Antioxidants protect PINK1-dependent dopaminergic neurons in Drosophila

Results and Discussion

PINK1 gene encodes a 581-aa putative serine/threonine kinase with unknown substrates and function. Sequence analysis revealed a single Drosophila homolog (CG4523) of hPINK1with 52% similarity at the amino acid level that was designated as Drosophila PINK1 (dPINK1) (Fig. 5, which is published as supporting information on the PNAS web site). Quantitative real-time PCR showed that dPINK1 is expressed throughout the lifespan of Drosophila, with the highest levels in pupal stages and slightly reduced levels in adults (Fig. 6A, which is published as supporting information on the PNAS web site). In adult flies, dPINK1 mRNA is highly abundant in brain including the retina of the eye (Fig. 6B), suggesting a role for dPINK1 in the developing or adult nervous system of Drosophila.

To investigate the mechanism of PINK1 mutants in PD pathogenesis using Drosophila as a model, we reduced dPINK1 function with a transgenic dPINK1-RNAi by using the GAL4/UAS system (16). Knockdown of dPINK1 in multiple transgenic lines was verified by RT-PCR (Fig. 6C). Inactivation of dPINK1 driven by the ubiquitous driver daughterless (da)-GAL4 resulted in embryonic lethality of second-instar larvae. To determine RNAi specificity for dPINK1, we generated transgenic Drosophila lines expressing hPINK1. Expression of hPINK1 variants in flies driven by da-GAL4 was verified by immunoprecipitation followed by immunoblotting (Fig. 6D). Expression of hPINK1, but not lacZ or GFP, fully rescued the lethality caused by dPINK1 RNAi (Table 1, which is published as supporting information on the PNAS web site; data not shown). In contrast, expression of a mutant form of PINK1, hPINK1G309D, which was identified in patients with PD (4), failed to rescue (data not shown). The observed rescue is likely to be specific to hPINK1 activity, because little dPINK1 was detected in flies expressing both dPINK1 RNAi and hPINK1 cDNA (Fig. 6E). These results suggest that dPINK1 apparently is the functional homolog of hPINK1.

We next examined the effect of dPINK1 inactivation on DA neurons. There are 13 well defined DA neuronal clusters normally present in Drosophila adult brain, including six paired clusters in each brain hemisphere and an unpaired ventral medial cluster (VUM) (Fig. 1I). dPINK1 inactivation in DA neurons or all neurons was achieved by using Ddc-GAL4 or elav-GAL4 drivers, respectively. DA neurons were detected by immunofluorescent staining of whole-mount fly brains at 1 and 10 days of age by using an anti-tyrosine hydroxylase (TH) antibody (17). When dPINK1 was inactivated by using the elav-GAL4 driver, flies at 1 day of age showed little difference in the total number and distribution of DA neurons in brain compared to control flies expressing the elav-GAL4 driver alone, elav-GAL4 driven lacZ, or elav-GAL4 driven hPINK1 (Fig. 1 A–D). In contrast, a dramatic reduction in the number of DA neurons was seen in fly brains with dPINK1 inactivation at 10 days of age compared to age-matched controls (Fig. 1 E–H). Significant neuronal loss was observed in most DA neuron clusters including in PAL, PPM1/2, PPM3, PPL2, and less so in PPL1 (Fig. 1; see Fig. 3E). In contrast, the number of DA neurons in the VUM regions of the brain were not significantly affected, although the intensity of TH staining in this and other clusters seem to be reduced. These results agree with previously reported selective loss of DA neurons seen as early as 10 days of age in transgenic flies overexpressing human α-synuclein (18, 19). A similar loss of DA neurons was observed in flies with Ddc-GAL4 driven dPINK1 RNAi (data not shown). Loss of DA neurons in flies expressing dPINK1 RNAi in both cases is likely specific to the loss of dPINK1 function because it was rescued by coexpression of hPINK1 (Fig. 1K; see Fig. 3E). Staining of serotonergic neurons showed little change between flies expressing elav-GAL4 alone and elav-GAL4 driven dPINK1 RNAi (Fig. 1 L–O). Thus, inactivation of dPINK1 results in progressive and apparently selective degeneration of DA neurons in the Drosophila brain.

