Human Molecular Genetics Advance Access originally published online on February 2, 2005
Human Molecular Genetics 2005 14(6):799-811; doi:10.1093/hmg/ddi074
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Genetic and genomic studies of Drosophila parkin mutants implicate oxidative stress and innate immune responses in pathogenesis
Department of Genome Sciences, University of Washington, Seattle, WA 98195, USA
* To whom correspondence should be addressed. Tel: +1 2066165997; Fax: +1 2065430754; Email: pallanck{at}gs.washington.edu
Received December 3, 2004; Accepted January 18, 2005
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Loss-of-function mutations of the parkin gene, which encodes a ubiquitin-protein ligase, are a common cause of autosomal recessive juvenile parkinsonism (ARJP). Previous work has led to the identification of a number of Parkin substrates that implicate specific pathways in ARJP pathogenesis, including endoplasmic reticulum (ER) stress and cell cycle activation. To test the involvement of previously implicated pathways, as well as to identify novel pathways in ARJP pathogenesis, we are using genetic and genomic approaches to study Parkin function in the fruit fly Drosophila melanogaster. In previous work, we demonstrated that Drosophila parkin null mutants exhibit mitochondrial pathology and flight muscle degeneration. To further explore the mechanisms responsible for pathology in parkin mutants, we analyzed the transcriptional alterations that occur during muscle degeneration and performed a genetic screen for parkin modifiers. Results of these studies indicate that oxidative stress response components are induced in parkin mutants and that loss-of-function mutations in oxidative stress components enhance the parkin mutant phenotypes. Genes involved in the innate immune response are also induced in parkin mutants. In contrast, our studies did not reveal evidence for cell cycle or ER stress pathway induction in parkin mutants. These results suggest that oxidative stress and/or inflammation may play a fundamental role in the etiology of ARJP.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Parkinson's disease (PD) is a common neurodegenerative disorder characterized by the loss of dopaminergic neurons in the substantia nigra and the accumulation of intracellular protein aggregates known as Lewy bodies. Previous work suggests that oxidative stress (1
–3), mitochondrial dysfunction (4–7), aberrant protein degradation (8,9) and inflammation (10,11) may be involved in the pathogenesis of PD. However, much of the previous work on PD pathogenesis is correlative and it remains unclear whether previously implicated factors are causative or consequences of pathogenesis. Although most PD is sporadic, recent work has led to the identification of specific gene defects responsible for simple Mendelian forms of this disorder (reviewed in 12,13). Studies directed at an understanding of the functional roles of these genes present an opportunity to define molecular mechanisms of pathogenesis in PD (14).
Loss-of-function mutations in the parkin gene have been shown to be a major cause of autosomal recessive juvenile parkinsonism (ARJP) (15–17) and increasing evidence suggests that parkin dysfunction may also be involved in late-onset and sporadic PD (17–21). Parkin functions as an E3 ligase in the ubiquitin protein degradation pathway, which has led to the suggestion that accumulation of Parkin substrates results in the death of dopaminergic neurons in ARJP individuals (22). To date, nine substrates of Parkin have been identified, suggesting a variety of hypotheses about the role that Parkin plays in the cell and the etiology of ARJP (reviewed in 23,24). For example, the Parkin substrate Pael-R is a putative G-protein coupled receptor that has a propensity to accumulate in an unfolded form in the endoplasmic reticulum (ER). Overexpression of Pael-R in cultured cells appears to induce an unfolded-protein response (UPR)/ER stress response and ultimately results in cell death. Co-overexpression of Parkin suppresses Pael-R-induced cell death, suggesting that Parkin plays a functional role in the UPR/ER stress response (25). Another Parkin substrate, Cyclin E, promotes entry of cells into the S-phase of the cell cycle. It has been proposed that a defect in Cyclin E degradation resulting from inactivation of Parkin could lead to inappropriate activation of the cell cycle and subsequent apoptosis of dopamine (DA) neurons (26). Additional substrates of Parkin implicate dysfunctional Lewy body metabolism (27,28), altered neurotransmission (29–31) and cytoskeletal defects (32) in ARJP pathogenesis. Although the identification of Parkin substrates is integral to an understanding of ARJP pathogenesis, an important goal of future work is to definitively identify which, if any, of the current Parkin substrates play active roles in ARJP pathogenesis.
To explore the biological role of parkin, we are using genetic approaches to study a Drosophila parkin ortholog. Previous work has shown that Drosophila parkin null mutants display partial pupal lethality, decreased adult lifespan, apoptotic muscle degeneration and male sterility (33,34). Mitochondrial pathology is the earliest cellular manifestation of tissue degeneration in parkin mutants, suggesting a role for Parkin in mitochondrial integrity. To better understand the pathways responsible for muscle degeneration in parkin mutants, we carried out a genetic screen for parkin modifiers and used cDNA microarrays to characterize transcriptional alterations in parkin mutants. Results from these studies suggest that loss-of-function mutations in parkin sensitize cells to the deleterious effects of reactive oxygen species. Furthermore, innate immune response components are a major category of genes that are induced early in the timecourse of muscle degeneration in parkin mutants, suggesting that inflammation may also play a role in muscle degeneration. No clear evidence for pathways implicated by previously identified Parkin substrates was detected in these analyses. Given the substantial circumstantial evidence for involvement of oxidative stress and inflammatory pathways in sporadic PD, our results raise the possibility that similar mechanisms underlie ARJP pathogenesis and sporadic forms of PD.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Transcriptional profile of 1-day-old parkin mutants
To identify transcriptional alterations in parkin mutants, cDNA microarrays were used to compare transcript abundance between 1-day-old adult parkin mutants and control flies. This timepoint was chosen because 1-day-old parkin mutants show clear phenotypes, including muscle degeneration, indicating that transcriptional differences between parkin mutants and controls are likely present (33
). The microarrays used for these experiments contain 
6000 spotted Drosophila cDNAs (Drosophila gene collection 1), 45% of the predicted genes in the genome. Control and mutant RNAs from whole adult flies were labeled, hybridized to the arrays and the ratio of the signals was determined for each spot. After normalization, the Cyber T algorithm (35) was used to identify any spots with significantly altered expression (altered intensity ratios) in parkin mutants relative to controls.
