Drosophila parkin mutants have decreased mass and cell size and increased sensitivity to oxygen radical stress

Parkinson disease (PD) is a progressive neurodegenerative disorder and individuals with PD present symptoms of resting tremors, rigidity and akinesia. Pathological examination of the brains of individuals with idiopathic PD show loss of dopaminergic (DA) neurons in the substantia nigra of the midbrain with the formation of cytoplasmic protein inclusions called Lewy bodies (LBs). The loss of these DA neurons might underlie the clinical symptoms of PD. Although the cause of idiopathic PD is unknown, various environmental toxins, such as pesticides and those structurally similar to MPTP, have been strongly implicated(Lockwood, 2000). MPTP(1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine),when converted to the active toxin MPP(+) by monoamine oxidase B, can induce parkinsonism in primates. MPP(+) interferes with mitochondrial complex I,resulting in an inhibition of ATP synthesis and the generation of free radicals, causing DA neuron damage (Dawson et al., 2002). In addition, endogenous toxins may also be a primary or an additional cause of DA neuron loss in a susceptible population. Specifically, dopamine readily oxidizes to produce free radicals, which can inflict damage on the cell. Mitochondria use several defense systems such as catalase, superoxide dismutase and glutathione as protection against oxygen radicals. However, these systems can be overwhelmed, resulting in mitochondrial dysfunction and triggering apoptosis(Orth and Schapira, 2002).

Recent identification of a number of familial PD cases and the genes involved therein has provided new insights into the pathogenesis of PD. Missense mutations in α-Synuclein, identified in a small number of families, result in an autosomal dominant familial PD(Kruger et al., 1998; Polymeropoulos et al., 1997).α-Synuclein is a presynaptic protein and is a major component of LBs(Clayton and George, 1998). Misexpression of α-Synuclein in cultured cells results in aggregate formation, abnormal mitochondrial morphology, and increased levels of free radicals (Hsu et al., 2000). A number of transgenic models of PD have been generated using overexpression ofα-Synuclein in both mice and flies(Feany and Bender, 2000; Giasson et al., 2002; Lee et al., 2002; Masliah et al., 2000).

Mutations in PARK2 have been linked to Autosomal Recessive-Juvenile Parkinson's (AR-JP)(Kitada et al., 1998). The dopaminergic loss in AR-JP is generally not associated with LB formation(Mizuno et al., 1998). PARK2 is highly expressed throughout the brain, is present in both the cytoplasm and the nucleus, and is also associated with mitochondrial membranes (Darios et al.,2003; Kubo et al.,2001; Shimura et al.,1999; Stichel et al.,2000). The human PARK2 protein consists of four domains: an N-terminal ubiquitin-homology domain and two RING finger domains flanking a cysteine-rich in between RING fingers (IBR) domain(Fig. 1A)(Morett and Bork, 1999). Proteins containing RING finger domains frequently act as ubiquitin E3 ligases(Joazeiro and Weissman, 2000). E3 proteins function as part of the ubiquitin-proteasome system required to degrade proteins that are misfolded, damaged by oxidative stress, nitrated or ubiquitinated (Grune et al.,1998; Halliwell,2001; Sitte et al.,2000; Stadtman and Berlett,1998). Vertebrate parkin also acts as an E3 ligase, helping to ubiquitinate and degrade several proteins, including Pael receptor (Pael-R),α-Synuclein, CDCrel-1 and Synphilin 1. Familial mutations in PARK2 abolish its E3 activity (Chung et al.,2001; Imai et al.,2000; Shimura et al.,2000; Zhang et al.,2000). This and other data (see below) suggest that loss of PARK2 may result in increased amounts of damaged and misfolded proteins, leading to cellular dysfunction and apoptosis of DA neurons(Bence et al., 2001; Mezey et al., 1998). A familial mutation in another ubiquitin/proteasome system member, ubiquitin carboxy-terminal hydrolase I, also results in an early onset PD, underscoring the importance of this system in PD (Leroy et al., 1998).

