|Home | About | Journals | Submit | Contact Us | Français|
Mutations in DJ-1, PINK1 (PTEN-induced putative kinase 1) and parkin all cause recessive parkinsonism in humans, but the relationships between these genes are not clearly defined. One event associated with loss of any of these genes is altered mitochondrial function. Recent evidence suggests that turnover of damaged mitochondria by autophagy might be central to the process of recessive parkinsonism. Here, we show that loss of DJ-1 leads to loss of mitochondrial polarization, fragmentation of mitochondria and accumulation of markers of autophagy (LC3 punctae and lipidation) around mitochondria in human dopaminergic cells. These effects are due to endogenous oxidative stress, as antioxidants will reverse all of them. Similar to PINK1 and parkin, DJ-1 also limits mitochondrial fragmentation in response to the mitochondrial toxin rotenone. Furthermore, overexpressed parkin will protect against loss of DJ-1 and, although DJ-1 does not alter PINK1 mitochondrial phenotypes, DJ-1 is still active against rotenone-induced damage in the absence of PINK1. None of the three proteins complex together using size exclusion chromatography. These data suggest that DJ-1 works in parallel to the PINK1/parkin pathway to maintain mitochondrial function in the presence of an oxidative environment.
Although the mechanism of neurodegeneration in Parkinson's disease (PD) is not well understood, mitochondria have been proposed to play a role in the disorder. For example, toxins 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) and rotenone, which target complex I of mitochondria, can recapitulate the nigral dopamine neuron loss characteristic of PD (1).
A genetic component of PD is increasingly recognized (2). Three genes parkin, DJ-1 and PINK1(PTEN-induced putative kinase 1) cause early-onset recessive parkinsonism and all three proteins associate with mitochondria, at least in some circumstances. PINK1 is a mitochondrial serine/threonine kinase the kinase domain of which faces the cytoplasm on the outer mitochondrial membrane (3). Parkin is an E3 ubiquitin ligase that is recruited to mitochondria after loss of polarization of the mitochondrial membrane (4). DJ-1 is relocalized to mitochondria under conditions of oxidative stress (5). All three genes also functionally impact mitochondria. PINK1 (6,7), DJ-1 (5,8–10) and parkin (11,12) influence the sensitivity of dopamine neurons to toxins such as rotenone or MPTP. Loss of dopaminergic neurons does not occur in knockout mice, but there is evidence of altered mitochondrial function (13,14).
As well as this convergent impact on organelle function, PINK1 and parkin are in the same genetic pathway. Parkin- and PINK1-deficient flies have similar mitochondrial phenotypes and although parkin overexpression can rescue PINK1 deficiency, the reverse is not true, indicating that PINK1 is genetically upstream of parkin (15–18). Studies in mammalian cells confirm that PINK1 acts upstream of parkin (19). PINK1 is required for the recruitment of parkin to depolarize mitochondria prior to selective clearance of the functionally inactive organelles by autophagy (20–23). DJ-1 cannot rescue phenotypes resulting from lack of PINK1 or parkin (16,19), suggesting that DJ-1 works upstream of PINK1/parkin or in an independent pathway but the exact relationship has not been resolved.
Here, we have explored the mitochondrial phenotypes of human neuroblastoma cell lines deficient in DJ-1. We find that although loss of either DJ-1 or PINK1 produces similar phenotypes in cells, the three genes likely work in independent parallel pathways.
M17 human dopaminergic neuroblastoma cell lines stably expressing short hairpin RNA (shRNA) against DJ-1 (24) and nonsense shRNA controls selected and cloned in parallel were used to characterize the effects of loss of DJ-1 on mitochondrial function. After confirming DJ-1 protein was lowered (Fig. 1A), we used two different experimental approaches to estimate mitochondrial membrane potential (Δψm) in a clone that we confirmed had lower DJ-1 protein (Fig. 1A). Using flow cytometry, the proportion of cells with low tetramethylrhodamine ethyl ester (TMRE) staining was greater in DJ-1-deficient cells than in controls (Fig. 1B). This was confirmed using live cell confocal microscopy, revealing that DJ-1-deficient cells had lower TMRE staining in mitochondria than control cells (Fig. 1C and D). Collectively, these results show that loss of DJ-1 results in lower Δψm.