At a stringent false discovery rate of 1% there were 1000 genes with significantly altered expression in 1-day-old parkin mutants relative to controls, representing about one-sixth of the genes on the microarray. Among this large group of altered transcripts, approximately half of the transcripts were decreased in parkin mutants relative to controls and approximately half were increased, though the magnitudes of change for the decreased transcripts were greater. The genes whose transcripts were altered in 1-day-old parkin mutants represent 44% of all annotated biological processes in Drosophila. This large number of changes is likely indicative of the advanced stage in the degenerative process and suggests that many of the transcriptional alterations represent downstream events in pathology. To capture transcriptional alterations earlier in the timecourse of muscle degeneration, another set of experiments was initiated using parkin mutants at an earlier stage in development.
Transcriptional profile of 48 h parkin mutant pupae
To identify a developmental stage that precedes the onset of visible muscle degeneration, parkin mutant pupal flight muscles were examined by transmission electron microscopy (TEM). TEM sections were taken of the indirect flight muscle of parkin mutants and controls at various timepoints after puparium formation (APF). As previously reported, at 96 h APF, parkin mutant pupal flight muscles display mild mitochondrial pathology (33). However, at 48 h APF, the flight muscles in parkin mutants were morphologically indistinguishable from controls (Fig. 1). In particular, the morphology of the developing myofibrils and overall structure and size of the mitochondria, as well as the integrity of the mitochondrial cristae, are indistinguishable between parkin mutants and age-matched controls. Given the morphological similarity of mutants and controls at this developmental stage, transcriptional differences between mutants and controls at this timepoint likely represent an early response to loss of Parkin function. Thus, further transcriptional profiles were obtained at this developmental stage.
|
RNA was extracted from 48 h parkin mutant and control pupae and hybridized to cDNA arrays containing
12 000 cDNA spots (Drosophila gene collections 1 and 2). This represents a large portion of the number of predicted genes in the Drosophila genome, which is 13 700 (reviewed in 36). Hybridizations and statistical analyses were carried out in the same manner as for the 1-day-old adult timepoint, and for this analysis, a false discovery rate of 1% was also used. Twenty-six genes showed altered expression in parkin mutants relative to controls at this developmental stage (Table 1).
|
The majority of genes showing altered expression in parkin mutants were upregulated relative to controls. These upregulated genes could be grouped into two categories including immune-response related (six genes) or oxidative-stress and electron-transport related (four genes). The remaining 11 upregulated genes and the five downregulated genes could not be grouped into a specific category or are of unknown function.
To validate the array findings, the alterations in transcript levels were verified using slot blot analysis for several of the genes found to be altered in parkin mutants. RNA from parkin mutant and control pupae was blotted and hybridized with probes corresponding to selected genes. Signal intensity was quantified in both parkin mutants and controls, and representative results are shown in Figure 2. Good agreement was observed between the difference in transcript levels predicted by the array and that observed using the slot blot apparatus indicating that the results from the array accurately represent differences in transcript abundance.
|
To systematically identify which pathways and processes are altered in parkin mutant pupae, the EASE program (37
) was used to categorize the altered transcripts. The EASE program identifies gene ontology categories that are significantly over-represented among the transcripts that display altered abundance in parkin mutants (see Materials and Methods for details). At a 1% false discovery rate, the most significantly over-represented category of upregulated transcripts is that of immune and defense response, followed by oxidative stress response. These same categories are seen if the false discovery rate is increased further to 5, 10, 15 or 20%. The downregulated transcripts could not be grouped into any significant category, and downregulated genes do not fall into statistically significant categories at any of these false discovery rates.
Analysis of pathways previously implicated in parkin pathology
Previous studies have implicated Parkin in several well-defined processes including ER stress (25) and cell cycle regulation (26). To investigate the involvement of these processes in pathology in Drosophila parkin mutants, the transcriptional alterations in parkin mutants were examined to look for changes in components of these pathways. ER stress is associated with a defined transcriptional profile, which can be examined in the data obtained from parkin mutants. Previous work in yeast has identified 381 genes identified to be upregulated in response to ER stress (38). Of the 381 yeast genes induced in response to ER stress, 188 were found to have Drosophila homologs with cDNA spots represented on the array. At the 48 h pupal stage, the median fold expression change among these ER-stress related homologs in parkin mutants was –1.07, with 81% of these homologs having expression ratios between 1.2 and –1.2, indicating no significant change in transcript levels. In addition, approximately half of these genes showed a slight negative ratio indicating a tendency towards downregulation. This contrasts with the finding that the yeast homologs of these genes are upregulated in response to ER-stress-inducing treatment (38). Similar results were seen upon analysis of these genes in the adult profile data. In addition, Table 2 depicts several of the key mediators of this pathway (39) and demonstrates that none of these components are altered in parkin mutants. These data strongly suggest that the pathology seen in Drosophila parkin mutants is not associated with induction of the ER stress response pathway.
|
The identification of Cyclin E, which is involved in the G1–S transition in the cell cycle, as a putative Parkin substrate provokes an additional hypothesis about pathology in patients with parkin mutations: the inappropriate activation of the cell cycle in post-mitotic cells results in apoptosis (40
). Unlike the ER stress pathway, the precise transcriptional changes that result from Cyclin E activation are not well defined, and the regulation of the cell cycle is dependent on a number of transcriptional and post-transcriptional alterations. Therefore, a survey of many different genes that are known to function in cell cycle activation and progression was undertaken. Results of this analysis show that the average expression of these genes is not changed from controls in parkin mutant pupae (Table 3). In 1-day-old parkin mutant adults, there is evidence for a downregulation of genes involved in cell cycle, which appears contrary to what would be expected from Cyclin E activation. In addition to the transcriptional data, Cyclin E protein levels were found to be indistinguishable between parkin mutants and controls at all developmental stages analyzed, including a stage that precedes muscle pathology and in 1-day-old adults where pathology is prominent (Fig. 3). In addition, Cyclin E levels are not altered in actively dividing cultured Drosophila cells that have reduced parkin expression through the use of RNA interference, despite >90% reduction of Parkin protein levels in these cells (Fig. 3). Thus, it appears that the cell cycle is not activated in the absence of Drosophila parkin function.