Screens to identify PARK2 targets and overexpression studies have provided evidence that PARK2 plays an important role in protecting the cell from toxic insults. Misexpression of Pael-R, a substrate of human PARK2, causes it to become unfolded, insoluble and ubiquitinated, thereby resulting in cell death. An insoluble form of Pael-R was also found to accumulate in AR-JP brains. PARK2 binds and ubiqutinates Pael-R, leading to the degradation of its insoluble form, and protects cells from Pael-R-induced cell death (Imai et al., 2001). Furthermore, misexpression of Pael-R in Drosophila dopaminergic neurons results in DA cell loss that is suppressed by co-expression with parkin. By contrast, co-expression of Pael-R with RNAi knockdown of endogenous Drosophila parkin results in enhanced DA cell loss (Yang et al.,2003). α-Synuclein-induced DA cell loss is also suppressed by co-expression of parkin(Auluck et al., 2002; Yang et al., 2003). In addition, Parkin can promote the ubiquitination and degradation of an O-glycosylated form of α-Synuclein(Shimura et al., 2000). Overexpression of PARK2 in tissue culture cells results in increased proteasomal activity and decreased levels of protein carbonyls and 3-nitrotyrosine-containing proteins (markers of proteins modified by reactive nitrogen species) (Hyun et al.,2002). Although the roles of two other Parkin-interacting proteins, CDCrel-1 and Synphilin 1, in AR-JP are still unclear, it is possible that Synphilin 1 may mediate interaction between Parkin and a more abundant non-glycosylated form of α-Synuclein. Finally, PARK2 expression is sufficient to prevent mitochondrial swelling and cytochrome c release in ceramide-induced apoptosis and Parkin is associated with the outer mitochondrial membrane (Darios et al.,2003). Taken together, these results suggest that Parkin may be important in mitochondrial functions such as oxygen radical defense or the cell death pathway.

Previous work has shown that Drosophila parkin is required for longevity, male fertility and survival of flight muscle fibers(Greene et al., 2003). In this report, we confirm these observations and extend the requirement for parkin to several other processes. Specifically, we show that parkin mutant flies have reduced mass, cell size and number; female infertility; and increased sensitivity to chemical and environmental stress. We also confirm that loss of parkin does not result in any significant DA neuron loss.

We obtained the EP(3)3515 P-element insertion in the orbit gene from the Bloomington Drosophila Stock Center. This element, which causes recessive sterility, was mobilized usingΔ 2-3 transposase and new insertions were identified using inverse PCR as described (Tavsanli et al.,2001). We identified a new insertion 1 kb 3′ of parkin (named dpk^P30) that maintained an intact EP(3)3515 element. Reversion of sterility was used to select for precise excision of the EP element and PCR was employed to identify those lines with an intact dpk^P30 insertion. We then mobilized dpk^P30 and screened for new insertions in or near parkin. We identified a new insertion 29 bp 5′ of the parkin translational start site that also maintained an intact dpk^P30 element. This line is called dpk^P21. In addition, we also identified two other lines, dpk^P23 and dpk^P24, that truncate parkin 127 and 155 amino acids 3′ of the N terminus and behave as genetic nulls. To generate parkin deletion chromosomes, the dpk^P30-dpk^P21 double insertion was mobilized with Δ2-3 transposase as described above. Progeny were screened for the absence of w⁺ eyes. 66 w^- males were identified and balanced, and their progeny were screened by Southern blot analysis. Five out of 66 lines were found to have parkin DNA deleted. These five deletion lines were screened by PCR and sequencing for the presence of intact genomic DNA flanking the two starting elements. We identified one parkin deletion line meeting these criteria, dpk^Δ21(Fig. 1B). The c179-Gal4 that drives Gal4 expression in the adult muscle as well as other tissues, was obtained from the Bloomington collection (stock BL–6450) (Manseau et al.,1997).