Decrease in Δψm can result in changes in mitochondrial morphology. Using live cell confocal microscopy of yellow fluorescent protein (YFP)-labeled mitochondria, DJ-1-deficient cells had subtle morphological differences with an accumulation of fragmented mitochondria (Fig. 1E). To investigate this phenomenon, we used fluorescence recovery after photobleaching (FRAP), to quantify mitochondrial connectivity (Fig. 1F). Summarizing the FRAP curves by calculating the mobile fraction of mitochondrially directed YFP (mito-YFP) demonstrates that mitochondria in DJ-1-deficient cell lines have significantly lower functional connectivity (Fig. 1G), which is similar to observations of PINK1-deficient cells (7).
Previous studies have shown that stable knockdown of PINK1 also results in mitochondrial defects (7,19,25) and the accumulation of markers of autophagy (25), a process by which depolarized mitochondria are selectively removed from the cell (26). MAP-LC3 (microtubule-associated protein 1 light chain 3, denoted as LC3 in the remainder of the text) is a major constituent of the autophagosome, which sequesters the target for degradation by fusing with lysosomes. To date, LC3 is the only reliable marker of autophagosomes. We used accumulation of LC3-GFP as a marker of autophagy and found that DJ-1-deficient cells had higher numbers of LC3-GFP-positive punctatae than controls, which were often near or adjacent to mitochondria (Fig. 1H and I). Fully depolarizing mitochondria with carbonyl cyanide 3-chlorophenylhydrazone (CCCP) caused further increase in the number of LC3-GFP-positive structures to a similar extent in both control and DJ-1-deficient cell lines (Fig. 1H and I), suggesting that the accumulation of this autophagic marker is a consequence of loss of Δψm.
To be certain that these phenotypes were due to loss of DJ-1, we performed a series of control experiments. First, we confirmed that the phenotypes of lower connectivity, Δψm, and accumulation of markers of autophagy were seen in an independent DJ-1 shRNA clone (Supplementary Material, Fig. S1). Second, expressing mouse DJ-1, which is highly conserved but resistant to the human DJ-1 shRNA, reversed both the FRAP defects and accumulation of autophagic markers (Fig. 2). Finally, mouse embryonic fibroblast (MEF) lines from wild-type and DJ-1-/- animals also had lower Δψm and decreased mito-YFP FRAP (Figs 4 and and7B).7B). Collectively, these results demonstrate that DJ-1 deficiency is associated with mitochondrial dysfunction in two independently generated cell models.
DJ-1-deficient cells have increased levels of oxidative stress (27–29). DJ-1 binds RNA for glutathione metabolism genes (24) and glutathione status is influenced by DJ-1 expression (30). Therefore, we examined the role of oxidative stress in this model, specifically focusing on glutathione. We confirmed that DJ-1-deficient cells showed increased oxidative stress using DFFDA (5-(and-6)-carboxy-2′,7′-difluorodihydrofluorescein diacetate) fluorescence which could be rescued by glutathione ethyl ester (GSH-EE; Fig. 3A), a cell-permeable precursor of glutathione. We next addressed whether GSH-EE would recover the mitochondrial phenotypes seen in the same cells. Using both flow cytometry (Fig. 3B) and live cell imaging (Fig. 3C) of TMRE staining, GSH-EE fully recovered the Δψm in DJ-1-deficient cells to the level measured in control cells. Additionally, adding GSH-EE improved mitochondrial morphology (Fig. 3D) and increased mobile fraction in FRAP experiments (Fig. 3E) in DJ-1-deficient cells but had no effect in control cell lines. Finally, the increased LC3 punctae in DJ-1 shRNA lines were also reversed by the addition of GSH-EE (Fig. 3F). The enhanced oxidative stress (Fig. 4A) and rescue of TMRE staining were confirmed in DJ-1-/- MEF lines (Fig. 4B and C). Together, these data suggest that endogenous oxidative stress induced by lack of DJ-1 accounts for mitochondrial phenotypes.
