PD (Parkinson's disease) is a complex disorder that is associated with neuronal loss or dysfunction caused by genetic risks, environmental factors and advanced aging. It has been reported that DJ-1 mutations rendered neurons sensitive to oxidative damage, which led to the onset of familiar PD. However, the molecular mechanism is still unclear. In the present study we show that DJ-1 interacts with RACK1 (receptor of activated C kinase 1) and increases its dimerization and protein stability. The DJ-1 transgene protects cortical neurons from H2O2-induced apoptosis, and this protective effect is abrogated by knocking down RACK1. Similarly, deletion of DJ-1 in cortical neurons increases the sensitivity to H2O2, and the damage can be significantly rescued by DJ-1 or DJ-1/RACK1 co-transfection, but not by RACK1 alone. We observed further that the interaction of DJ-1 and RACK1 is disrupted by H2O2 or MPP+ (1-methyl-4-phenylpyridinium) treatment, and the protein levels of DJ-1 and RACK1 decreased in neurodegenerative disease models. Taken together, the DJ-1–RACK1 complex protects neurons from oxidative stress-induced apoptosis, with the implication that DJ-1 and RACK1 might be novel targets in the treatment of neurodegenerative diseases.
Known as the second most common progressive neurodegenerative disease, PD (Parkinson's disease) is one type of locomotion disorder of the central nervous system and is characterized by dopaminergic neuronal loss in the substantial nigra and by the deficiency of dopamine in the striatal region [1–3]. Although many clinical cases of PD are idiopathic, at least 5% are caused by mutations in specific genes. These genes, including SNCA (α-synuclein), LRRK2 (leucine-rich repeat kinase 2), PARK2, PINK1 and PARK7, have been confirmed and have provided tremendous insights into the signalling pathways underlying neurodegeneration shown in PD [4–6]. Nevertheless, the precise molecular mechanisms of these gene products in the pathogenesis of PD remain largely unknown.
DJ-1, coded by the gene PARK7, is a ubiquitous protein, highly conserved in all organisms. Deletion or several mutations in the DJ-1-encoding gene are closely linked to autosomal recessive early onset PD, and its up-regulation has been found in many types of tumours [7–10]. DJ-1 has multiple functions ranging from transcriptional regulator  and chaperone with protease activity [12–15], to antioxidant scavenger and redox sensor [16–22]. General consensus agrees that the protective role of DJ-1 results from its ability to counteract oxidative stress. Furthermore, DJ-1 has been found to protect against other toxic agents by modulating the Akt survival pathway [23–25] by interacting with the MAPK (mitogen-activated protein kinase) cascade and the p53 pathway [26–29] or by binding and stabilizing the anti-apoptotic protein Bcl-XL [30,31]. Recently, we report that DJ-1 protects neurons from oxidative stress-induced cell death through promoting Fis1 degradation .
RACK1 (receptor of activated C kinase 1), also known as guanine-nucleotide-binding protein subunit β-2-like 1, is a multi-faceted scaffolding protein that mediates the shuttling of activated protein kinase C to cellular membranes. It has also been demonstrated that RACK1 participates in many signal pathways and plays an important role in multiple cellular functions [33,34]. RACK1 is highly expressed throughout the brain and is involved in various kinds of neurological disorders [35,36]. In addition, it has been reported that RACK1 is involved in neuronal stress responses [37–39]. However, the regulatory mechanism of RACK1 in stress response is largely elusive.
Although it has been shown that the neurotoxicity of oxidative stress could be sensed and abolished by the reductive protein DJ-1, the molecular mechanism of this survival pathway is unclear. In the present study we show that DJ-1 interacts with RACK1 to form a functional complex. This complex plays a very important role in resisting ROS (reactive oxygen species) in cultured cortical neurons. Furthermore, the level of DJ-1 and RACK1 protein dramatically declined in the cellular and mouse models of PD.