Using a BLAST search of the Drosophila genome database, we identified an EST clone (SD01679) that encodes Drosophila parkin. We used this EST to probe a Drosophila genomic library (kindly provided by Ron Davis) and identified lambda phage clones that span the parkinlocus. Phage integrity was verified by restriction digests as well as sequence analysis of the parkin locus. These phage were used to generate the dpkR (Rescue) genomic rescue construct that encompasses the following genomic coordinates: chromosome 3L, 21098849-21124696(www.ensembl.org)(Fig. 2A). This rescue fragment was cloned into the pCasper 4.0 vector to generate transgenic flies. The dpkSR (Stop-Rescue) construct was created by introducing stop codons in all three reading frames of parkin using PCR mutagenesis but is otherwise identical to dpkR. These stop codons are located 66 base pairs 3′ of the parkin AUG. Two and five independent transgenic lines of dpkSR and dpkSR, respectively, were obtained. In addition to the parkin mutant chromosome, dpk^Δ21, we also used an independent chromosome that had a deficiency uncovering the entire parkin gene, Df(3L) pc-MK (Bloomington, stock BL-3068). UAS-HA::parkin transgene was built using the pUAST vector with an engineered HA-tag sequence (courtesy or Beril Tavsanli) fused to the parkin-coding sequence from SD01679 EST clone. This EST clone contained an A→T difference at base pair 760 (of 1446) of the coding sequence compared to the genome database and our genomic phage clones. Therefore, we replaced a NarI fragment of SD01679 with the same fragment from EST clone AT25577 to correct for this change.

RNA was isolated from staged embryos, larvae, pupae, and adult flies using 3 M LiCl, 6 M UREA and 0.2% SDS as previously described(Kelley, 1993). Poly(A)⁺ RNA was isolated using oligo(dT)-cellulose columns, subjected to electrophoresis in a 1% agarose, 6.25% formaldehyde gel and blotted. We hybridized blots with a parkin cDNA probe and subsequently with an RP49 cDNA probe using the RediPrime II Random Prime Labeling System (Amersham Pharmacia Biotech).

Whole-mount immunohistochemistry was performed as previously described(Davis et al., 2003). Brains from parkin mutant and heterozygous adults, aged 2 and 21 days post eclosion, were fixed in PLP, washed in PAXDG and stained with rabbit polyclonal anti-Drosophila-tyrosine hydroxylase antibody (courtesy of Wendi Neckameyer) (Mardon et al.,1994; Neckameyer et al.,2001). Subsequently, brains were stained with Alexa goat anti-rabbit antibodies in PAXDG. Brains were mounted in Vectashield and dorsomedial clusters identified and the number of cells counted using confocal microscopy (Zeiss LSM 510 microscope).

Two-day old adult flies were dissected by removing the head and abdomen in ice cold fixative (2% paraformaldehyde/3% glutaraldehyde/0.1 M sodium cacodylate/0.05% CaCl₂). Whole thoraces were fixed overnight at 4°C and then washed with 0.1 M cacodylate. Postfixation was performed with 2% OsO₄ for 2 hours at room temperature and the sample was washed with water before going through a series of progressive dehydration steps in ethanol:water mixtures. The sample was then embedded in Spurr's resin(Spurr, 1969). Thoraces were then trimmed to retain only the dorsal longitudinal muscles. Ultrathin sections of 50 nm were then made with a diamond knife and stained with Toluidine Blue dye, which labels nucleic acids, thereby marking both nuclei and cytoplasm (Balabanova et al.,2003). For TEM, sections were stained on the grid first with 4%uranyl acetate, then with 2.5% lead nitrate, five minutes each at room temperature. The sections were observed with a JEOL 1010 transmission electron microscope. Late pupae were removed from their pupal cases in the same fixative as above and processed as described above.

The courtship behavior assay was carried out on parkin mutant and heterozygous males that were aged individually for one week as previously described (Hall, 1994). Individual virgin w^- females, comparable in age to the males, were placed in a food vial with individual males, and the elements of the courtship behavioral sequence were noted for a total of 1 hour. Mating pairs were considered to be successful when the courtship behavior ended with copulation. Fertility was assayed by placing single parkin males with 3-4 virgin w^- females and by placing single virgin parkinfemales with two w^- males. Vials were then checked 3-7 days later for the presence of larvae.

Mass analysis was performed as described(Bohni et al., 1999). Twenty individual males and females were weighed on a precision scale (range 0.01 mg to 50 g; Mettler Toledo AG245). In addition, groups of 20 and 30-100 flies of each genotype were weighed to confirm the accuracy of the individual fly weighing. Flies were reared under the same conditions and were age matched(48-72 hours post eclosion). Cell counts to determine the differences in cell size in the wing were assessed by counting the number of wing hairs on the dorsal wing surface in the same arbitrary area just posterior to the posterior cross vein for all genotypes (n=20 wings/genotype, P<0.001 one-way ANOVA test). Wing area was determined using NIH image 1.60 software(http://rsb.info.nih.gov/nih-image/Default.html)(n=20 wings/genotype, P<0.001 one-way ANOVA test).