We were interested to understand whether loss of DJ-1 was related to mitochondrial dynamics, which are reported to be sensitive to oxidative stress. We used mitochondrially directed photoactivatable green fluorescent protein (GFP) to monitor fusion rates, which are proportional to loss of signal over time (Fig. 5A). Quantification showed that DJ-1-deficient cells had lower mitochondrial fusion (Fig. 5B). This predicts that increasing rates of fusion or limiting fission should reverse mitochondrial phenotypes in the DJ-1-deficient cells. We overexpressed the fusion proteins Opa1 and Mfn1 or a dominant negative mutant of the fission GTPase Drp1 (Drp1 K38A) in control or DJ-1-deficient cells and confirmed expression by western blots (Supplementary Material, Fig. S3A). These constructs, especially Drp1 K38A, resulted in mitochondrial fusion (Fig. 5C). Either increasing fusion or blocking fission reversed the loss of mitochondrial connectivity in DJ-1-deficient cells (Fig. 5D). These results suggest that the observed lack of mitochondrial fusion underlies the morphological phenotypes in DJ-1-deficient cells. We also found that DJ-1-deficient cells had lower phosphorylated Drp1 (Supplementary Material, Fig. S3B and C), and therefore may have an increased tendency for mitochondrial fission as Drp1 is known to be regulated by phosphorylation (7,31,32).
Previous work has shown that mitochondrial dynamics can influence the autophagic pathway and catabolism of mitochondria (33), so we examined whether fusion defects affected autophagy markers in the same cells. Expression of Drp1 K38A in control cells produced an increase in the number of LC3-GFP-positive structures per cell. However, the opposite effect was seen in DJ-1-deficient cells (Supplementary Material, Fig. S4). These results suggest that some of the effects seen with markers of autophagy may be downstream of alterations in fusion.
We also examined whether autophagy might contribute to the loss of mitochondrial connectivity in these cells. We exposed cells to the bafilomycin A1, which inhibits vacuolar H+-ATPase and thus blocks lysosomal acidification and prevents removal of autophagosomes by fusion with lysosomes. We counted LC3-GFP punctatae, which increased as expected when degradation of autophagosomes was blocked (Supplementary Material, Fig. S5A and B). The same treatment decreased mobile fraction values for mito-YFP (Supplementary Material, Fig. S5C). Therefore, mitochondrial turnover by autophagy is required to maintain mitochondrial connectivity.
To further examine the progression of autophagy (autophagic flux), we monitored the conversion of LC3-I to LC3-II by immunoblotting for LC3. The amount of LC3-II correlates with the number of autophagosomes, and the characteristic conversion of LC3 can be used to examine autophagic activity in the presence or absence of lysosomal activity (34). To verify the increase in LC3-GFP-positive punctatae that was observed with the loss of DJ-1, we transiently transfected LC3-GFP in control and DJ-1 shRNA cells and immunoblotted for LC3. In comparison to the control shRNA cells, DJ-1 silencing promotes more LC3 lipidation under basal and CCCP conditions (Supplementary Material, Fig. S2A) which is indicative of increased autophagy (34). It was determined that lysosomal inhibition increased the number of LC3 punctatae (Supplementary Material, Fig. S5A and B) and LC3-II protein (Supplementary Material, Fig. S2B), which shows efficient autophagy in control cells. Moreover, the amount of LC3-II protein increased in the presence of bafilomycin in DJ-1-deficient cells (Supplementary Material, Fig. S2B) which also suggests an increase in autophagy.