However, dPINK1 inactivation-mediated neurodegeneration is not restricted to DA neurons of the brain. Expression of dPINK1 RNAi in Drosophila eyes using the GMR-GAL4 driver resulted in age-dependent progressive ommatidial degeneration as manifested in black lesions in the external eyes that is rescued by coexpression of wild-type hPINK1 (Fig. 2A–E and Fig. 7, which is published as supporting information on the PNAS web site). Scanning electron microscopic imaging also revealed a “rough eye” phenotype, disorganized interommatidial bristles, and degeneration of ommatidia in flies with dPINK1 knockdown (Fig. 2 A′–D′). However, ommatidial degeneration induced by dPINK1 RNAi could not be rescued by coexpression of a PD-associated PINK1G309D mutant (Figs. 2 A″–E″ and 6D). The results suggest that mutant forms of PINK1 that are associated with PD in humans, unlike the wild-type counterpart, are no longer able to protect ommatidia with dPINK1-RNAi from degeneration. To elucidate the cellular manifestation of dPINK1-RNAi-induced ommatidial degeneration, we examined the arrangement and integrity of ommatidial cells and pupal retinal neurons 44 h after pupae formation (APF). Immunostaining for Discs Large (Dlg), a Drosophila membrane-associated guanylate kinase protein (MAGUK) (20), revealed morphologically disrupted ommatidia, increased and misoriented mechanosensory bristle groups, and disorganized and enlarged pigment cells after dPINK1 inactivation compared with controls (Fig. 2 F–I). Colabeling with phalloidin (revealing actin organization) and an anti-elav antibody (marking neuronal nuclei) shows a significant loss of photoreceptor neurons 44 h APF in pupal retinas with dPINK1-RNAi (Fig. 2 J–M). These results reveal that dPINK1 inactivation in the Drosophila eye induces degeneration of photoreceptors. This rough-eye and ommatidial degeneration phenotype appears to be autonomous to photoreceptors because dPINK1-RNAi driven by pan-neuronal elav-GAL4 was indistinguishable to that of GMR-GAL4 driven dPINK1-RNAi (Fig. 8, which is published as supporting information on the PNAS web site).

Rough eyes and the loss of ommatidia induced by dPINK1-RNAi is unlikely to occur via apoptosis, because TUNEL assay detected little signal in the pupal retina of both RNAi flies and their controls 44 h after pupae formation (data not shown). Moreover, expression of dIAP1, a Drosophila inhibitor of apoptosis protein (21), did not inhibit ommatidial degeneration induced by dPINK1-RNAi (Fig. 9, which is published as supporting information on the PNAS web site). Interestingly, expression of PD-associated PINK1 mutants potentiates oxidative stress-induced death of transfected SH-SY5Y cells in vitro (4). To investigate possible molecular mechanisms underlying neuronal loss induced by dPINK1 inactivation, we determined whether expression of SOD1 could prevent degeneration of DA neurons and ommatidia in flies with dPINK1 inactivation. SOD1 is an antioxidant enzyme found in the cytosol, nucleus, peroxisomes, and mitochondrial intermembrane space of eukaryotic cells (22). Expression of human SOD1, but not lacZ, in flies with dPINK1 knockdown under the control of the GMR-Gal4 driver markedly suppressed ommatidial degeneration (Fig. 3 A–D). Consistent with this finding, expression of human SOD1 driven by elav-GAL4 remarkably inhibited dPINK1 inactivation-induced degeneration of DA neurons (Fig. 3 A′–D′ and E). These results suggest that dPINK1 inactivation is likely to induce neuronal death via an oxidative stress pathway.