Flies were anesthetized with CO₂ and individually placed in vials 24 hours before the assays were conducted to allow time for a full recovery from effects of CO₂. Each fly was dispelled into a plastic petri dish by gently tapping the vial. The dish was covered with a lid to prevent flies from escaping. The petri dish was then tapped lightly on the bench to initiate a jumping reaction. Flight tests were performed by removing the lid of the petri dish and inverting the dish at a height of 29 inches. The lid was lightly tapped to loosen the fly. Flies either fell in a straight line, fell by veering or flew. The number of flies exhibiting each behavior was recorded. Twenty flies of each genotype were assayed.

Paraquat (methyl viologen) toxicity was performed as previously described(Clancy et al., 2001). Two-day old animals were placed in vials (10 per vial) with food media made of 0.8%agar containing 5% sucrose and 2 mM paraquat (Sigma). Control flies for each genotype were placed under the same conditions minus paraquat. All experiments were performed in the dark. Animals were scored daily and the media was replaced every 3-4 days.

Resistance to cold stress of 2-day old flies was performed as previously described (Mockett et al.,2003). Flies were placed (n=10) in a vial and kept at 4°C, except for a 1 hour recovery period (25°C) daily prior to scoring mortality.

Flies were maintained on standard media at 25°C, ten per vial, and transferred to new food medium daily. Mortality was scored daily.

Analysis of EST clone sequences and genomic data indicates that parkin consists of six exons (CG10523, www.bdgp.org). The predicted open reading frame encodes a 482 amino acid protein with four domains that follow the same order as in human Parkin: an N-terminal ubiquitin-homology domain, two ring finger domains and an in between ring(IBR) domain which are all highly conserved(Fig. 1A). The human and Drosophila proteins have an overall identity of 46% and an E-value of 3.1E^–102 (Fig. 1B). No other obvious human Parkin homolog is present in the Drosophila genome.

To determine the developmental expression of parkin we performed northern blot analysis of embryonic, larval, pupal and adult RNAs. This reveals a 1.6 kb transcript, which is similar in length to the longest parkin EST clone identified (SD01679)(Fig. 1C). parkintranscript is first detected early (0- to 2-hour-old embryos) and is probably due to a maternal contribution. Although parkin transcript is not detected at high levels during later embryonic stages, it is present in third instar larvae, as well as throughout pupal development and into adulthood(Fig. 1C).

We used P-element mutagenesis to generate null mutations in parkin(Fig. 2A,B)(Kaiser and Goodwin, 1990). A P-element located 26 kb 3′ of parkin [EP(3)3515] was used to conduct a local hop screen for new insertions in or near the parkin locus (see Materials and methods)(Fig. 2A). We obtained a new insertion 1 kb 3′ of parkin, termed dpk^P30. After precisely excising the starting element [EP(3)3515], the new element(dpk^P30) was mobilized to generate new insertions in and flanking parkin (dpk^P21, dpk^P23, and dpk^P24) (Fig. 2B and data not shown). Subsequent simultaneous mobilization of dpk^P30 and dpk^P21 resulted in deletion of the entire parkin-coding region. PCR and Southern analyses were used to confirm deletion of the gene (data not shown). We also obtained two insertions directly in the parkin-coding region, dpk^P23 and dpk^P24, which behave genetically as null alleles (data not shown). To confirm that observed phenotypes are specific to parkin we generated two rescue transgenes. The first transgene contains parkin and 10-12 kb flanking it on either side. The second transgene is identical to the first with the exception that stop codons were introduced in the coding sequence of parkinshortly after the translational start site. Rescue by the first transgene, dpkR (Rescue), but not the second, dpkSR (Stop Rescue), indicates that observed phenotypes are specific to loss of parkin function.

parkin mutant animals are homozygous viable but eclose one day later than heterozygous controls. In addition, loss of parkin results in a significant pupal lethality (Fig. 2C). Although there is little or no lethality prior to pupation,68% of parkin mutant animals fail to eclose from their pupal cases,suggesting that parkin plays an important role during pupal development (Fig. 2C). parkin null animals that do eclose exhibit a significant reduction in body size (Fig. 3A). Mutant flies present a mass a reduction of 34% in females and 30% in males when compared with controls of the same age and sex. This phenotype is partially but significantly rescued with dpkR but not dpkSRtransgenes, confirming that the mass reduction is due to parkin and not any of the surrounding genes that could have been affected by the parkin deletion (Fig. 3B).