The above results show that endogenous DJ-1 provides protection against oxidative stress and that one of the outputs of loss of DJ-1 is impaired mitochondrial function. To test this further, we increased DJ-1 expression by transient transfection and then exposed cells to oxidative stress by adding rotenone to the cells. Rotenone caused loss of Δψm as estimated by TMRE staining (Supplementary Material, Fig. S6A). We also found that this loss of Δψm was associated with increased numbers of LC3-GFP-positive punctatae in cells, again near mitochondria, as was seen for CCCP (Supplementary Material, Fig. S6B and C). These results show that rotenone produces loss of Δψm and activates the autophagy machinery.
Exposure to rotenone resulted in shortened, disconnected mitochondria but this was minimized if DJ-1 was added (Fig. 6A). FRAP of mito-YFP showed decreased mitochondrial connectivity in cells with vector alone after rotenone exposure but no loss of FRAP signal after rotenone exposure in cells overexpressing DJ-1 (Fig. 6B and C). These results confirm the ability of DJ-1 to protect against oxidative damage.
We have previously reported that stable expression of PINK1 in human cells also protects against loss of FRAP signal after exposure to rotenone and that loss of endogenous PINK1 has the opposite effect (7). We confirmed the same effect in a transiently overexpressed PINK1 system. Transient transfection of human wild-type PINK1 did indeed limit mitochondrial fragmentation after exposure to rotenone (Fig. 6D) and, furthermore, parkin shared this ability (Fig. 6E). These results confirm that DJ-1 (35) and PINK1 (7) protect mitochondria against rotenone-induced fragmentation and extend this to all three known genes for autosomal-recessive parkinsonism.
Given the shared and similar effects of DJ-1, PINK1 and parkin, we evaluated the relationships between their respective contributions to mitochondrial protection. We transiently transfected control or DJ-1-deficient M17 cells with myc-tagged parkin or V5-tagged PINK1. DJ-1-deficient cells had lower mitochondrial connectivity, but increased expression of PINK1 or parkin was able to fully rescue this phenotype (Fig. 7A). We confirmed these results in DJ-1-/- MEF lines (Fig. 7B), showing that in two independent contexts, the pink1/parkin pathway can protect against DJ-1 deficiency.
This places PINK1/parkin either downstream of DJ-1 or in a parallel pathway. We next examined PINK1-deficient cells that, as previously demonstrated (7), have lower mitochondrial connectivity (Fig. 7C). Mitochondrial connectivity was further decreased after exposure to rotenone but rescued by DJ-1 in either control or PINK1-deficient cells. This shows that DJ-1 is active against rotenone-induced damage irrespective of PINK1 expression. Therefore, DJ-1 is not upstream of PINK1/parkin but works in parallel to this pathway.
It has been suggested that DJ-1, PINK1 and parkin form a complex with E3 ligase activity (36). We used size exclusion chromatography to distinguish native complexes of endogenous DJ-1 with overexpressed PINK1 and YFP-tagged parkin. Under these conditions, PINK1, YFP-parkin and DJ-1 were separated into distinct complexes (Fig. 8A and C). The peak of DJ-1 was similar to the molecular weight of a dimer (45–50 kDa) as previously reported (37) irrespective of whether PINK1 and parkin were co-expressed or not. Parkin was found in a peak of ~110 kDa, similar to results reported elsewhere (38). PINK1 was found in a much higher molecular weight complex distinct from parkin or DJ-1. As mitochondrial depolarization promotes parkin relocalization to mitochondria (4,20–22), we repeated these experiments in the presence of CCCP. We confirmed by microscopy that YFP-parkin relocalized to mitochondria in these cell lines in the same experiments, prior to extracting proteins for size exclusion chromatography (data not shown). The three proteins retained their distinct migration patterns and remained separated (Fig. 8B and D), suggesting that the similar effects of DJ-1, PINK1 and parkin are not mediated through a co-complex.
