Oxidative stress contributes to the development of neurodegenerative diseases. DJ-1, a protein genetically linked to Parkinosn's disease (PD), has been implicated in oxidative stress defense and transcriptional regulation. However, it is unclear whether these two aspects of the DJ-1 function are connected. Here, we show that the inactivation of DJ-1 causes decreased expression of the human MnSOD. DJ-1 stimulates the activity of a master regulator of mitochondrial biogenesis and stress response, peroxisome proliferator-activated receptor-γ co-activator 1α (PGC-1α), in the transcription of the MnSOD. Although DJ-1 does not interact with PGC-1α directly, it inhibits the SUMOylation of a transcriptional repressor, pyrimidine tract-binding protein-associated splicing factor (PSF). PSF binds PGC-1α and suppresses its transcriptional activity. In contrast, a SUMOylation-deficient PSF mutant exhibits reduced binding to PGC-1α and promotes its activity. SUMO-specific isopeptidase SENP-1 further enhances the synergy between DJ-1 and PGC-1α, whereas an SUMO E3 ligase protein inhibitor of activated STAT Y completely blocks the synergy. Conversely, oxidative modification renders DJ-1 unable to inhibit SUMOylation, resulting in attenuated transcriptional synergy between DJ-1 and PGC-1α. Therefore, our results validate DJ-1 as a transcriptional regulator in mitochondrial oxidative stress response and imply that the oxidation-mediated functional impairment of DJ-1 leads to gradual dysregulation of the SUMO pathway. Consequent abnormal mitochondrial gene expression may contribute to the development of sporadic PD.
Mutations in DJ-1 cause early onset autosomal recessive juvenile Parkinsonism ( 1 ). Originally identified as an oncogene, DJ-1 was subsequently found to promote androgen receptor-mediated transcription by preventing the recruitment of repressor complexes ( 2 , 3 ). After DJ-1 was linked to Parkinosn's disease (PD), a number of studies have demonstrated the neuroprotective activity of DJ-1. DJ-1 activates the AKT pathway ( 4 ), disrupts apoptotic signaling ( 5 ) and relieves protein-associated splicing factor (PSF)-mediated transcriptional repression and apoptosis ( 6 ). DJ-1 is induced by oxidative stress signals and can be readily oxidized. This oxidative modification leads to a PI shift of DJ-1 from 6.2 to 5.8 in cultured cells exposed to H 2 O 2 or paraquat ( 7–9 ). Oxidized DJ-1 accumulates in the brain tissues from patients with Parkinson's or Alzheimer's disease ( 10 ) and increases with age in Drosophila ( 11 ). Furthermore, recent evidence indicates that the levels of mitochondrial H 2 O 2 increase in DJ-1 knockout mice ( 12 ). Thus, at least part of the neuroprotective activity of DJ-1 is mediated by its role as a cellular-free radical scavenger. However, whether oxidative modification of DJ-1 affects its other reported functions, such as transcriptional regulation, remains unknown.
Mounting evidence indicates that mitochondrial dysfunction causes PD ( 13 , 14 ). Mitochondrial toxins, such as MPTP and rotenone, induce PD-like symptoms in human ( 15 ) and in rodent models ( 16 ). In addition, pathogenic mutations in a number of PD genes, including α-synuclein , Parkin , DJ-1 and PINK1 , lead to increased sensitivity to oxidative stress and mitochondrial poisoning ( 17–22 ). Despite the differences in their functions, proteins encoded by some of these PD genes act in common pathways to regulate mitochondrial activity. For instance, inactivation of a serine/threonine kinase PINK1 or a ubiquitin E3 ligase Parkin in Drosophila resulted in almost identical phenotypes including mitochondrial abnormalities and muscle degeneration. Further, abnormal phenotypes caused by the loss of PINK1 were rescued by the expression of human Parkin ( 20–22 ). Although DJ-1 may not share the same biochemical pathways as PINK1 or Parkin, it appears to regulate the mitochondrial oxidative stress response ( 14 ). In addition to serving as an antioxidant, it induces the expression of the rate-limiting enzyme for glutathione synthesis, glutamate cysteine ligase, in a rat dopaminergic cell line ( 23 ). Further, DJ-1 has been shown to stabilize nuclear factor erythroid 2-related factor (Nrf2), a transcriptional regulator involved in the induction of mitochondrial antioxidative proteins ( 24 ). Previously, we have shown that DJ-1 is a transcriptional co-activator suppressing protein SUMOylation ( 25 ). SUMOylation is a reversible enzymatic process to attach the small ubiquitin-like modifier (SUMO) to a lysine residue (K) in the protein substrate typically at a conserved SUMO consensus motif, ΨKXD/E, where ψ is any large hydrophobic residue, X can be any residue between the target lysine and aspartate or glutamate ( 26 , 27 ). SUMOylation affects the subcellular localization and transcriptional activity of the target protein ( 26 , 27 ). However, it is unclear whether the SUMO-regulating ability of DJ-1 is associated with its role in the mitochondrial oxidative stress response.