|
|
Genetic interactions between parkin and array-identified genes
To test the functional significance of genes identified in the transcriptional profile analysis, genetic interactions between parkin and genes showing altered expression in parkin mutants were examined. In agreement with previously published work, we have found that, under standard laboratory conditions, parkin mutants display a partial lethality phenotype; in a cross of parkin heterozygotes, only
75% of the expected Mendelian numbers of parkin homozygotes are observed (data not shown and 34). The partial lethality appears to be related to muscle dysfunction as the lethality can be rescued by ectopic expression of parkin in muscle (data not shown). Because the partial lethality phenotype can potentially be suppressed or enhanced by mutations in other genes, and is simple to assess, this phenotype was used to test for genetic interactions between parkin and genes identified by transcriptional profile analysis. Available mutations in several genes with significantly altered transcript levels in parkin mutants were obtained and crossed into a parkin background. The percentage of parkin mutants eclosing from a cross of parkin heterozygotes was determined in a genetic background heterozygous or homozygous for alleles of array-identified genes. None of the alleles of the array-identified genes tested exhibit viability defects as heterozygotes or homozygotes in a wild-type parkin background. Therefore, an alteration of parkin mutant viability by these alleles should reflect a genetic interaction of these alleles with parkin. Loss-of-function alleles of the genes GstE1, CG2789, CG12505 and Dgp-1 substantially reduced the viability of parkin mutants (Fig. 4). As loss-of-function mutations of these genes enhance the parkin partial lethal phenotype, it appears that the upregulation of these genes detected in parkin mutants represents a protective response. One additional allele of Dgp-1, as well as an allele of CG32758, was tested in the context of a genetic screen for parkin modifiers (discussed later). These alleles did not have an effect on parkin mutant viability. There was also no reduction in viability of parkin mutants when a Map205 allele was in the heterozygous state, and there was only a slight decrease in viability when this allele was homozygous (Fig. 4). The functional effects of the P element insertions of Dgp-1, CG32758 and Map205 that were without effect on parkin viability are unknown, and the lack of alteration of the viability of parkin mutants by these alleles may reflect weakness of these alleles or a lack of involvement of these genes in Parkin pathogenesis. CG16820 transcript levels are decreased in parkin mutant pupae, raising the possibility that downregulation of CG16820 may be protective to parkin mutants. In accordance with this hypothesis, parkin mutant flies that were heterozygous for an allele of CG16820 displayed increased viability: the percentage of parkin mutants increased to 36% in this genetic background, which corresponds well to the expected Mendelian percentage of 33%. However, making this allele homozygous reduced parkin mutant viability dramatically, which suggests that this modification may be dosage sensitive.
|
The finding that (1) alleles of five out of the six array-identified genes tested have an effect on the partial lethality phenotype, (2) loss of function of upregulated genes enhances the lethality and (3) loss of function of a downregulated gene suppresses lethality strongly supports the involvement of these genes in parkin pathogenesis.
Genetic screen for parkin modifiers
Results from testing genetic interactions between parkin and array-identified genes revealed interactions consistent with expectations based on alterations in transcript levels. This suggested the possibility that a genetic screen for modifiers of the parkin partial lethality phenotype could be used to identify pathways relevant to Parkin pathogenesis. The screen for modifiers was carried out by crossing a collection of P element transposon insertions into a parkin mutant background and scoring for suppression or enhancement of the partial lethality phenotype. The collection of enhancer P (EP) transposon insertion lines (41,42) makes use of a transposon that contains GAL4-responsive promoter sequences at one end which allow overexpression of the genomic sequences adjacent to the transposon insertion. This strategy takes advantage of the propensity of P element transposons to insert into the 5' ends of genes, often in the 5'-untranslated region (UTR) (43). The resulting insertions can either drive overexpression of the downstream genes when crossed to a GAL4-producing strain (if oriented correctly) or create a loss-of-function allele through insertional inactivation. When crossed to GAL4-producing strains, these EP elements can be screened to look for modification of phenotypes of interest. The mesodermal GAL4 driver, 24B-GAL4, which fully rescues the partial lethality when used to drive expression of a parkin cDNA, was used to drive expression from the EP elements in this screen.
The EP insertions used in this analysis are distributed among the X, 2nd and 3rd chromosomes of Drosophila. As the parkin gene resides on chromosome 3, only EP insertions on the X and 2nd chromosomes (1400 total) were used in the screen for modification of partial lethality, which is a parkin loss-of-function phenotype. Although parkin heterozygotes lack a detectable phenotype, the 1000 EP insertions on chromosome 3 were screened to attempt to identify enhancers of parkin that result in a phenotype. This was done by generating parkin heterozygote flies bearing the EP insertion and the GAL4 driver. These flies were compared with flies bearing the EP insertion and GAL4 driver in a wild-type parkin background. All flies were counted to look for any induced lethality and assayed for flight defects. No modifiers were recovered from this haploinsufficiency 3rd chromosome screen.
To screen the insertions on the X and 2nd chromosomes, a two generation crossing scheme was used (see Materials and Methods for details). Offspring bearing the 24B-GAL4 driver, the EP being tested, and two copies of the park25allele (parkin homozygotes) were compared with flies bearing the 24B-GAL4, the EP, and one copy of the park25 allele (parkin heterozygotes) to account for any effects on viability due to expression from the EP. To score modification of the partial lethality, a ratio of parkin heterozygotes to homozygotes was obtained for each EP line. P element insertions found to increase or decrease the number of parkin homozygotes by 2 standard deviations (SD) from the expected outcome were considered modifiers. Those lines whose ratio was greater than that of the mean were considered enhancers as there were fewer parkin homozygous mutant progeny than expected, and those lines whose ratio was less than that of the mean were considered suppressors as more parkin homozygous mutant progeny than expected were produced. In the absence of modification, the ratio is expected to be 1.3 : 1 (parkin heterozygotes: parkin homozygote). To meet the 2 SD requirement, the ratio for enhancers was greater than 2.2 : 1 (parkin heterozygotes: parkin homozygote), whereas for suppressors, the ratio was less than 0.53 : 1 (parkin heterozygotes: parkin homozygote). All putative modifiers recovered from screening were retested at least twice and were required to have met the same degree of modification to be included in the final results.