We have shown that lack of the recessive parkinsonism protein DJ-1 causes several defects in mitochondria related to oxidative stress. Specifically, mitochondria in DJ-1-deficient cells are depolarized and have an increased tendency to fragment, which agrees with other studies (39,40). Furthermore, we defined a common phenotype and relationship between DJ-1 and two other genes for recessive parkinsonism, PINK1 and parkin.
Numerous reports demonstrate that the lack of DJ-1 increases sensitivity to reactive oxygen species (ROS) and mitochondrial complex I inhibitors (5,9,41). In response to oxidative stress, DJ-1 can upregulate glutathione synthesis (30), which may be related to the ability of DJ-1 to interact with mRNA transcripts related to glutathione metabolism (24). Supporting the concept that ROS metabolism is important in mediating loss of DJ-1 phenotypes, increasing intracellular glutathione fully rescued the mitochondrial phenotypes. We have previously shown that the ability of DJ-1 protein to respond to oxidative stress is critical for maintenance of mitochondrial morphology (35). These results suggest that DJ-1 influences mitochondrial function and morphology through an oxidative stress pathway.
The two principal mitochondrial phenotypes seen in these DJ-1-deficient cells, namely loss of membrane potential and an increased tendency to fragment, are similar to observations made for PINK1 knockdown (7,25,42). For both PINK1 and DJ-1, the morphological effects can be rescued by blocking mitochondrial fission. However, the fusion effects may be secondary to other mechanisms. For example, in PINK1-deficient cells, calcineurin inhibitors can block the tendency for mitochondrial fragmentation, whereas loss of Δψm cannot (7), and thus fragmentation is a secondary event.
In a previous report in mammalian cells, PINK1 deficiency triggers increased fragmentation of mitochondria, which can be rescued by parkin but not by DJ-1 (19). Because parkin also rescues the DJ-1 deficiency phenotype in this study, we might conclude that DJ-1 is upstream of PINK1/parkin. However, because we see that DJ-1 is still active at preventing loss of mitochondrial connectivity induced by rotenone in the absence of PINK1, we suggest instead that DJ-1 and PINK1/parkin are parallel pathways. We do not find evidence for a native co-complex of the three proteins (36), which indirectly supports a parallel relationship.
DJ-1 now shares an additional similarity to PINK1 and parkin, an association with autophagy. Autophagy has been extensively addressed in the field of recessive parkinsonism. Reductions in mitochondrial membrane potential initiate parkin-mediated autophagy (4,25), which is inhibited in PINK1-deficient cells (21,22,43,44). However, our study does not find a complex, whereby PINK1 and parkin physically interact before or after mitochondrial depolarization despite other reports showing such evidence (45).
The association of DJ-1 with autophagy is varied. DJ-1 silencing has resulted in the upregulation (39,46,47) and downregulation (40,48) of autophagy. Our results suggest that the loss of DJ-1 promotes an increase in autophagy, as an increase in LC3-GFP-positive punctatae and LC3-II protein levels are seen. These data indicate that DJ-1 function may either contribute to the regulation of autophagy or mitigate the downstream effects of ROS, as studies have shown that ROS can upregulate autophagy (49–53).
Future studies should address the role of autophagy in mediating damage caused by ROS. It has been suggested that ERK1/2 regulates lysosomal degradation and more specifically, ERK2 kinase activation is necessary for autophagy (25). Recently, decreased ERK2 phosphorylation was noted in the mitochondrial fraction of DJ-1-deficient cells (40) and it was inferred that DJ-1-mediated ERK2 phosphorylation could putatively control autophagic and lysosomal functions. Our data differ from those of that study as we used affinity chromatography to separate phosphorylated proteins from the unphosphorylated whole cellular fraction. Although we did not observe a change in ERK1/2 phosphorylation in DJ-1-deficient cells, it would be worthwhile to examine the phosphorylation status of the mitochondrial pool of ERK1/2.