We have previously identified PSF as a major DJ-1-interacting protein ( 6 ). This interaction was subsequently validated in a large-scale proteomic study ( 28 ). Essential for pre-mRNA splicing in vitro ( 29 ), PSF harbors RNA-binding motif ( 30 ) and is a component of the spliceosome C complex ( 31 ). Besides regulating mRNA processing, PSF binds DNA and acts as a transcriptional corepressor ( 32 ). PSF is SUMOylated at the lysine residual 338, and the SUMOylation of PSF facilitates the recruitment of a transcriptional repressor histone deacetylase 1 (HDAC 1) ( 25 ). By inhibiting the SUMOylation of PSF, DJ-1 promotes the transcription of tyrosine hydroxylase (TH), the rate-limiting enzyme for dopamine synthesis, in a human dopaminergic neuroblastoma cell line ( 25 ).
Peroxisome proliferator-activated receptor-γ co-activator 1α (PGC-1α) is a potent transcriptional co-activator regulating the mitochondrial biogenesis and oxidative stress response by inducing and activating the transcriptional factors nuclear respiratory factor (NRF) 1 and 2 (33). NRF 1 and 2 then promote the expression of nuclear-encoded mitochondrial genes, such as β-ATP synthase, cytochrome c and cytochrome- c -oxidase subunit IV. In addition, NRFs induce mitochondrial transcriptional factor A, a key activator of mitochondrial DNA replication and transcription ( 33 ). Thus, PGC-1α triggers a signaling cascade to boost mitochondrial function. Recently, inactivation of PGC-1α has been implicated in neurodegeneration. PGC-1α knockout mice exhibit decreased expression of mitochondrial enzymes involved in oxidative stress response, such as MnSOD and catalase, and are more susceptible to MPTP-induced dopaminergic neuronal loss ( 34 ). In mouse models of Huntington's disease, mutant huntingtin suppresses the expression of PGC-1α or PGC-1α target genes and leads to mitochondrial dysfunction and thermoregulatory defects ( 35 , 36 ). These results strengthen the link between neurodegenerative diseases and mitochondrial and transcriptional dysfunction.
In this study, we report a novel functional link between DJ-1 and PGC-1α in the transcriptional regulation of the human MnSOD. DJ-1 enhances the transcriptional activity of PGC-1α. The synergy between these two proteins is caused by DJ-1-mediated de-SUMOylation of PSF. SUMOylation-deficient PSF exhibits reduced binding affinity for PGC-1α. In addition, the oxidation of DJ-1 blunts its ability to inhibit SUMOylation and leads to decreased synergy with PGC-1α in the induction of the human MnSOD. These findings further validate the role of DJ-1 in the regulation of mitochondrial oxidative stress response and demonstrate the functional impairment of DJ-1 by oxidation. In addition, transcriptional dysregulation caused by the inactivation of DJ-1, either by pathogenic mutations or oxidation, results in defective mitochondrial gene regulation that may contribute to the development of sporadic PD and other neurodegenerative diseases.