The reduction in body size could be accounted for by a reduction in cell number and/or a reduction in the size of individual cells. We used the Drosophila wing to determine whether cell number, cell size or both is reduced in parkin mutants. Each wing cell normally secretes cuticle and a single hair. Therefore, cell number and cell size in the wing can be determined by counting wing hairs and determining their density over an area (Bohni et al., 1999). Compared with heterozygous controls, cell size in parkin mutant female wings is reduced by 10% and the wing area itself is reduced by 23 %(n=20, P<0.001 one-way ANOVA test)(Fig. 3C; Table 1). This suggests that parkin may play a role in cell growth, proliferation and/or survival.

Similar to previously published work, we observe that loss of parkin function also reduces longevity(Greene et al., 2003). We analyzed lifespans of w; dpk^Δ21and w; dpk^Δ21 /Df(3)pc-MK flies. Df(3)pc-MK is an independent deficiency that completely uncovers the parkin locus. The mean life spans for w;dpk^Δ21 and for w;dpk^Δ21 /Df(3)pc-MK flies are 13 and 32 days, respectively, while heterozygous controls have a mean life span greater than 60 days (Fig. 4A). This phenotype is rescued by dpkR but not dpkSR transgenes and w; dpk^Δ21 and w;dpk^Δ21 /Df(3)pc-MK animals are rescued to the same degree by dpkR. These results suggest that the less severe phenotype of w; dpk^Δ21/Df(3)pc-MK compared with w;dpk^Δ21 is due to a second site suppressor on the Df(3)pc-MK chromosome(Fig. 4A).

A high rate of oxygen radical production (as part of dopamine oxidation)and corresponding mitochondrial damage to DA neurons is thought to be related to the death of these cells in individuals with PD(Braunwald et al., 2001). Furthermore, overexpression of PARK2 protects against cell death(Darios et al., 2003). Therefore, we hypothesized that loss of parkin may lead to a reduction in stress resistance, perhaps contributing to reduced longevity. We subjected parkin mutant flies to two types of stress: chemical(paraquat) and environmental (cold stress). To assay oxygen radical resistance, parkin mutant animals were placed on 2 mM paraquat(methyl viologen)-supplemented food(Phillips et al., 1989). Paraquat is an MPTP-like toxin that produces oxygen radicals. Mutant animals,compared with controls, show high sensitivity to this low dose of paraquat as most mutants died by one week of age while heterozygous controls showed minimal lethality (Fig. 4B). We used cold to further test the stress resistance of parkin-null animals and discovered that parkin mutant flies show a significant reduction in resistance to cold temperature. The mean lifespan of parkin mutant flies maintained at 4°C is 5-7 days compared with 50 days for heterozygous controls (Fig. 4C) (Mockett et al.,2003). Taken together, these results suggest that parkinmay be required for stress resistance and protection against oxygen radical damage.

Previous work has shown that parkin is required for male fertility and in particular for proper sperm maturation(Greene et al., 2003). We confirmed that parkin mutant males are sterile and have found that females are infertile as well. Neither sex yields progeny when crossed with wild-type animals of the opposite sex. Furthermore, only three out of 20 mutant males exhibited one of seven tested courtship behaviors (data not shown, see Materials and methods) (Hall,1994). This suggests that parkin is important in either sexual behavior, function or both.

We also confirmed that parkin null animals are unable to fly(Table 2) and exhibit a drooped wing phenotype: mutant wings are positioned at an approximately 45° angle to the ground, while wild-type flies hold their wings parallel to the ground(Greene et al., 2003). To investigate the flight and drooped wing phenotypes, we analyzed thoracic indirect flight muscles (IFMs) using light microscopic analysis of thin resin sections and transmission electron microscopy (TEM) at both pupal and adult stages (Figs 5, 6). At a late pupal stage (94 hours after pupation) all IFMs appear normal(Fig. 5A-D). By contrast, 2 days post eclosion parkin mutant animals have significant IFM degeneration (Fig. 5F,H,asterisks). By 11 days, most of the IFMs show signs of degeneration(Fig, 5J,L, asterisks). Heterozygous controls and mutant animals containing the rescue transgene, dpkR, did not show any signs of a muscle loss, while parkinmutants carrying the dpkSR transgene were not rescued(Fig. 5E,G,I,K).