MATERIALS AND METHODS
Plasmids, antibodies and RNAi
The plasmids used were as follows: pCMV-3×FLAG-RACK1 and pCMV-3×FLAG-DJ-1 were cloned from the cDNA of human HEK (human embryonic kidney)-293T cells, GFP–RACK1, GFP–DJ-1 and Myc–DJ-1 were cloned from the 3×FLAG tag plasmids. The antibodies used were as follows: polyclonal rabbit anti-GFP (A11122, Invitrogen), monoclonal mouse anti-FLAG (F3165, Sigma), monoclonal mouse anti-Myc (Sc-40, Santa Cruz Biotechnology), polyclonal rabbit anti-DJ-1 (5933P, Cell Signaling Technology), monoclonal mouse anti-RACK1 (610177, BD Biosciences), polyclonal rabbit anti-Akt1 (2938S, Cell Signaling Technology), polyclonal rabbit anti-(Akt Ser473) (4060S, Cell Signaling Technology), monoclonal mouse anti-β-tubulin (CW0098A, CWBiotech), monoclonal mouse anti-β-actin (60008-1-Ig, Proteintech Group) and monoclonal mouse anti-GAPDH (glyceraldehyde-3-phosphate dehydrogenase) (CW0266A, CWBiotech). The hairpin RNA targeting sequence of mouse RACK1 was cacaatggatgggt-aacacag, which was cloned into the PLKO.1-Venus vector.
HEK-293T and COS7 cell lines were cultured in DMEM (Dulbecco's modified Eagle's medium) supplemented with 10% FBS (Gibco), 50 units/ml penicillin and 50 μg/ml streptomycin, in 5% CO2 at 37°C. Cells were transfected with Vigofect (Vigorous) according to the manufacturer's description.
Animals were maintained in the Animal Care Facility at our institute, and all experiments involving animals were approved by Institutional Animal Care and Use Committee at the Institute of Biophysics, Chinese Academy of Sciences.
CHX (cycloheximide) was purchased from Beyotime Institute of Biotechnology. Hydrogen peroxide and MPP+ (1-methyl-4-phenylpyridinium) were purchased from Sigma–Aldrich.
Primary cortical neuron culture
The cortical neurons were dissected from embryos at day 14. At 72 h after initial plating, transfections were performed using the calcium phosphate method. At 24 h after transfection, or 72 h for RNAi experiments, cells were left untreated or treated with H2O2 for 16 h and then subjected to immunocytochemistry assays. To measure the apoptosis of transfected neurons, GFP-positive neurons were counted by investigators blinded to the experimental conditions and cells with condensed nuclei were considered to be apoptotic neurons (Figure 3).
Immunoprecipitation and Western blotting
Cells were washed once with PBS and lysed in buffer A, consisting of 50 mM Tris/HCl (pH 7.5), 150 mM NaCl, 1 mM EGTA and 0.5% Nonidet P40 with protease inhibitor mixture (Roche). Lysates were centrifuged at 20000 g for 10 min. For IP (immunoprecipitation), 200 μg of cell extract was incubated with 1 μg of primary antibody and 30 μl of Protein G–Sepharose beads for 1 h at 4°C. The beads were washed three times in Buffer A. The samples were resolved by SDS/PAGE, generally using Tris/Tricine gels, and were detected by Western blotting. The lane intensity was scanned and quantified by ImageJ (NIH).
Freshly fixed neurons were washed with PBS three times and blocked with 20% goat serum in PBS containing 0.2% Triton X-100 for 1 h at room temperature. Neurons were then incubated with the primary antibody at 4°C overnight. After washing with PBS three times, Alexa Fluor® 488- or 546-conjugated secondary antibody (Invitrogen) was used to detect the signal. The secondary antibody was incubated for 1 h at room temperature, and then nuclear morphology was visualized using Hoechst 33258 (Sigma) .
Statistical analysis of the data was performed with a two-tailed Student's t test, or one-way ANOVA followed by Fisher's PLSD (protected least-squares difference) post-hoc test using Origin software (Version 8). Data are presented as the mean±S.E.M. except for analyses of apoptosis assays, and the number of experiments is indicated in each Figure. *P<0.05, **P<0.01 and ***P<0.001 or NS (no significance) denotes statistical significance.
DJ-1 interacts with RACK1 in primary cortical neurons
To investigate the members of the DJ-1 complex, we first cloned the human gene PARK7 into the pOZ-Flag-HA vector, then established the stably transfected HeLa-S (suspension) cell line. The cells were large-scale cultured and cell lysates went through tandem affinity purification followed by MS analysis. We discovered RACK1 in the DJ-1 complex (Figure 1A). Next, to confirm the interaction between DJ-1 and RACK1, we co-expressed 3×FLAG–RACK1 and GFP–DJ-1 in HEK-293T cells and found that DJ-1 immunoprecipitated with RACK1 (Figure 1B). We also observed that endogenous DJ-1 interacts with RACK1 in HEK-293T cells (Figure 1C). To confirm the result in neurons, we cultured cortical neurons from embryonic day 14 C57/B6 mice embryos and found that DJ-1 interacts with RACK1 in neurons (Figure 1D). Additionally, immunostaining showed that DJ-1 co-localized with RACK1 in the soma and neurites of cortical neurons (Figure 1E). Taken together, DJ-1 and RACK1 physically form a complex in cell lines and primary cortical neurons.