DJ-1 induces the expression of human MnSOD
As DJ-1 is an oxidative stress responsive protein and has been implicated in transcriptional regulation and mitochondrial function, we examined whether genes involved in mitochondrial detoxification were affected by DJ-1 inactivation. We transfected a human dopaminergic neuroblastoma cell line CHP-212 with siRNA constructs specifically targeting DJ-1. Compared with the more commonly used SH-SY5Y neuroblastoma cell line, CHP-212 cells generate higher levels of l -DOPA, proliferate at a slower rate and retain more characteristics of dopaminergic neurons ( 25 , 37 ). Transfection of the DJ-1 siRNA specifically and consistently reduced the protein levels of the endogenous DJ-1 by more than 70% ( 25 ) (Fig. 1 B) and resulted in ∼20–50% decrease of the mRNA expression of genes involved in reactive oxygen species (ROS) removal, such as MnSOD, catalase, glutamate cysteine ligase catalytic subunit (GCLC) and glutathione peroxidase 1 (Gpx1) (Fig. 1 A). As the mRNA expression of MnSOD was the one most affected by DJ-1 knockdown, we decided to focus on this key detoxification enzyme as a major DJ-1 target gene in mitochondria. Consistent with decreased mRNA levels, the protein expression of MnSOD was inhibited by 50% in CHP-212 cells transfected with DJ-1 siRNA constructs (Fig. 1 B and C). To confirm the regulation of human MnSOD by DJ-1 in vivo , we assessed the expression of MnSOD in the lymphoblasts from a normal individual and two patients with familial PD carrying either the homozygous L166P point mutation or the exons 1–5 deletion mutation in the DJ-1 gene. These pathogenic mutations were originally described by Bonifati et al . ( 1 ) in the report that linked DJ-1 to PD. Although the DJ-1 level was undetectable in lymphoblast cells from the two PD patients, the MnSOD expression was significantly reduced (Fig. 1 D and E). Conversely, MnSOD was induced in SH-SY5Y cells stably expressing the wild-type (WT) DJ-1, but not in cells expressing either a control vector or the homozygous M26I mutant DJ-1 ( Supplementary Material, Fig. S1 ). D149A DJ-1, a heterozygous mutant that exhibited similar degree of neuroprotection as the WT DJ-1 ( 6 ), slightly induced MnSOD expression. Therefore, these results are consistent with DJ-1 as a regulator of the human MnSOD.
DJ-1 and PGC-1α synergistically activate the MnSOD promoter
Prior studies have indicated that DJ-1 is a transcriptional co-activator ( 3 , 6 , 25 ). Thus, we examined whether DJ-1 induced transcription from the human MnSOD promoter. Interestingly, transient expression of DJ-1 did not significantly affect the expression of a luciferase reporter gene directed by the human MnSOD promoter (Fig. 2 , first two columns). We then considered the involvement of other transcriptional regulators. Recently, several reports have demonstrated potential roles of the transcriptional co-activator PGC-1α in the prevention of neurodegeneration in mouse models. PGC-1α knockout mice, such as DJ-1 knockout mice, are susceptible to MPTP toxicity and demonstrate deficits in mitochondrial oxidative stress responses ( 34 ). Furthermore, MnSOD, catalase and Gpx1 are target genes regulated by PGC-1α ( 34 ). Taken together, these data suggested that DJ-1 and PGC-1α might affect mitochondrial function by transcriptionally regulating common target genes. To test this possibility, we co-transfected DJ-1, PGC-1α with MnSOD-luciferase reporter gene and determined luciferase activity 24 h post-transfection. Although PGC-1α, but not DJ-1, induced a 3-fold increase in the luciferase activity, co-expression of DJ-1 and PGC-1α resulted in an 8-fold synergistic activation (Fig. 2 , column 4).
Next, we examined whether this synergy between DJ-1 and PGC-1α in the MnSOD promoter activation was the result of their direct interaction. Co-immunoprecipitation experiments using lysates from native neuroblastoma cells or human embryonic kidney (HEK) 293 cells overexpressing both DJ-1 and PGC-1α failed to demonstrate convincing interactions between these two proteins (data not shown). As PGC-1α harbors sequences matching the consensus site for SUMOylation, we evaluated whether PGC-1α was a SUMO substrate and whether DJ-1 might affect the SUMOylation of PGC-1α. However, we failed to detect SUMOylated PGC-1α species by immunoprecipitating either the overexpressed PGC-1α or native protein with specific antibodies, followed by western blotting using an anti-SUMO-1 antibody (data not shown).