This screen led to the recovery of a total of 15 genetic modifiers of the partial lethality phenotype (Table 4). These modifiers make up 1.1% of all the lines screened. To test whether the observed modification results from overexpression of flanking genes (gain-of-function enhancers or suppressors) or from insertional inactivation by the EP insertion (loss-of-function enhancers or suppressors), the effects of EP modifiers on the parkin lethality were explored in the absence of a GAL4 driver. For all loss-of-function parkin modifiers recovered from screening, the gene closest to the EP element responsible for genetic modification was assumed to be causally involved in the genetic interaction with parkin. For gain-of-function enhancers, the orientation of the EP element was also evaluated to identify the gene responsible for genetic interaction with parkin. Only genes positioned in the correct orientation to allow overexpression by the P element were considered candidates for the gain-of-function category of modifiers. The results of these analyses are included in Table 4.
|
The identified modifiers represent genes active in a variety of functional pathways. The most striking finding from this study is that all three of the loss-of-function enhancers recovered from the screen involve oxidative stress response components. One of the enhancers in this category (GstS1) had the greatest effect on parkin viability among the modifiers recovered from this screen. The identification of a second allele of GstS1 from this analysis serves as independent confirmation of its modification of parkin reduced viability.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Our current study describes the results of genetic and genomic approaches to explore the biological role of Drosophila parkin. Although our previous work indicates that apoptotic muscle degeneration and spermatid developmental defects are the primary phenotypes associated with loss-of-function mutations of parkin in this organism, there are several reasons to believe that the molecular mechanisms underlying these phenotypes may have similarity to the mechanism of DA neuron degeneration in ARJP individuals. In particular, mitochondrial pathology is a central hallmark of both the muscle and germline defects in Drosophila parkin mutants (33
,34) and recent work indicates that mitochondrial dysfunction is also a feature of ARJP (44), suggesting that the core biological function of Parkin may be conserved in both Drosophila and humans. Although it is presently unclear why muscle and germline tissues are the most sensitive to loss of parkin function in Drosophila, a unique feature of insect flight muscle is its high metabolic activity (45). This high metabolic rate likely results in the significant production of reactive oxygen species, and previous work suggests that DA neurons also produce significant levels of reactive oxygen species (reviewed in 46). Thus, a common feature of insect flight muscle and DA neurons that may render these tissues sensitive to loss-of-function mutations of parkin is the high rate of reactive oxygen species formation. Results of our current study provide support for this hypothesis.
Evidence for oxidative stress in parkin mutants
The major finding from our genetic and genomic studies is that oxidative stress pathways appear to play a role in Parkin pathogenesis. All of the loss-of-function enhancers of parkin recovered from screening participate in detoxification of oxidatively damaged molecules, providing direct functional support for the importance of this pathway in parkin pathogenesis. Oxidative stress response components also represent a major category of genes that are upregulated shortly before overt cellular manifestation of mitochondrial dysfunction in parkin mutants. Although direct functional evidence for the involvement of genes implicated from microarray studies in parkin pathogenesis is currently limited to only a small number of genes, these studies strongly support their functional significance. Loss-of-function alleles in five out of seven genes identified from transcript analyses (71% of the genes tested), including an EP allele of CG32758 that was not recovered in our genetic screen for parkin modifiers, were found to exhibit a genetic interaction with parkin. The genetic interactions observed with these genes were robust and in several cases resulted in synthetic lethality. In contrast, only 1% of an impartial set of lines tested in the genetic screen for parkin modifiers displayed genetic interactions equaling or exceeding the severity of interactions displayed by alleles of the array-identified genes. Furthermore, the genetic interactions observed between parkin and alleles of the array-identified genes indicate that the transcriptional changes reflect compensatory stress responses to loss of parkin function. Alleles of genes that were upregulated in parkin mutants enhanced the parkin lethality phenotype, and an allele of a gene that was downregulated in parkin mutants suppressed the parkin lethality phenotype. Thus, the genes identified from the transcriptional profiling studies appear to be highly enriched for those involved in parkin pathogenesis.
The two loss-of-function enhancers recovered from genetic screening, glutathione S-transferase S1 (GstS1) and thioredoxin, are part of an important biological defense system against oxidative damage (47–49). In addition to GstS1, a loss-of-function allele of another glutathione S-transferase gene, GstE1 also enhanced the parkin partial lethal phenotype, although the magnitude of the genetic interaction was just below the statistical cutoff used to select genetic modifiers. Interestingly, GstE1 is also upregulated early in the timecourse of muscle degeneration in parkin mutants and an allele of GstE1 that was not part of the collection used in the genetic screen displayed a dominant synthetic lethal interaction with parkin. These findings strongly implicate glutathione metabolism in parkin pathogenesis. Previous work has shown that the substantia nigra in PD patients has significantly reduced glutathione levels relative to controls (1–3), and that glutathione S-transferase gene inactivation results in enhancement of cell death in a yeast model of 
-synuclein toxicity (50). Furthermore, in an allelic association study, two glutathione S-transferases were recently found to play a role in modification of age of onset of PD and Alzheimer's Disease (51). Thus, our current findings suggest a possible mechanistic link between ARJP and sporadic PD.
