The mitochondrial pellet was washed once in buffer A and lysed in RIPA-DOC buffer (50 mm Tris, pH 7.2, 150 mm NaCl, 1% Triton X-100, 1% deoxycholate and 0.1% SDS) with 1× fresh protease inhibitor cocktail. For total cell lysate, cells were harvested in 1× cold PBS and directly lysed in RIPA-DOC buffer. Equal amounts (30 µg) of mitochondrial or total proteins were loaded in 4–20% Tris–glycine gels (Invitrogen). A mouse anti-DJ-1 monoclonal antibody (Stressgen) was used to detect Myc-His-tagged DJ-1. A rabbit polyclonal antibody against mitochondrial HSP60 and a goat polyclonal antibody against β-actin (both from Santa Cruz) were used to detect these proteins as loading controls. Mitochondrial DJ-1 content was determined using desitometry and Image J software (NIH).
SH-SY5Y cells stably expressing wild-type and mutant DJ-1 were washed twice and pre-incubated for 20 min in methionine- and cysteine-free medium. The cells in 10 cm culture dishes were then pulse labeled for 40 min with 2 ml of 0.2 mCi/ml [35S]-Met and Cys (Easy Tag Express 35S protein labeling mix, Perkin–Elmer Life Sciences). After being labeled, the cells were chased for 0, 0.5, 1, 2 and 4 h with growth medium containing 0.5 mg/ml of unlabeled methionine and 0.2 mg/ml cysteine. Following the chase, the cells were harvested in ice-cold PBS buffer at different time points, and lysed in non-denaturing lysis buffer (50 mm Tris–HCl, pH 7.5, 300 mm NaCl, 5 mm EDTA, 0.02% sodium azide, 1% Triton X-100 and 1× protease inhibitor cocktail). Equal amounts of lysate from each time point were pre-cleared and incubated with a rabbit anti-His polyclonal antibody (Santa Cruz) overnight at 4°C with rocking. Protein A/G beads were used to precipitate the immuno-complex, and were washed five times before SDS–PAGE analysis. Images were acquired with Storm Phosphoimager (GE Biosciences) and quantified with Image J software (NIH). Instat 3.0 was used for the statistical analysis.
Transcription reporter assay
HeLa or SH-SY5Y cells in 24-well plates were transfected with reporter genes G5–Luc and pCMV-β-Gal (0.05 µg each) and 0.5 µg of Gal4–p54nrb, 0 or 0.25 µg of CMV–PSF, and 1 µg (Fig. 4A) or indicated amount of wild-type or mutant DJ-1 plasmids when needed. GFP or pcDNA3 plasmids were used to balance the amount of input DNA in each condition. Cells were harvested 48 h after transfection for luciferase and β-galactosidase assays (Promega). All readings were in linear range. Results represent luciferase activity normalized with β-galactosidase activity from at least three experiments in duplicates or triplicates.
Transfection, immunofluorescence microscopy and analysis of neuronal apoptosis
SH-SY5Y cells plated on coverslips in 24-well dishes were transfected using Lipofectamine 2000 (Invitrogen) or Transfectin (Bio-Rad) reagents. To assess the protective effect of DJ-1 on α-synuclein toxicity, 2 µg of A30P α-synuclein was co-transfected with 1 µg of wild-type or mutant DJ-1 expression plasmids. Similarly, 1 µg of PSF was co-transfected with 1 µg of DJ-1 plasmids in PSF toxicity assays. To examine the neuroprotective effect of p54nrb, 1 µg of Flag-tagged p54nrb were transfected with 2 µg of myc-tagged A30P α-synuclein or treated with 250 µm of H2O2 for 24 h before harvest. For controls, equal amount of either GFP or pcDNA3 empty vector was used to match transfected DJ-1. Forty-eight hours after transfection, cells were fixed and labeled with appropriate primary and Cy2/Cy3 conjugated secondary antibodies, followed by immunofluorescence microscopy and analysis of neuronal apoptosis as described (22). α-Synuclein, PSF, DJ-1-His and p54nrb were labeled with Syn-1 (1 : 600, BD Bioscience) or anti-myc (1 : 600, Santa Cruz), a rabbit polyclonal anti-PSF (1 : 500, kindly provided by Dr James Patton), anti-His probe (sc-804 or sc-8036, 1 : 600, Santa Sruz) and anti-Flag (1 : 500, Sigma), respectively.
DJ-1 siRNA transfection
SH-SY5Y cells plated in six-well cultured dishes were transfected with 100 nm pooled human DJ-1 specific siRNA constructs or non-specific control siRNA constructs (siGene siRNA, Dharmacom) using Transfectin reagent (Bio-Rad). Cells were trypsinized and re-plated on coverslips in 24-well dishes 36 h after siRNA transfection. After another 12 h, siRNA-treated cells were re-transfected with either 0.5 µg of GFP or PSF, and were incubated for another 48 h before fixation and apoptosis analysis as described earlier. Separate samples collected 48 h after control or DJ-1-specific siRNA transfection were used to assess DJ-1 expression.
Supplementary Material is available at HMG Online.
We thank Drs James Patton and Randall Urban for valuable reagents; Nick Nagykery for technical assistance; Drs Alvin Lyckman, Kenneth Rosen, Adrian Isaacs, Shyan-yuan Kao for critical reading of the manuscript and for helpful discussion. This work was supported by a research grant from American Parkinson Disease Association to (J.X), and in part by National Institutes of Health grant AG17573 (to B.A.Y.).