The SUMOylation status of PSF affects the transcriptional activity of PGC-1α
Previously, we showed that DJ-1 interacts with a transcriptional repressor PSF and blocks the transcriptional repression by PSF ( 6 ). Furthermore, we found that this de-repression by DJ-1 is mediated by its ability to suppress the SUMOylation of PSF ( 25 ). Therefore, it is conceivable that PSF and its SUMOylation are involved in the transcriptional synergy between PGC-1α and DJ-1. We evaluated whether nuclear proteins PSF and PGC-1α formed a complex. HEK293 cells were co-transfected with the human WT PGC-1α and PSF or control pcDNA3 and were lysed in non-denaturing buffer after 24 h. In co-immunoprecipitation assays, an anti-PSF monoclonal antibody specifically pulled down PGC-1α (Fig. 3 A).
We then assessed whether the SUMOylation status of PSF would affect the interaction between PSF and PGC-1α. We previously identified the SUMOylation site within PSF at the lysine residue 338 and showed that a lysine to alanine mutation at this site (K338A) abolished the SUMOylation of PSF ( 25 ). Further, this SUMO-deficient PSF mutant exhibited reduced ability to repress transcription due to its decreased binding affinity for HDAC 1 ( 25 ). Flag-tagged WT or K338A PSF was co-expressed with Flag-tagged PGC-1α in HEK293 cells, and non-denaturing lysates were collected and immunoprecipitated with either a control IgG or a monoclonal anti-PSF antibody. In multiple experiments, the K338A PSF was consistently more stable and expressed at higher levels than the WT PSF in the presence of PGC-1α (Fig. 3 B, lysate panel). Although a much higher amount of K338A PSF was pulled down than that of the WT PSF (immunoprecipitated Flag-tagged PSF marked by * in Fig. 3 B), the mutant PSF co-precipitated slightly less amount of PGC-1α (marked by # in Fig. 3 B). Quantitative analysis of the ratio between co-immunoprecipitated PGC-1α and total immunoprecipitated PSF from two independent co-IP experiments indicated a reduced binding affinity of PGC-1α for the K338A PSF (unSUMOylated) versus WT PSF (Fig. 3 C). As DJ-1 inhibits the SUMOylation of PSF, we examined the direct effect of DJ-1 on PGC-1α and PSF binding. Consistent with the results shown in Fig. 3 B and C, overexpression of the WT DJ-1 inhibited the binding between PGC-1α and PSF in co-IP assays ( Supplementary Material, Fig. S2 ).
To evaluate whether the interaction between PSF and PGC-1α affected transcription from the MnSOD promoter, we co-transfected PGC-1α and the WT PSF or K338APSF with MnSOD-luciferase reporter. Consistent with its binding to PGC-1α and its role as a transcriptional repressor recruiting HDAC 1 ( 25 ), WT PSF inhibited the activation from the human MnSOD promoter by PGC-1α by 50% (Fig. 4 A). Interestingly, the SUMO-deficient mutant PSF significantly enhanced the transactivation by PGC-1α. These results agree with DJ-1 as an inhibitor of PSF SUMOylation ( 25 ) and an activator for PGC-1α, and suggest that the SUMOylation status of PSF affects the induction of MnSOD by PGC-1α.
Given the SUMO-dependent regulation of PGC-1α activity by DJ-1 and PSF, we reasoned that other known regulators of SUMOylation would either enhance or repress PGC-1α-mediated activation of MnSOD. We tested the effects of Sentrin/SUMO-specific proteases-1 (SENP-1), a general SUMOylation inhibitor that removes the SUMO group from SUMOylated substrates ( 38 ), and protein inhibitor of activated STAT Y (PIASy), an E3 SUMO ligase that has been shown to promote PSF SUMOylation ( 25 ), on the synergistic activation of MnSOD-luciferase by DJ-1 and PGC-1α (Fig. 4 B). SENP-1 increased the luciferase activity by another 27% ( P < 0.05). In contrast, PIASy repressed the DJ-1–PGC-1α-induced activation of MnSOD by 55% ( P < 0.01). These results confirmed the role of SUMO regulation in the transcription of human MnSOD.