We used TEM to further investigate the nature of this IFM defect. Wild-type muscle fibers have a highly regular and uniform structure(Fig. 6A). At the late pupal stage, parkin mutant IFM fibers have normal morphology that is indistinguishable from control animals (data not shown). However, at 2 days post eclosion, parkin mutants IFMs have roughened edges, increased spacing between fibril groups, and a reduced number of fibrils(Fig. 6B,D). There is also a reduced number of mitochondria which are of a generally smaller and variable size with a significant loss of christae(Fig. 6B, asterisk). In addition, signs of apoptosis, including chromatin condensation and nuclear envelope breakdown, are readily apparent in parkin mutant muscle tissues (Fig. 6F) (Cotran,1999). The dpkR, but not dpkSR, transgene rescues these phenotypes (Table 2; Fig. 6). In addition, we were able to rescue the flight defect with a UAS-HA-parkin transgene using both ubiquitous (Ubi-Gal4) and muscle (C179-Gal4) drivers(data not shown). These data further confirms that loss of parkin is responsible for the IFM degeneration phenotype. At 2 days post-eclosion not all IFM groups within each animal exhibit this phenotype, suggesting that muscle degeneration is progressive. parkin mutant flies can survive for several weeks and are able to walk, but not jump or fly, suggesting that only specific muscle groups are affected. As these flies are able to feed,smooth muscle is also unlikely to be significantly affected. We conducted neuromuscular junction (NMJ) electrophysiological studies to determine if there is a neural and/or muscular defect at the NMJ during the third larval stage (L3). Our tests show that parkin mutants have normal L3 recordings (data not shown). As third instar larval neuromuscular junction recordings in parkin null animals are normal, the primary role for parkin in muscle development and maintenance probably occurs during pupal stages.

The pathological presentation of familial cases of Parkinson disease is an age-dependent degeneration of dopaminergic neurons in the substantia nigra of the midbrain (Cotran, 1999). Therefore, we were interested to determine if there was a corresponding age-dependent loss of dopaminergic neurons in parkin mutant animals. We specifically focused on the dorsomedial clusters (DMC) of the Drosophila brain dopaminergic system. DMC neurons were previously reported to be affected in transgenic Drosophila misexpressing α-Synuclein and Paelreceptor. These clusters contain 14-18 tyrosine hydroxylase (TH)-positive cells that were shown to degenerate in an age-dependent manner in the previously reported transgenic models(Auluck and Bonini, 2002; Auluck et al., 2002; Feany and Bender, 2000; Yang et al., 2003). To assay DA neurons in these clusters, we used whole-mount brain staining and confocal microscopy (Fig. 7A,B). We analyzed 2- to 3-day-old and 21-day-old animals, and did not observe any significant difference in the number of TH-positive cells(Fig. 7C). Furthermore, we did not observe any defects in DA cell body appearance. These findings are consistent with previously published results in flies and mice, and suggest that parkin is not essential for DA neuron survival at least during the 21 days post eclosion (Greene et al.,2003; Itier et al.,2003). In addition, parkin mutants do not have an abnormal electroretinogram (ERG) shortly after eclosion or at 2 weeks of age(data not shown). This suggests that photoreceptor neurons are also intact.