Several genes identified from our studies that are not formally classified as oxidative stress response components also potentially implicate oxidative stress in parkin pathogenesis. For example, the gain of function parkin suppressor RluA-1 contains domains that are involved in the biosynthesis of the antioxidant vitamin B2. Thus, increased expression of RluA-1 may lead to higher levels of vitamin B2 production and enhanced protection from oxidative stress. A recent study has shown that vitamin B2 can have a beneficial effect in patients suffering from PD (52). The Dgp-1 gene, which is also upregulated in parkin mutants, was recently shown to be upregulated in Drosophila that have been challenged with paraquat, an inducer of oxidative stress (53). Dgp-1 is a translation elongation factor with homology to the human GTP-binding-protein 1, which appears to play a role in the cellular defense against oxidative stress (54). Interestingly, the same study that demonstrated Dgp-1 upregulation in response to paraquat also found upregulation of Thor, a translational regulator, following paraquat treatment. Thor mutants were shown to be highly sensitive to paraquat-induced oxidative stress, strongly suggesting a functional role for Thor in the oxidative stress response. Thor was also identified in our genetic screen for parkin modifiers, possibly reflecting the role of this factor in the oxidative stress response. However, our data indicate that Thor overexpression enhances the parkin mutant phenotype so further work will be required to understand the mechanism of Thor involvement in parkin pathogenesis.
There are several potential explanations for the source of oxidative stress in parkin mutants. One possibility is that Parkin is responsible for degrading oxidatively damaged proteins, and in the absence of Parkin their accumulation activates oxidative stress pathways and triggers cellular dysfunction. Drosophila parkin mutants are sensitive to paraquat treatment, a known inducer of oxidative damage [our unpublished data and (34)], and HOIL-1, another E3 ligase that is structurally similar to Parkin, has been shown to degrade an oxidatively modified substrate (55), suggesting that Parkin may function to protect cells from oxidatively damaged molecules. Another potential explanation for the source of oxidative stress in parkin mutants is that Parkin plays a direct or indirect role in mitochondrial integrity. ARJP patients have been shown to have reduced mitochondrial complex I activity (44), and parkin mutant mice exhibit altered abundance of particular mitochondrial components, thus providing support for this model (56). One of the transcripts increased in parkin mutants is that of CG2789. The CG2789 protein is homologous to the mammalian peripheral benzodiazapene receptor (PBR), which has been linked to the regulation of mitochondrial integrity and is thought to be involved in the regulation of mitochondrial swelling, respiration, transmembrane potential and the prevention of oxidative damage to the mitochondria (reviewed in 57). In the absence of Parkin function, mitochondria become dysfunctional, likely resulting in the increased production of reactive oxygen species. The upregulation of CG2789 could represent a cellular attempt to limit this damage. Further genetic and biochemical analysis of the genes recovered in this study is needed to resolve these models.
Activation of the innate immune response in parkin mutants
The other major category of upregulated genes in parkin mutants contains genes involved in the innate immune response. This finding suggests that Parkin may normally play an inhibitory role in the immune response. The innate immune response in Drosophila utilizes many of the same factors and processes known to induce the innate immune response in mammals, including macrophage-like cells that respond to external stimuli and the production of anti-microbial peptides that are released into the circulatory system. The immune-response genes identified in this analysis largely represent these released factors.
Previous work has shown that feeding Drosophila compounds that induce the production of nitric oxide (NO) results in transcriptional activation of immune response genes (58), particularly Diptericin (Dpt), which is one of the most highly upregulated genes in parkin mutants. The mechanism of NO induction of the immune response in Drosophila remains unknown. Parkin's E3 ligase activity has also been shown to be modulated by nitrosylation (21,59), raising the possibility that NO-mediated inhibition of Parkin function may be the mechanism of NO-mediated induction of the innate immune response pathway. Cytotoxic production of NO and other metabolites by activated microglia have been documented in PD cases, and dopaminergic neurons in the substantia nigra appear especially sensitive to toxins produced by these inflammatory response cells (reviewed in 10). A microglial reaction to treatment of MPTP (a chemical that induces DA neuron loss) in damaged areas of mouse brains has also been documented (60–65). In addition, a polymorphism in Interleukin-1ß (IL-1ß) has been identified as a factor for developing PD (11). Thus, our findings suggest the hypothesis that ARJP pathogenesis may involve a dysfunctional immune/inflammatory response.
There are also several alternative interpretations for the induction of the immune response in parkin mutants. While this response may reflect a direct role of Parkin in the innate immune response, the induction of this pathway also occurs in response to oxidative stress and possibly in response to more general cellular stresses. For example, a study examining the transcriptional profiles of aged or oxygen-stressed flies, as compared to controls, detected induction of the immune response along with other stress-response genes under both of these conditions (66). In addition, studies of long-lived mutants of Caenorhabditis elegans have shown that immune response genes and genes involved in protection from oxidative damage are induced upon aging (67–69). One possible mechanism of induction of the innate immune response in parkin mutants involves excess production of NO by dysfunctional mitochondria (reviewed in 70,71). This mode of activation of immune response genes may reflect a normal biological role for these genes in the oxidative stress response, or alternatively, it may represent a passive transcriptional response that is functionally unrelated to the oxidative stress response. Further study of the role of the immune response genes in parkin pathology will allow dissection of these possibilities.
Relevance of our findings to Parkin substrates
Although it is unclear whether Parkin targets substrates for proteolytic degradation, or whether Parkin acts in a ubiquitin-mediated signaling capacity, a commonly posited mechanism of ARJP pathogenesis suggested by the former possibility involves toxic accumulation of Parkin substrates. As the transcriptional profile experiments evaluate mRNA levels and not protein levels, the accumulation of Parkin substrates would not be directly detected by this method; however, Parkin substrates may be represented among the modifiers recovered from the genetic screen. The toxic accumulation model predicts that mutations affecting the abundance of Parkin substrates would modify the parkin mutant phenotypes. In particular, loss-of-function mutations of genes encoding Parkin substrates might be expected to suppress the parkin phenotypes, whereas overexpression gain-of-function mutations might enhance the parkin mutant phenotypes. These predictions suggest that the mutations recovered from genetic screening that fall into these two categories potentially represent Parkin substrates.
None of the modifiers identified from this screen corresponds to substrates previously identified for Parkin. There are EP alleles of two genes that are homologous to previously identified substrates, CycE and ßTub56D: however, the EP alleles of these genes were not identified as modifiers in this analysis. This could be because these genes are not involved in pathogenesis or because the EP insertions do not significantly affect the functions of the corresponding genes. For the other previously identified Parkin substrates, there are either no clear Drosophila homologs or the homologs are not represented in the EP collection. Further work will be required to evaluate the involvement of Drosophila homologs of other previously identified Parkin substrates in parkin pathogenesis.