Oxidation impairs DJ-1’s ability to regulate SUMOylation and transcription
DJ-1 is susceptible to oxidation during oxidative stress. To test whether the oxidation of DJ-1 affected its synergy with PGC-1α in the transcriptional activation of MnSOD, we pre-treated SH-SY5Y cells with increasing amount of H 2 O 2 for 24 h and then co-transfected the cells with MnSOD-luciferase and DJ-1, PGC-1α or both (Fig. 5 A). Luciferase activity was determined 24 h post-transfection. Although no significant difference in the transactivation was found in cells exposed to lower doses of H 2 O 2 (up to 200 µ m ), a significant reduction (∼30%, P < 0.05) of the synergy between DJ-1 and PGC-1α was observed in cells treated with 400 µ m of H 2 O 2 . A parallel analysis using two-dimensional PAGE indicated a clear increase of oxidized DJ-1 in the lysate from cells treated with 400 µ m of H 2 O 2 , but not in cells exposed to lower doses of H 2 O 2 (Fig. 5 B). These results support the idea that oxidation of DJ-1 attenuates its function.
As DJ-1 is an inhibitor of SUMOylation ( 25 ), we examined whether oxidative stress affected SUMOylation via the functional impairment of DJ-1. To assess the global effect on SUMOylation by oxidative stress, we exposed the SH-SY5Y cells stably expressing WT DJ-1 to increasing doses of H 2 O 2 and determined the abundance of total SUMOylated protein. In cells untreated with H 2 O 2 , overexpression of DJ-1, but not an empty vector, resulted in decreased levels of high molecular weight SUMOylated proteins (Fig. 6 A, lanes 1 and 2). Although lower doses of H 2 O 2 (100 and 200 µ m ) did not affect DJ-1’s ability to suppress SUMOylation, 400 µ m of H 2 O 2 prevented the SUMO inhibition by DJ-1 (Fig. 6 A, lanes 3–5). Furthermore, SUMOylated PSF species increased in these cells even though the total PSF or DJ-1 levels were not affected by 400 µ m of H 2 O 2 (Fig. 6 B). These observations are consistent with our previous data showing the accumulation of SUMOylated PSF in lymphoblast cells from the PD patients carrying either the L166P or exons 1–5 deletion mutation in the DJ-1 gene ( 25 ). Collectively, these results indicate that oxidative modification renders DJ-1 incapable of suppressing SUMOylation and results in attenuated transcriptional synergy between DJ-1 and PGC-1α in the activation of the human MnSOD promoter.
Although DJ-1 has been shown to regulate transcription and mitochondrial function, this is the first report directly demonstrating the transcriptional regulation of a key mitochondrial detoxification enzyme by DJ-1 in human cells. The ability of DJ-1 to promote the transcriptional activity of PGC-1α validates the involvement of DJ-1 in the mitochondrial function. Furthermore, this study underscores DJ-1’s role as a transcriptional co-activator and a regulator of SUMOylation. Upon oxidative modification, DJ-1 loses its ability to suppress the SUMOylation of PSF and this results in attenuated transcriptional synergy with PGC-1α (Fig. 7 ). Given the evidence showing increased DJ-1 oxidation in AD and PD patients, the current study implies that aging-related oxidative impairment of DJ-1 leads to gradual dysregulation of the SUMO pathway and affects the transcription of genes involved in mitochondrial functions and neuroprotection.