Human PARK2 is involved in several processes that may provide important insights concerning the pathogenesis of Parkinson disease. Parkin is an E3 ubiquitin ligase that is thought to target several proteins for degradation, including α-Synuclein and Pael receptor(Shimura et al., 2001; Yang et al., 2003). Misexpression of α-Synuclein results in mitochondrial damage and increased levels of free radicals while misexpression of Pael-Rresults in cell death (Hsu et al.,2000). This suggests that human Parkin may play a role in protecting the cell from cytotoxic insults. In addition, Parkin interacts with cyclin E as part of the SCF^hSel–10 complex(Staropoli et al., 2003). This may be important in PD as accumulation of cyclin E is associated with apoptotic death in postmitotic neurons that are exposed to the glutamatergic excitotoxin kainate. Glutamatergic toxicity has been previously implicated in the pathogenesis of PD (Olanow and Tatton,1999). Overexpression of PARK2 suppresses both cyclin E accumulation and excitotoxin-induced apoptosis of granular neurons. In addition, PARK2 deficiency potentiates and PARK2overexpression protects against kainate-medicated toxicity in cultured dopaminergic neurons (Staropoli et al.,2003). Furthermore, overexpression of PARK2 in tissue culture protects cells from mitochondrial swelling and cytochrome c release during apoptosis (Darios et al.,2003). Therefore, PARK2 may be involved in the regulation of cell death. Furthermore, interaction of human Parkin with cyclin E suggests that it may also be involved in regulation of the cell cycle. Nevertheless,the molecular mechanisms of Parkin function and its precise role in PD are poorly understood. In this report, we describe a Drosophila model of PD that will help our understanding of the molecular mechanisms of parkin function and may thereby provide insights into the pathogenesis of PD. To this end, we generated loss of function mutations in Drosophila parkin. Many of the phenotypes of parkin mutant animals parallel its inferred roles based on in the vitro experiments outlined above. First, parkin mutant animals have reduced body size and cell size at eclosion, consistent with a possible role for parkin in regulating the cell cycle via interaction with cyclin E. Second, parkin mutants are highly sensitive to oxidative and environmental stress, suggesting possible defects in oxygen radical defense. Finally, parkin null animals show apoptotic flight muscle degeneration supporting other data that parkin may be involved in the regulation of cell death (Greene et al.,2003).

parkin mutant animals have reduced body and cell size at eclosion,suggesting possible defects in cell growth, proliferation and/or cell survival. This phenotype is particularly interesting given the interaction of human Parkin with cyclin E, an important regulator of cell cycle progression(Staropoli et al., 2003). The reduced body size of parkin mutants is similar to the phenotypes of insulin growth factor (IGF) receptor mutant flies and mice(Bruning et al., 2000). Physiological effects of insulin in the brain are not limited to regulation of food intake and control of glucose uptake, but are also important in trophic actions on neurons and glial cells. Administration of the N-terminal tripeptide of IGF1 prevents loss of DA neurons after chemically(6-hydroxydopamine)-induced DA cell lesion in rats(Guan et al., 2000). In addition, IGF1 also protects against DA-induced neurotoxicity in vitro(Offen et al., 2001). This,together with our data, suggests parkin may play a role in the insulin signaling pathway during development or in adults.

Our results show that parkin mutant flies have increased sensitivity to paraquat toxicity. Paraquat, with its two N-methyl pyridinium moieties, is structurally similar to MPP(+), a toxic metabolite of the MPTP. In humans, toxins such as MPTP cause DA neuron-specific death in the substantia nigra because production of oxygen radicals, resulting in Parkinsonian symptoms. Glutathione and superoxide dismutase (SOD) inactivate H₂O₂ and superoxide radicals, respectively, thereby reducing MPTP neurotoxicity in mice(Przedborski et al., 2001). Paraquat also causes high oxygen radical production and interferes with mitochondrial respiration, resulting in cell death. The effects of paraquat are not specific to DA cells and sensitivity to its toxicity reflects a general impairment of oxygen radical defense(Ding and Keller, 2001; Phillips et al., 1989). Loss of Drosophila SOD also results in sensitivity to low doses of paraquat, reflecting a similar defect in oxygen radical defense(Kirby et al., 2002; Phillips et al., 1989). Our results suggest that antioxidant defenses in parkin mutant flies are also impaired and that parkin may play a role in the oxidative stress response. One likely mechanism of Parkin function is to help rid the cell of specific proteins that are misfolded because of oxygen radical damage. This is consistent with the identification of human Parkin as an E3 ligase(Imai et al., 2000; Shimura et al., 2000; Staropoli et al., 2003; Zhang et al., 2000).