To further explore the involvement of oxidative stress in dPINK1 inactivation-dependent neurodegeneration, we treated GMR-GAL4 driven dPINK1-RNAi flies with compounds that exhibit antioxidant activity, including recombinant SOD1 protein (as in refs. 38 and 39) and vitamin E, by adding the antioxidants to the flies' diet. Treatment with SOD or vitamin E inhibited ommatidial degeneration in a dose-dependent manner (Fig. 4 A–C and A′–C′ and Fig. 10, which is published as supporting information on the PNAS web site). In GMR-GAL4 driven dPINK1 RNAi flies, SOD treatment also inhibited degeneration of DA neuron, especially of the PPL3 cluster (Fig. 11, which is published as supporting information on the PNAS web site). These results suggest that inactivation of dPINK1 results in oxidative stress in Drosophila and that specific antioxidants can suppress dPINK1 inactivation-induced neurodegeneration. Consistent with this notion, overexpression of hPINK1 results in flies with reduced sensitivity to treatment with paraquat, an environmental toxin linked to sporadic PD (23), and of H₂O₂ (Fig. 4 D and E), known inducers of oxidative stress. In contrast, expression of hPINK1 did not protect flies from damage induced by protein unfolding-promoting 2-mecaptoethanol (not shown). Together, these results indicate that PINK1 plays an important role in preventing oxidative stress-induced neuronal injury and death.

Our results provide strong in vivo evidence that PINK1 plays an important role in the survival of DA neurons. PD-associated PINK1 mutants are likely to fail in protecting neurons from stress and potentiate susceptibility to neuronal death. These results agree with previously reported inhibition of basal and staurosporine-induced death of SH-SY5Y cells by overexpressing wild-type PINK1 but not PD-associated PINK1 mutants in vitro (24). Our findings that SOD1 and antioxidant treatments suppress Drosophila neuronal death due to dPINK1 knockdown indicate that PINK1 in vivo acts to protect neurons that seem to be particularly susceptible to oxidative stress. The mechanism for this remains to be determined, but a potential clue may be the mitochondrial localization of PINK1 (4, 13, 25). Mitochondria are the main source of reactive oxygen and nitrogen species, and are the cellular compartments critical for integration of intrinsic death pathways. PD-associated PINK1 mutants may disrupt homeostasis of mitochondria, resulting in oxidative stress and eventual neuronal death. Alternatively, PINK1 could functionally collaborate with parkin and DJ-1, two other recessive, early-onset PD-linked genes, to regulate neuronal sensitivity to oxidative stress. In vitro, parkin and DJ-1 have been shown to protect cells from oxidative stress (26–31). Moreover, PINK1 interacts with DJ-1 and digenic mutations of PINK1 and DJ-1 are associated in early onset familial form of PD (15). Consistent with our findings that antioxidants can ameliorate the dPINK1-dependent PD pathology in Drosophila are two recent reports that show that deletion mutants of dPINK1 result in abnormally functioning mitochondria, sensitization to oxidative stress, and mild degeneration of DA neurons (32, 33). The phenotype of the dPINK1 deletion mutants in these studies appears to be milder, with no or less DA neuron loss than we observe. Possible reasons include: (i) maternal dPINK1present in zygotic mutants may be less in our “systemic” RNAi flies and thus cause earlier (larval) lethality, and importantly, (ii) the genetic background perhaps compounded by slight environmental differences may be sufficiently different to render our knock-down flies more sensitive to a reduced PINK1 function.

This study also introduces a fly model for Parkinson's disease that complements and extends previous models based on the expression of human α-synuclein or rotenone treatment (18, 34). Flies undergoing dPINK1 inactivation show degeneration of both DA neurons and ommatidia. The obvious ommatidial degeneration seen after dPINK1 inactivation should enable efficient screening of compounds preventing PD-related neurodegeneration and greatly facilitate identification of novel factors involved in PD pathogenesis.

Materials and Methods

Supplementary Material