Although the human MnSOD is transcriptionally regulated by DJ-1 and PSF, mouse MnSOD is not significantly affected by DJ-1 inactivation in vitro ( 39 ) or in vivo ( 12 ). Owing to compensatory response, the protein levels of MnSOD increase in older DJ-1 knockout mice ( 12 ). However, we have confirmed decreased expression of MnSOD both in human dopaminergic neuroblastoma CHP-212 cells treated with DJ-1-specific siRNA and in lymphoblast cells from two patients with familial PD carrying loss-of-function mutation in the DJ-1 gene. We have reported a similar species-specific regulation of the TH by DJ-1. Although DJ-1 silencing leads to decreased expression of the human TH and causes reduced l -DOPA production in CHP-212 cells ( 25 ), the expression of TH is unchanged in DJ-1 knockout mice ( 19 , 40 ). Although the molecular basis for the species-specific transcriptional regulation by DJ-1 remains unclear, the muted regulation of these DJ-1 target genes in mice may contribute to the subtle motor deficits and the lack of histological abnormalities in DJ-1 knockout mice ( 19 , 40 ).
This study revealed the regulation of the PGC-1α activity by the DJ-1–PSF protein complexes. Although DJ-1 did not directly interact with PGC-1α nor regulated the SUMOylation of PGC-1α, DJ-1 promoted the transcriptional activity of PGC-1α via a common binding partner, PSF. We have previously demonstrated that DJ-1 enhances the transcription of the human TH via a similar mechanism. By inhibiting the SUMOylation of PSF, DJ-1 prevents the recruitment of transcriptional repressor HDCA 1 to the TH promoter by PSF ( 25 ). In addition to PGC-1α, the DJ-1–PSF complexes affect the transcriptional activity of androgen receptor. We found that androgen receptor interacted with DJ-1 and induced the expression of the human TH (N.Z. and J.X., unpublished observation). This observation is consistent with the result showing DJ-1 as a co-activator of androgen receptor ( 3 ), and recent evidence demonstrating the repression of androgen receptor by PSF ( 41 ). Furthermore, SUMO-specific isopeptidase SENP-1 enhances androgen receptor-mediated transcription by deSUMOylating HDAC 1 ( 38 ). It is conceivable that DJ-1 may also inhibit the SUMOylation of HDAC 1 directly, thus preventing both the recruitment and repressive activity of HDAC 1. As PGC-1α and androgen receptor are transcriptional regulators that affect energy metabolism, oxidative stress response, aging and apoptosis ( 33 , 42 ), the pro-survival activity of DJ-1 in tumorgenesis and PD prevention may be mediated by its activity as a transcriptional co-activator for androgen receptor and PGC-1α.
The WT and SUMOylation-deficient (K338A) PSF exhibit differential effects on transactivation from the human MnSOD promoter by PGC-1α. Although the WT PSF causes transcriptional repression, K338A PSF significantly enhances the transactivation by PGC-1α. These observations provide a mechanistic basis for the synergy between DJ-1 and PGC-1α. As DJ-1 suppresses the SUMOylation of PSF, the expression of either DJ-1 or K338A PSF results in increased pool of unSUMOylated PSF species, which are poor recruiters of the transcriptional repressor HDAC 1 ( 25 ). Despite its low binding affinity with PGC-1α, the SUMOylation-deficient K338A PSF still competes with WT PSF for PGC-1α, thus leading to decreased recruitment of HDAC 1. Like DJ-1, the SUMO protease SENP-1 promotes the transcriptional activity of PGC-1α. Conversely, the SUMO E3 ligase PIASy suppresses the transcriptional activation of MnSOD-luciferase by PGC-1α. These results strongly suggest a SUMO-dependent regulation of the PGC-1α activity by DJ-1 and PSF in the transcription of human MnSOD, and suggest a general role for SUMO inhibitors in the regulation of mitochondrial function and oxidative stress response.
The interaction between PGC-1α and PSF is intriguing. Like PSF, PGC-1α contains RNA-binding motif that are characteristics of splicing factors. Moreover, PGC-1α co-localizes with splicing factors in nuclear speckles and affects RNA processing when it is loaded onto the promoters of target genes ( 43 ). It will be of interest to evaluate whether the interaction between these two proteins affects RNA processing besides transcription and whether SUMOylation is involved.