parkin mutant flies show a progressive apoptotic degeneration of indirect flight muscles (IFMs). Furthermore, this degeneration is accompanied by mitochondrial disintegration and loss of christae(Greene et al., 2003). Drosophila IFMs are groups of specialized muscle that are in a constant state of vibration. They require a high oxygen supply to sustain their respiratory activity, making this tissue especially susceptible to mitochondrial dysfunction. It is possible that the mitochondrial degeneration we observe in parkin mutants results in increased susceptibility to oxygen radical damage because of the impairment of antioxidant defenses by mitochondria, culminating in cell death. The data from animal and tissue culture models of PD outlined above suggest that mitochondrial dysfunction and oxygen radical damage are two crucial factors in the development of PD pathology. In addition, they also underscore the importance of the apoptotic pathway in DA cell loss. Oxidative stress such as that produced by MPTP can trigger apoptosis (Kruman et al.,1998). Transgenic mice overexpressing the anti-apoptotic gene Bcl2 and mice null for the proapoptotic gene Bax are resistant to MPTP (Offen et al.,1998; Vila et al.,2001). Loss of parkin results in similar phenotypes in flies: increased sensitivity to oxygen radical stress and IFM apoptosis,suggesting that underlying mechanisms of cellular dysfunction maybe be similar between flies and humans.

Individuals with AR-JP exhibit a loss of DA neurons in the substantia nigra similar to that of idiopathic PD. Therefore, we expected to see a similar phenotype in parkin mutant animals. We focused on the dorsomedial cluster (DMC) neurons because they were shown to be affected in other Drosophila models of PD (Auluck et al., 2002; Feany and Bender,2000). However, we do not observe any significant loss of DA neurons at three weeks of age in parkin mutant animals compared with controls. In addition, in contrast to a previously published report, we do not see any changes in DA cell morphology in parkin mutants(Greene et al., 2003). Loss of mouse PARK2 also does not result in DA cell loss. However, increased extracellular DA concentration, abnormal neurophysiology and motor and cognitive behavioral deficits are observed in PARK2 mutant mice(Goldberg et al., 2003; Itier et al., 2003). By contrast, only motor deficits have been observed thus far in Drosophila (Greene et al.,2003). In another model of PD, overexpression of α-synuclein in mice does not result in DA cell loss but does cause DA neuron dysfunction(Giasson et al., 2002; Masliah et al., 2000; van Der Putten et al., 2000). Similarly, in Alzheimer's disease mouse models employing overexpression of amyloid precursor protein, cognitive defects are present in the absence of cortical neuronal loss (Hock and Lamb,2001). The absence of cell loss in multiple models of neurodegenerative disease may reflect the shorter life span of flies and mice compared with humans. Although the mean onset of PD symptoms in AR-JP is the earliest of all genetically defined forms of Parkinson's disease, it is not clear how to compare fly, mouse and human life spans in this regard(Gasser, 1998).

Even though we do not observe DA loss in parkin mutant animals,the mechanism of cell loss may be similar between humans and Drosophila. Neurons and muscle are two of the most energy-dependent tissues, and therefore have high numbers of mitochondria and are highly sensitive to mitochondrial insults. Although the mechanisms of cell death in IFMs and in DA neurons might be similar, it is possible that Drosophila IFMs are more sensitive than DA neurons to mitochondrial defects. The nature of this defect is still unclear; however, three mechanisms are likely. First, it is possible that parkin functions in a trophic factor pathway that promotes cell survival, and in its absence cells become more susceptible to insults such as oxygen radicals. Second, parkinmight be important in the stress response pathway, and in its absence the cell becomes more susceptible to various stimuli such as oxygen radical damage that trigger apoptosis. Finally, it is possible that parkin is part of the cell death pathway, and its absence results in susceptibility to proapoptotic insults. Drosophila parkin mutants will serve as an invaluable model for understanding the biological role of parkin and may provide important clues concerning the molecular mechanisms of PD.

We thank Richard J. Davis and Kartik S. Pappu for critical reviews of the first draft of this manuscript. Y.P. is also grateful to Richard J. Davis and Beril Tavsanli for teaching him how to conduct a local hop screen. We would also like to thank Ron Davis for providing us with a Drosophilagenomic library and Beril Tavsanli for providing a northern blot filter. This work was funded by The Michael J. Fox Foundation for Parkinson's Research. Y.P. was also funded by National Research Service Award F31 NS41852-02 from NINDS. He is also a Hudson Scholar as part of the Medical Scientist Training Program.