Although none of the genes recovered from genetic screening encodes previously identified Parkin substrates, there are domains present in some of these proteins that are found in Parkin substrates. For example, previous work has shown that Parkin interacts with its substrate Synaptotagmin XI by binding the C2 domains of this protein (30). CG14045, a loss-of-function suppressor of parkin identified in our work, also appears to have a C2 domain, raising the possibility that this protein is a direct substrate of Parkin. CG14045 also has a PDZ domain which is also found in the Parkin-interacting protein CASK (72). However, CG14045 and many other genes recovered from genetic screening encode proteins of unknown function and further work will be required to understand how mutations in these genes influence parkin pathogenesis. These genes of unknown function potentially represent novel pathways that may impinge on disease progression and thus may reveal novel therapeutic targets.
Relationship of our findings to other models of PD
Transcriptional profile analyses have also been performed in several other models of PD. Studies examining the transcriptional profiles of MPTP-treated mice show that transcriptional alterations involve genes associated with, among other things, oxidative stress, NO metabolism and inflammatory processes (73–75). Treatment of a dopaminergic cell line with MPTP induces the intracellular accumulation of reactive oxygen species and oxidatively modified proteins (76). In addition, the genes identified upon microarray analysis of a transgenic alpha-synuclein Drosophila model of PD (77) display some general similarities to the genes identified in our studies of parkin mutants. alpha-synuclein-expressing flies show upregulation of several cytochrome P450 genes, other genes related to energy metabolism and Tetraspanin42E (Tsp42E), which encodes a membrane protein of unknown function. Tsp42E was identified in this study as a genetic modifier of parkin lethality. Although these models differ from Drosophila parkin mutants, it is intriguing that similar pathways in these different PD models are implicated in pathogenesis.
Our work suggests the involvement of oxidative stress and immune system dysfunction in Parkin pathogenesis, however, we did not detect evidence for pathways previously implicated in Parkin pathogenesis. In particular, none of the genetic modifiers of the parkin loss-of-function phenotype implicates ER stress or cell cycle activation. Analysis of transcriptional profile data from parkin mutant pupae also failed to detect activation of genes related to either of these pathways. Finally, we were unable to detect alterations in the abundance of Cyclin E, a substrate of vertebrate Parkin, in Drosophila parkin mutants relative to wild-type controls. These results indicate that cell cycle dysfunction and ER stress are not involved in the majority of tissue degeneration in Drosophila parkin mutants. Although these findings appear to challenge the involvement of cell cycle dysfunction and ER stress in parkin pathogenesis, there are several caveats to extrapolating these results to humans. First, our studies involve a muscle degeneration phenotype, and the work implicating parkin in ER stress and cell cycle dysfunction involved vertebrate neuronal tissues. Secondly, the parkin substrate thought to trigger ER stress, Pael-R, does not appear to have a clear Drosophila counterpart. Thus, the lack of an ER stress response in Drosophila parkin mutants may reflect the absence of a Pael-R ortholog in this organism. Finally, a general limitation of our genetic screening method is that only a small fraction (5–10%) of Drosophila genes are represented in the collection of mutants analyzed. Thus, genes that are functionally involved in pathways important for ARJP pathogenesis may not be represented by the collection of mutants utilized in this study. Future saturation genetic screening may reveal pathways involved in parkin pathogenesis that were not detected in the current study.
The pathways that are implicated in parkin-mediated pathogenesis from our current and previous studies strongly correlate with factors implicated in sporadic PD, including mitochondrial dysfunction, apoptosis, oxidative stress and induction of the immune response. Thus, our findings suggest an underlying mechanistic similarity in the pathogenesis of ARJP and sporadic PD. Although many of the factors identified in our study appear to exhibit genetic interactions with parkin, our studies represent a starting point and an important goal of future work will be to further characterize the mechanisms by which these genes influence parkin pathogenesis. In particular, it will be important to explore the effects of alleles of these genes on parkin phenotypes other than partial pupal lethality. Further experiments with one of the genes identified from this work, GstS1, have shown that altered GstS1 abundance strongly modulates multiple parkin phenotypes, including behavioral and neuronal phenotypes (Whitworth et al., manuscript in preparation). Similar experiments with the other factors identified from this work should contribute to our understanding of parkin pathogenesis and may define avenues of therapeutic intervention in ARJP and possibly sporadic PD.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Drosophila strains and culture
All Drosophila strains were maintained on standard cornmeal/molasses medium at 25°C. The park25, parkrvA and UAS-park lines were described previously (33
). The EP collection was obtained from Exelixis. All other strains were obtained from the Bloomington Drosophila stock center at Indiana University.
Electron microscopy
Tissues for electron microscopy were prepared by dissecting aged pupae in 2% paraformaldehyde, 2.5% gluteraldehyde and fixing overnight. After rinsing in 0.1 M cacodylate buffer with 1% tannic acid, samples were post-fixed in 1 : 1 2% OsO4 and 0.2 M cacodylate buffer for 1 h. Samples were rinsed, dehydrated in an ethanol series and embedded using Epon.
Array analysis
One-day-old whole adult flies of the appropriate genotype (w; park25/park25 for mutants, w; parkrvA/park25 for controls) were collected, frozen in liquid nitrogen and stored at –80°C until use. Pupae of the same genotypes were collected at the white prepupal stage and allowed to age for 48 h at 25°C. The pupae were then frozen in liquid nitrogen and stored at –80°C until use. Total RNA was extracted from the appropriate tissue, either from 30 flies or 35–40 aged pupae per RNA prep using Trizol Reagent (Invitrogen). Three to four independent RNA samples (biological replicates) of mutants and controls were prepared.