DJ-1 is oxidized during oxidative stress conditions. Using mass spectrometry analysis, Choi et al. ( 10 ) have demonstrated that oxidized DJ-1 species accumulate in the brains of patients with PD or Alzheimer's disease. In addition, oxidized DJ-1 species increase with age in Drosophila , and exhibit reduced neuroprotective activity ( 11 ). These studies raise the possibility that the neuroprotective function of DJ-1 is impaired by the oxidative modification. In this report, we demonstrated that oxidative modification abolished DJ-1’s ability to inhibit SUMOylation and led to decreased synergy between DJ-1 and PGC-1α in the transcriptional regulation of the human MnSOD. Interestingly, oxidative modification, reduced SUMO-inhibiting activity and decreased transcriptional activity of DJ-1 only became apparent when the cells were exposed to higher dose of H 2 O 2 . These results are consistent with the concept that DJ-1 contributes to the ROS removal via two functional aspects, as both a free radical scavenger and a transcriptional regulator promoting the expression of genes involved in oxidative stress defense, such as MnSOD. A transcriptionally active DJ-1 promotes the expression of MnSOD and reduces the cellular pool of free radicals, thus preventing the accumulation of oxidized DJ-1. However, the build-up of free radicals during aging may overwhelm the cellular ROS defense machinery and lead to the accrual of oxidized DJ-1. Subsequently, the oxidation of DJ-1 impairs its ability to regulate SUMOylation and transcription, and sets up a vicious cycle that has the potential to exacerbate oxidative stress, thus contributing to the development sporadic PD and other neurodegenerative diseases.
MATERIALS AND METHODS
Cell culture and plasmids
Human neuroblastoma CHP-212 cells (ATCC) and SH-SY5Y cells were maintained in 10% FBS containing EMEM/F-12 (50/50%) and DMEM, respectively. SH-SY5Y cells stably expressing empty vector or WT DJ-1, as well as pcDNA3-Myc-His-DJ-1, pcDNA3-Flag-PSF, pcDNA3-Flag-PSF338 and pcDNA3-Myc-PIASy were described previously ( 6 , 25 ). pcDNA4-Myc-PGC-1α was purchased from Addgene, Inc. pcDNA3-Flag-SENP-1 was kindly provided by Yeh ( 38 ). The MnSOD-luciferase reporter construct was a gift from Khochbin ( 44 ).
Transfection of siRNA, RNA extraction and quantitative RT-PCR (Q-PCR)
CHP-212 cells were plated in six-well culture dishes in triplicates and were transfected with 100 n m of siRNA specifically for human DJ-1 or non-specific control siRNA constructs (Dharmacon) using the Transfectin reagent (Biorad, Hercules, CA) as described ( 25 ). Cells were harvested 48 h post-transfection for RNA extraction using the TriZol reagent and RNAeasy kit (Qiagen, Valencia, CA). On-column treatment of RNA with RNase-free DNaseI (Qiagen) was performed to remove any possible remaining genomic DNA. RNA quality was examined by gel-electrophoresis. Two-step, syber green-based approach (Applied Biosystem, Foster City, CA) was applied in Q-PCR. cDNAs were generated from equal amount of RNA template and serially diluted for PCR quantification. Equal amount of cDNAs of all analyzed samples (i.e. CTL and DJ-1 siRNA treated) were pooled and serially diluted to generate samples for reference standard curve. Q-PCRs of experimental and reference samples were performed in triplicate using 7300 system (Applied Biosystem), and data were acquired using 7300 system SDS software version 1.3. PCR product dissociation curve analysis was included for each run as control for amplification specificity. Results were normalized to internal control of each sample amplified with β-actin primers. Primers used in Q-PCR analysis: human β-actin, forward, 5′-tcaccaactgggacgaca-3′, reverse, 5′-ggctggggtgttgaagg-3′; MnSOD, forward, 5′-ttcaataaggaacggggaca-3′, reverse, 5′-gtaagcgtgctcccacaca-3′; GPX1, forward, 5′-ggagaacgccaagaacga-3′, reverse, 5′-cgcaggaaggcgaagag-3′; catalase, forward, 5′-catcgccacatgaatgga-3′, reverse, 5′-gccgcatcttcaacagaaa-3′; GCLC, forward, 5′-ggacaagaatacaccatctcca-3′, reverse, 5′-gcagcactcaaagccataac-3′.