DJ-1 interacts with RACK1
It has been shown that DJ-1 and RACK1 could form dimers [12,41–43]. To investigate the DJ-1–RACK1 interaction further, we detected the interaction of RACK1 with different DJ-1 cysteine to serine residue mutations, which were made on the basis of the prediction that DJ-1 and RACK1 interact (Zdock analysis followed by Rosetta dock) [44,45]. We found that some of the DJ-1 cysteine residue mutations could reduce the interaction with RACK1, indicating that DJ-1 forms a complex with RACK1 possibly through Cys46 and Cys106 (Figure 1F).
DJ-1 enhanced the stability of RACK1 and promoted RACK1 dimerization
The interaction between DJ-1 and RACK1 usually affects the protein stability of the complex members [46,47]. To test this possibility, the same amount of RACK1 was co-transfected with increasing amounts of DJ-1 in HEK-293T cells, and we found that the RACK1 protein level positively correlated with DJ-1 expression level when the DJ-1/RACK1 ratio is up to 1:1 (Figure 2A). To examine further the stability of RACK1, we co-expressed 3×FLAG–RACK1 and GFP–DJ-1 (or GFP vector as control) and treated cells with CHX followed by immunoblotting. We found that DJ-1 dramatically increased the protein stability of RACK1 (Figure 2B).
DJ-1 enhances the stability of RACK1 protein and promotes its dimerization
It has been shown that DJ-1 or RACK1 forms dimers in cells. To investigate whether this interaction altered the protein dimerization, we transfected differently tagged DJ-1 or RACK1 in cells followed by IP and immunoblotting. To make sure the protein quantity was the same, we chose the transfected DJ-1/RACK1 ratio of 0.5:1. The results showed that DJ-1 expression enhanced the homodimers of RACK1 (Figure 2C), whereas dimerization of DJ-1 was not affected by RACK1 (Figure 2D). Taken together, our data suggest that DJ-1 not only stabilizes RACK1, but also enhances the RACK dimerization.
DJ-1 protected cortical neurons from H2O2-induced apoptosis through the RACK1/Akt pathway
We next investigated the biological function of the DJ-1–RACK1 interaction. In our previous study, we confirmed that DJ-1 transgenic mice mitigated dopamine neuron loss induced by MPTP (1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine) . Consistently, we found that the DJ-1 transgene protects cortical neurons from H2O2-induced apoptosis. Interestingly, knockdown of RACK1 abrogated the protective function of the DJ-1 transgene (Figure 3A). Furthermore, re-expression of DJ-1 into the cortical neurons from DJ-1-knockout mice could reduce H2O2-induced cell death (Figure 3B). Overexpression of RACK1 alone failed to protect neurons from H2O2-induced apoptosis, whereas simultaneous expression of RACK1 and DJ-1 confers the protective function. Together, these data indicate that DJ-1 and RACK1 function in a mutually dependent manner in neuronal protection.
DJ-1 protects cortical neurons from apoptosis induced by H2O2
To investigate the relation of DJ-1 and RACK1 as well as DJ-1/RACK1 signalling in vivo, we found that the protein level of RACK1 was up-regulated and the phosphorylation level of Akt Ser473 was significantly increased in DJ-1 transgenic mice (Figure 3C). Consistently, the activity of Akt1 and protein level of RACK1 markedly decreased in the brain tissue of DJ-1-knockout mice (Figure 3D). Together with in vitro data, these experiments suggested that the DJ-1/RACK1/Akt1 signalling pathway existed in vivo and played an important role in neuron survival.
ROS disrupts the DJ-1–RACK1 complex
To investigate further the effect of the oxidative stress on the DJ-1–RACK1 complex, we applied H2O2 or MPP+ to the cultured cortical neurons. We observed that the interaction of DJ-1 and RACK1 was decreased upon treatment with H2O2 (Figure 4A) or MPP+ (Figure 4B). These experiments suggested that ROS had an influence on the interaction of DJ-1 and RACK1.