An aliquot of 30 µg of total RNA was oligo dT primed and first strand cDNA was synthesized in the presence of amino-allyl dUTP. Cy3 or Cy5 was coupled to these strands and the reactions were quenched with 4 M hydroxylamine, combined and the labeled strands purified using Quiagen PCR purification kit. Labeled control and mutant samples were then hybridized together to chips containing either 6000 (1-day-old flies) or 12 000 (48 h pupae) spotted Drosophila cDNAs (Drosophila gene collection 1 and Drosophila gene collections 1 and 2, respectively). Arrays were hybridized, scanned and washed at the Array Facility at the Fred Hutchinson Cancer Research Center using standard procedures.
GenePix Pro 3.0 software was used to analyze the scanned files including flagging of spots not meeting quality standards. Background-subtracted intensities were calculated for all spots. Ratios of fluorescence intensity were normalized in an intensity-dependent manner using loess smoothing to correct for non-linearity (78). A total of six to eight microarrays were used for each experiment, including dye swapping within each sample pair. Dye swap values were averaged within an RNA preparation prior to statistical analysis. The normalized ratios for each experiment were then imput into the Cyber T algorithm (http://visitor.ics.uci.edu/genex/cybert/) (35), which determines which spots have ratios that differ significantly from the mean using Bayesian statistics. False discovery rates were assigned using a method that is independent of the fold change in expression of the transcript (79).
The EASE program (37) analysis was performed by imputing Flybase accession numbers for genes identified as alternately regulated by the transcriptional profiling and comparing that list to Flybase accession numbers for all genes represented on the microarray. Significant categories were those identified as having a maximum EASE score of 1x10–4 and maximum Bonferroni score of 2x10–2.
Slot blot analysis
Total RNA samples were prepared as described earlier. An aliquot of 3 µg of total RNA from either control or mutant samples were loaded into slots on a Bio-Dot SF microfiltration apparatus (Bio-Rad) and transferred to Hybond-N membrane. Probes were labeled by digoxigenin-incorporation (Invitrogen) in an in vitro transcription reaction using cDNAs from the array as templates and hybridized to membrane using standard techniques. Probes were detected using AP-conjugated anti-DIG and CDP-star luminescence reagent. Signal intensity was captured and quantified using Labworks image acquisition and analysis software (UVP Bioimaging). A minimum of six spots (technical replicates) were used for the intensity calculation for each probe used. The data were calculated as a ratio of signal intensity of the gene of interest to a control whose transcript level is not altered on the array (SNAP) for both parkin mutant (w; park25/park25) and control (w; parkrvA/parkrvA) RNA samples. The mean and SD of the ratios was calculated for each gene analyzed in both park25and control samples.
Antibody generation
The antiserum used to detect Drosophila Parkin was generated by expressing 6X-HIS-tagged recombinant Parkin using the Pet11a vector system (Novagen). Recombinant protein was dissolved in urea buffer and purified over Ni-NTA beads (Quiagen) using standard procedures. This protein was used to inoculate rabbits to generate serum (R&R Rabbitry, Stanwood, WA, USA) using standard procedures.
RNAi in cell culture
Drosophila S2 cells were maintained in culture at room temperature in Schneider's medium (Sigma) supplemented with 10% fetal bovine serum. Double-stranded RNA against parkin was synthesized using the Megascript RNAi kit (Ambion) and corresponded to the region delineated by primers CTGTTGCAATTTGGAGGGA and CTTTGGCACGGACTCTTTCT. An aliquot of 15 µg of dsRNA was added to cells and incubated for 4 days before harvesting (80). Cells were lysed in SDS protein sample buffer for western analysis.
Western blot analysis
Flies were raised at 25°C. Genotypes of flies used for tissue samples were park o/exp: w; hs-GAL4/UAS-park, park+/+: w; parkrvA/parkrvA, park–/–: w; park25/park25. Tissues were homogenized in protein sample buffer, run through 4–20% polyacrylamide gels using SDS–PAGE, and transferred to nitrocellulose membrane using standard techniques. Blots were probed with anti-Cyclin E [1 : 50, a gift from B. Edgar (81)], anti-Adh [1 : 20 000, a gift from S. Brogna (82)], or anti-Parkin 74n (1 : 10 000, discussed earlier). Blots were developed using HRP-conjugated secondary antibodies (Bio-Rad) and Western Lightning chemiluminesence reagent (PerkinElmer).
Genetic analysis
Array-identified genes.
To investigate the functional involvement of genes identified from transcriptional profile analysis, available mutations were identified and crossed into the park25 mutant background. Heterozygous flies bearing the mutation of interest and the park25 allele were crossed together and the resulting progeny were counted to determine the number of heterozygotes and homozygotes. To calculate the effect on viability of the array-identified alleles, the percentage of flies homozygous for the allele was determined in a park+/+ background from a cross of heterozygotes.
Genetic screen.
Initial screening crosses were carried out in polystyrene vials using standard food. Repetition of results was carried out in glass vials. All crosses were carried out at 25°C as described in Figure 5. For the 2nd chromosome EP lines, two to three males from each EP stock were mated to three to four females from the marked parkin stock in the G1 cross. For the G2 cross, three to four males of the appropriate genotype were mated to five to six females of the 24B-GAL4, parkin stock. These crosses were placed at 25°C. Twelve days after the G2 cross was set up, the vials were emptied. Flies were allowed to eclose for 4 days, then collected and counted. parkin mutant flies recovered from these crosses were also examined qualitatively for wing posture and general movement ability.
|
For the X chromosome stocks, three to four virgin females were collected from each of the EP stocks and crossed to two to three males bearing the marked parkin chromosome. Appropriate male progeny from this cross were then set up in the G2 cross as described for the 2nd chromosome screen. Only the female progeny from the G2 cross carried the EP element and were included in the ratio calculations. Male G2 progeny from these crosses were examined to control for the viability of parkin mutants under the conditions used for these crosses.
|
ACKNOWLEDGEMENTS |
|---|
The authors would like to thank the Berkeley Drosophila Genome Project and the Bloomington Stock Center for providing information and fly stocks used in this work, J.A. Marty, J. Delrow, and B. Smith for help with technical aspects of this work, P. Wes, M. Babcock, F. Perez, P. Muchowski and members of the Pallanck laboratory for scientific discussion and critical comments on the manuscript. This work was supported by National Institutes of Health Grant 1RO1NS41780-01 (to L.J.P.).