Immunoprecipitation, western blotting and antibodies
To achieve higher transfection efficiency and protein expression, HEK293 cells were used for all the co-immnoprecipitation experiments. HEK293 cells were plated in 6 cm plate at 50% density and transfected with total of 20 µg of plasmids per well using Lipofectamine 2000 (Invitrogen, Carlsbad, CA). Ten micrograms of each effector plasmid (e.g. pcDNA3-PSF) were transfected per well, and pcDNA3 were used as balance plasmid when required. Cells were lysed in non-denaturing lysis buffer containing 1% Triton X-100 24 h after transfection. The procedures for immunoprecipitation and western blotting were described previously ( 6 ). Patient-derived lymphoblasts lysates were kindly provided by Heutink. Antibody used for IP: mouse monoclonal anti-PSF (Sigma, St. Louis, MO). Antibodies for western blotting include: a rabbit polyclonal anti-PGC-1α (Santa Cruz Biotechnology, Santa Cruz, CA, USA), a rabbit polyclonal anti-MnSOD (Stressgen, Ann Arbor, MI), a mouse monoclonal anti-DJ-1 (Stressgen), a rabbit polyclonal anti-DJ-1 ( 6 ), a rabbit polyclonal anti-Flag (Cell Signaling Technology, Danvers, MA), a monoclonal anti-SUMO-1 (Invitrogen) and a goat polyclonal anti-β-actin (Santa Cruz Biotechnology).
Luciferase reporter assay
SH-SY5Y cells were plated in 24-well plates at 70% density. One microgram of pcDNA3-Myc-His-DJ-1, pcDNA3-Flag-PSF, pcDNA3-Flag-PSFK338A mutant, pcDNA4-PGC-1α, pcDNA3-Flag-SENP-1 or pcDNA3-Myc-PIASy or combination of these plasmids as indicated in the figure legends were co-transfected with 100 ng of MnSOD-luciferase reporter plasmid and 100 ng of TK-Renilla luciferase construct (internal control for transfection efficiency). pcDNA3 vector or pcDNA3-GFP was used when necessary to balance the total amount of the effector plasmids in each well to 2 µg/well. Cells were harvested at 24, 36 or 48 h post-transfection in passive lysis buffer (Promega, Madison, WI) as indicated in the legends, and the luciferase activities were determined with Firefly Luciferase Assay System and Renilla Luciferase Assay System (Promega) using Wallac 1420 multilabel counter (Perkin-Elmer, Waltham, MA). The firefly luciferase activity from each sample was normalized to Renilla luciferase activity.
Two-dimensional gel electrophoresis
Human SH-SY5Y cells were lysed in sample lysis buffer (7 m urea, 2 m thiourea, 5% BME and 4% CHAPS). The ZOOM two-dimensional gel system was used following manufacturer's protocol (Invitrogen). Briefly, ZOOM strips (pH 3–10, non-linear) were rehydrated overnight in IPG runner cassette with equal amount of individual protein sample in re-hydration buffer (7 m urea, 2 m thiourea, 2% CHAPS, 2% Ampholytes (pH 3–10) and 0.002% Bromophenol blue, freshly added 10 m m Dithiotheritol). Isoelectric focusing was performed using voltage gradient electrophoresis (200 V, 20 min; 450 V, 15 min; 750 V, 15 min and 2000 V, 30 min). Next, ZOOM strips were equilibrated in 1X NuPAGE sample buffer containing 50 m m DTT for 15 min, and then in 1X NuPAGE sample buffer containing 23.2 mg/ml iodoacetamide for 15 min. Finally, ZOOM strips were loaded onto 4–12% NuPAGE Bis-Tris ZOOM gel for second dimension SDS–PAGE.
Statistical analyses were preformed using InStat 3.0.
This study is supported by a research grant from the Parkinson's Disease Foundation and research fund from the Caritas St Elizabeth's Medical Center.
The authors thank E. Yeh, S. Khochbin and P. Heutink for providing valuable reagents and Lyckman and Shtifman for critical reading of the manuscript.
Conflict of Interest statement . None declared.