H2O2 and MPP+ disrupt the interaction of DJ-1 and RACK1
DJ-1 and RACK1 were correlatively decreased in the cellular and genetic mouse neurodegenerative disease models
Usually the intracellular microenvironment is kept reduced under normal conditions. However, advanced aging, genetic mutations or long-term environmental exposure could accelerate the production of ROS, which activates pro-apoptotic signalling or inhibits pro-survival signal and finally leads to cell death [48–51]. To explore whether oxidative stress affected DJ-1 and RACK1, we treated SH-SY5Y cells with MPP+ for 12 h and found that the protein levels of DJ-1 and RACK1 were both reduced upon oxidative stress treatment (Figure 5A). Similarly we observed that both DJ-1 and RACK1 protein levels in prefrontal cortex dramatically decreased in α-synuclein transgenic mice (Figure 5B). Taken together, these data suggested that oxidative stress reduced expression levels of DJ-1 and RACK1 and indicated that the down-regulation of DJ-1/RACK1 might be involved in the pathogenesis of PD.
The protein levels of DJ-1 and RACK1 decline in neurodegenerative models
In the present study, we discovered a novel physical partner of DJ-1, which takes part in defying oxidative damage in neurons. Our findings indicate that DJ-1 and RACK1 work as a complex to mediate the activation of the survival protein Akt1, to counteract oxidative-stress-induced neuronal cell death. We also discovered that extracellular oxidative stress disturbs the DJ-1–RACK1 complex. In neurodegenerative models, the protein levels of DJ-1 and RACK1 remarkably decline. Taken together, the results show that the DJ-1–RACK1–Akt1 signalling pathway plays a key role in cellular responses to oxidative stress in the mammalian central nervous system.
The direct molecular link between intracellular signalling and DJ-1 protection has long been a major enigma in the neuronal apoptosis research field. Initially, it was reported that DJ-1 interacted with Daxx in the nucleus, preventing it from translocating to the cytoplasm, and binding to and activating apoptosis signal-regulating kinase 1 [52–54]. Other studies showed that DJ-1, Parkin and PINK1 formed a complex and co-localized to mitochondria to maintain its functional morphology [55–60]. However, DJ-1 is widely distributed in the neuron (Figure 1D), so the function of the majority of DJ-1 not located in the mitochondria is unknown. In the present study, RACK1, one of many multifunctional adapters, forms a complex with DJ-1 to promote neural survival in the cytoplasm. This complements the function of DJ-1 in neurons to some extent.
A recent study suggested that dopamine-derived quinones affect the dimerization and the redox sensor of DJ-1 through modifications at Cys106 and Cys53 . The oxidation of reductive residues on DJ-1 affected its structure and interaction. Other studies showed that DJ-1 L166P decreased its ability of dimerization and cell protection, indicating that the protective effect depended on DJ-1 dimerization [62–66]. In the present study, ROS undermine the DJ-1–RACK1 heterodimer or multimeric complex and the complex relies on some of the reductive residues of DJ-1, such as Cys106 and Cys53 (Figure 1F). When DJ-1 or RACK1 is absent, cortical neurons become more sensitive to H2O2 (Figures 3A and 3B). Interestingly, overexpressed RACK1 in absence of DJ-1 fails to prompt neuronal survival (Figure 3B), indicating that DJ-1 plays a critical role in RACK1′s neuronal protection under oxidative stress. Furthermore, the DJ-1–RACK1 interaction was largely disrupted by ROS (Figure 4), which is similar to DJ-1 cysteine mutations (C46S and C106S). Taken together, we conclude that DJ-1 and RACK1 form a signalling complex for neuronal protection and this complex is disrupted by excessive ROS, thus leading to neuronal cell death.
In summary, the present study defines a critical oxidative stress pathway in neurons mediated by DJ-1/RACK1/Akt1 signalling. Since DJ-1 mediates the protection from oxidative triggered neural apoptosis, our findings raise the possibility that the DJ-1–RACK1 complex might contribute to pathological states including neurodegenerative diseases. Elucidation of the DJ-1/RACK1/Akt1 signalling pathway as a key mediator of oxidative-stress-induced neuronal cell death proposes an important question of whether activating this signalling pathway might provide a new therapeutic programme for neurodegenerative diseases.
Jun Ma carried out most of the experiments. Rong Wu and Qiang Zhang participated in constructing some of the plasmids. Junbing Wu performed primary neuron culture. Rong Wu raised the DJ-1 transgenic mice. Jizhong Lou performed bioinformatics analysis. Zengqiang Yuan, Zheng Zheng and Jianqing Ding conceived the study. Jun Ma and Zengqiang Yuan wrote the paper.
We thank Dr Zhouhua Zhang at Xiangya School of Medicine, Central South University for providing us the α-synuclein transgenic mice.
This work was supported by the Ministry of Science and Technology of China [grant numbers 973-2012CB910701 and 2013DFA31990 (to Z.Y.)] and the National Science Foundation of China [grant numbers 81125010 and 81030025 (to Z.Y.)].