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
- Bharath, S., Hsu, M., Kaur, D., Rajagopalan, S. and Andersen, J.K. (2002) Glutathione, iron and Parkinson's disease. Biochem. Pharmacol., 64, 1037–1048.[CrossRef][ISI][Medline]
- Alam, Z.I., Jenner, A., Daniel, S.E., Lees, A.J., Cairns, N., Marsden, C.D., Jenner, P. and Halliwell, B. (1997) Oxidative DNA damage in the parkinsonian brain: an apparent selective increase in 8-hydroxyguanine levels in substantia nigra. J. Neurochem., 69, 1196–1203.[ISI][Medline]
- Alam, Z.I., Daniel, S.E., Lees, A.J., Marsden, D.C., Jenner, P. and Halliwell, B. (1997) A generalised increase in protein carbonyls in the brain in Parkinson's but not incidental Lewy body disease. J. Neurochem., 69, 1326–1329.[ISI][Medline]
- Janetzky, B., Hauck, S., Youdim, M., Riederer, P., Jellinger, K., Pantucek, F., Zochling, R., Boissl, K. and Reichmann, H. (1994) Unaltered aconitase activity, but decreased complex I activity in subbstantia nigra pars compacta of patients with Parkinson's disease. Neurosci. Lett., 169, 126–128.[CrossRef][ISI][Medline]
- Mizuno, Y., Yoshino, H., Ikebe, S.-i., Hattori, N., Kobayashi, T., Shimoda-Matsubayashi, S., Matsumine, H. and Kondo, T. (1998) Mitochondrial dysfunction in Parkinson's disease. Ann. Neurol., 44, S99–S109.[CrossRef][ISI][Medline]
- Mizuno, Y., Saitoh, T. and Sone, N. (1987) Inhibition of mitochondrial NADH-ubiquinone oxidoreductase activity by 1-methyl-4-phenylpyridinium ion. Biochem. Biophys. Res. Commun., 143, 294–299.[ISI][Medline]
- Mizuno, Y., Ohta, S., Tanaka, M., Takamiya, S., Suzuki, K., Sato, T., Oya, H., Ozawa, T. and Kagawa, Y. (1989) Deficiencies in complex I subunits of the respiratory chain in Parkinson's disease. Biochem. Biophys. Res. Commun., 163, 1450–1455.[CrossRef][ISI][Medline]
- Conway, K.A., Harper, J.D. and Lansbury, P.T. (1998) Accelerated in vitro fibril formation by a mutant alpha-synuclein linked to early-onset Parkinson disease. Nat. Med., 4, 1318–1320.[CrossRef][ISI][Medline]
- Liu, Y., Fallon, L., Lashuel, H.A., Liu, Z. and Lansbury, P.T., Jr (2002) The UCH-L1 gene encodes two opposing enzymatic activities that affect alpha-synuclein degradation and Parkinson's disease susceptibility. Cell, 111, 209–218.[CrossRef][ISI][Medline]
- Czlonkowska, A., Kurkowska-Jastrzebska, I., Czlonkowski, A., Peter, D. and Stefano, G.B. (2002) Immune processes in the pathogenesis of Parkinson's disease—a potential role for microglia and nitric oxide. Med. Sci. Monit., 8, RA165–RA177.[Medline]
- McGeer, P.L., Yasojima, K. and McGeer, E.G. (2002) Association of interleukin-1 beta polymorphisms with idiopathic Parkinson's disease. Neurosci. Lett., 326, 67–69.[CrossRef][ISI][Medline]
- Huang, Y., Cheung, L., Rowe, D. and Halliday, G. (2004) Genetic contributions to Parkinson's disease. Brain Res. Brain Res. Rev., 46, 44–70.[CrossRef][Medline]
- Bonifati, V., Oostra, B.A. and Heutink, P. (2004) Unraveling the pathogenesis of Parkinson's disease—the contribution of monogenic forms. Cell Mol. Life Sci., 61, 1729–1750.[ISI][Medline]
- Hardy, J. (2003) Impact of genetic analysis on Parkinson's disease research. Mov. Disord., 18 (Suppl. 6), S96–S98.[Medline]
-
Abbas, N., Lücking, C.B., Ricard, S., Dürr, A., Bonifati, V., De Michele, G., Bouley, S., Vaughan, J.R., Gasser, T., Marconi, R. et al. (1999) A wide variety of mutations in the parkin gene are responsible for autosomal recessive parkinsonism in Europe. Hum. Mol. Gen., 8, 567–574.
[Abstract/Free Full Text] -
Periquet, M., Latouche, M., Lohmann, E., Rawal, N., De Michele, G., Ricard, S., Teive, H., Fraix, V., Vidailhet, M., Nicholl, D. et al. (2003) Parkin mutations are frequent in patients with isolated early-onset parkinsonism. Brain, 126, 1271–1278.
[Abstract/Free Full Text] - West, A.B. and Maidment, N.T. (2004) Genetics of parkin-linked disease. Hum. Genet., 114, 327–336.[CrossRef][ISI][Medline]
-
Pawlyk, A.C., Giasson, B.I., Sampathu, D.M., Perez, F.A., Lim, K.L., Dawson, V.L., Dawson, T.M., Palmiter, R.D., Trojanowski, J.Q. and Lee, V.M. (2003) Novel monoclonal antibodies demonstrate biochemical variation of brain parkin with age. J. Biol. Chem., 278, 48120–48128.
[Abstract/Free Full Text] -
Lücking, C.B., Dürr, A., Bonifati, V., Vaughan, J.R., De Michele, G., Gasser, T., Harhangi, B.S., Meco, G., Denefle, P., Wood, N.W. et al. (2000) Association between early-onset Parkinson's disease and mutations in the parkin gene. New Engl. J. Med., 342, 1560–1567.
[Abstract/Free Full Text] - Lücking, C.B., Chesneau, V., Lohmann, E., Verpillat, P., Dulac, C., Bonnet, A.M., Gasparini, F., Agid, Y., Durr, A. and Brice, A. (2003) Cod