Additional studies are needed to identify the underlying molecular mechanisms relating PINK1/parkin and DJ-1. We suggest that oxidative stress is a likely common factor as it is important here for DJ-1 function, triggers phenotypes associated with loss of PINK1 (25) and is sufficient to activate parkin (4). If correct, this suggests that mitochondrial function under oxidative conditions is a determinant of neuronal survival in recessive parkinsonism.
Clonal M17 cell lines stably expressing DJ-1 shRNA constructs were grown and cultured as previously described (24). Expression plasmids for parkin (54) and PINK1 (55) have been described previously. The following plasmids were obtained from Dr Richard Youle (NINDS, Bethesda, MD) and were transfected into cells using Lipofectamine 2000 (Invitrogen): mito-YFP, pA-mito-GFP, Drp1-K38A-myc, Opa1-myc, Mfn1-FLAG (7) and LC3-GFP (56).
For Δψm measurements, cells were cultured in chambered coverglass (Nunc), washed once with phosphate-buffered saline and stained for 30 min at 37°C with 100 nm TMRE (Invitrogen). Images were captured using an inverted confocal microscope (LSM510, Zeiss) with a rhodamine filter (λexcitation 554 nm, λemission 576 nm). Exposure to 10 μm CCCP (Sigma) for 1 h was used as a control for mitochondrial depolarization.
For ROS measurements, cells were stained with 10 mm DFFDA (Invitrogen) and DAPI (4',6-diamidino-2-phenylindolez) for 40 min at 37°C. Images were captured using an inverted confocal microscope and a FITC filter set λexcitation 492–495 nm and λemission 517–527 nm. Exposure to 100 μm H2O2 for 1 h was used as a positive control for ROS production.
For flow cytometry, cells were trypsinized and collected by centrifugation prior to TMRE (100 nm) and DAPI (1 μg/ml, Invitrogen) staining for 30 min at 37°C. Analysis was performed using the FACSVantage SE (Becton-Dickinson) flow cytometer, and the results were analyzed using CellQuest software. Gating was determined empirically in one experiment and then applied in the same way to all subsequent experiments.
Mitochondrial connectivity and fusion assays were performed as described previously (7). Autophagosome formation was monitored by transfecting cells with LC3-GFP and counting the number of distinct punctatae per cell. Cells were fixed (4% paraformaldehyde, Sigma) and co-stained for TOM20 (Santa Cruz; 1:1000) and, in the case of co-transfection with Drp1-K38A, monoclonal anti-c-myc (Roche 1:200) with appropriate secondary antibodies.
General methods for western blotting for PINK1 expression using total sodium dodecyl sulfate (SDS) cell lysates have been described previously (55). Phosphoprotein enrichment was performed and validated as in previous studies (7,57). Size exclusion chromatography was also performed and calibrated as detailed (58). The following antibodies were used at given dilutions: monoclonal anti-β-actin (Sigma), 1:5000; monoclonal anti-Drp1 (BD Translabs), 1:1000; monoclonal anti-Opa1 (BD Translabs), 1:500; monoclonal anti-DJ-1 (Stressgen), 1:3000; polyclonal anti-LC3B (Novus), 1:1000; polyclonal anti-ERK1/2 (Cell Signal), 1:1000; monoclonal anti-FLAG M2 (Sigma), 1:1000; monoclonal anti-c-myc (Roche), 1:1000; polyclonal anti-Fis1 (Biovision), 1:1000.
This research was supported by the Intramural Research Program of the National Institute on Aging, National Institutes of Health (Project Z01-AG000953). This study was also supported by grants from the Parkinsonfonden, Stohnes foundation, Gamla Tjänarinnor Foundation, NHR foundation and Riksbanken Jubileumfonden, Sweden, The K&A Wallenberg Foundation and the Swedish Brain Power Initiative.
We thank Dr Richard Youle and Megan Cleland for helpful discussions and indicated plasmid constructs.
Conflict of Interest statement: The authors declare that there are no conflicts of interest.