Main

The mammalian HtrA (high temperature requirement) family of proteins show high homology to the bacterial chaperones DegS and DegP.1, 2 DegS, a serine protease, acts as a stress sensor by interacting with unfolded outer membrane porins via its PDZ domain.3 This releases DegS proteolytic activity, triggering a cascade that results in activation of stress response genes via the transcription factor σE. Mammalian HtrA2/Omi, is a mitochondrial protein with both proapoptotic and cell-protective roles.4 Its proapoptotic function is exerted through binding and cleavage of IAPs (inhibitor of apoptosis proteins) on its release into the cytoplasm following a proapoptotic stimulus.5, 6, 7, 8 In Drosophila, HtrA2 has also been reported to be released from the mitochondria on cellular insults such as UV irradiation, and to cleave DIAP1, the principal Drosophila IAP.9, 10, 11

However, genetic data have suggested that mammalian HtrA2, like its bacterial counterpart, normally functions as a stress response gene, preserving mitochondrial integrity. mnd2 (motor neuron degeneration) mice, which have an inactivating mutation in the HtrA2 protease domain, show muscle wasting and neurodegeneration.12 HtrA2 knockout mice have neuronal degeneration in a subset of striatal neurons and exhibit a parkinsonian phenotype, abnormal mitochondria and reduced lifespan.13

In further support of a protective role, growing evidence suggests a link between HtrA2 and Parkinson's disease (PD), a progressive neurodegenerative disorder of unknown aetiology. Two mutant alleles of HtrA2 (A141S and G399S) have been found in PD patients, leading to the classification of HtrA2 as PARK13 by OMIM.14 Although one of these genetic variants was later found in non-PD controls,15, 16 Bogaerts et al.17 identified a new mutation (R404W) in a large cohort of Belgian PD patients, confirming a role for HtrA2 in PD susceptibility. Importantly, recent studies have shown that HtrA2 forms a complex with the PD-related factor PINK1, a mitochondrial-targeted kinase.18 Moreover, HtrA2 is phosphorylated in a PINK1-dependent manner in response to p38 stress-activated protein kinase (SAPK) pathway activation, suggesting that PINK1 can modulate HtrA2 activity as part of mitochondrial stress response. Overexpression studies in flies suggest PINK1 and HtrA2 may be functionally related,19, 20 although their precise relationship remains unclear. We took advantage of Drosophila genetics to examine in vivo the function of HtrA2 in the PINK1 pathway and assess its putative role in apoptosis.

Results

The Drosophila HtrA2 homologue, encoded by CG8464, contains a predicted N-terminal mitochondrial targeting sequence (MTS), a transmembrane domain (TM), a central protease domain, a C-terminal PDZ domain and an unconventional IAP-binding motif (Supplementary Figure S1a).10 HtrA2 encodes a full-length protein of ∼46 kDa. On mitochondrial import, HtrA2 is cleaved to yield two products of 37 and 35 kDa. We have expressed and purified HtrA2 in bacteria and tested its activity towards an HtrA2 fluorescent peptide substrate (H2-Opt) as well as a control peptide as previously described.8 These experiments revealed that HtrA2 efficiently cleaves the H2-Opt substrate but not a control peptide, suggesting that Drosophila HtrA2 has similar substrate specificity to its mammalian homologue (Supplementary Figure S1b).

To address the in vivo function of HtrA2, we generated a mutant allele by imprecise P-element excision of G4907 (Genexel Inc.), which is inserted between HtrA2 and mRpL11 (mitochondrial ribosomal protein-like 11; CG3351; Figure 1a). The G4907 stock is homozygous lethal and could be mutant for either mRpL11 or HtrA2, or for both. We mobilized G4907 and generated a deletion removing 1037 bp from the insertion site to exon 1 of HtrA2 leaving 8 bp of exon 1 (Figure 1a). This allele (HtrA2Δ1) is homozygous lethal and lacks both the transcription and translation start sites of HtrA2, as well as the promoter region and transcription start site of mRpL11, and is therefore likely to be a null allele for both genes. We also recovered a precise excision as a control.

Figure 1
figure 1

Generation of HtrA2 mutants. (a) Schematic representation of the HtrA2 genomic region. The P-element G4907 (black triangle) was excised to produce a deletion of 1037 bp (red line). Position of start (ATG) and stop (*) codons are indicated. Lengths of the genomic rescue constructs are indicated. (b) The HtrA2Δ1 deletion mutant was balanced over TM3, primers from the genomic rescue construct were used to amplify a 2.5 kb (WT) product from TM3 and the resulting 1.8 kb deletion product (deletion). The positive control (+) was the amplification of the genomic product from wiso flies; negative control (−, no DNA). (c) RT-PCR shows the absence of HtrA2 transcript in mutants (HtrA2Δ1). The HtrA2 transcript can be amplified from the genomic rescued line (HtrA2Δ1 rescue), negative control (−, no RNA). (d) Western blot shows the absence of HtrA2 protein in mutants (HtrA2Δ1) and its presence in control and rescue flies

Full size image

We generated genomic rescue constructs for both HtrA2 and mRpL11 (gHtrA2 and gmRpL11; Figure 1a; Materials and Methods). The lethality of HtrA2Δ1 was rescued by gmRpL11 but not by gHtrA2, indicating that mRpL11 is required for viability, as expected for a ribosomal component. HtrA2Δ1 homozygotes bearing a gmRpL11 rescue transgene represent null alleles solely for HtrA2, indicating HtrA2 is non-essential.

Therefore, to further investigate the phenotypes of HtrA2 mutant flies, we compared flies mutant for HtrA2 and mRPL11 bearing two copies of the gmRpL11 construct (gmRpL11; HtrA2Δ1), which will be referred to as ‘HtrA2Δ1’, with precise excision control flies as a wild type. As an additional control, reversion of phenotypes was tested in flies carrying the HtrA2Δ1 deletion and bearing two copies each of rescue constructs for both mRpL11 and HtrA2 (gmRpL11, gHtrA2; HtrA2Δ1), which will be referred to as ‘HtrA2Δ1 rescued’.

The HtrA2 transcripts are expressed ubiquitously and uniformly in all the tested tissues (Supplementary Figure S2a–f). The precise excision, HtrA2Δ1 and HtrA2Δ1 rescued flies were tested by genomic PCR, RT-PCR and western blot to confirm the deletions and the absence of both transcript and HtrA2 protein in mutants (Figure 1b–d).

Several reports have suggested that HtrA2 can promote apoptosis by cleaving DIAP1, thus releasing the apical caspase Dronc in response to proapoptotic stimuli.9, 10, 11 We therefore monitored cell death using an antibody directed against cleaved Caspase 3 in wing imaginal discs treated with γ-rays (4Gy), staurosporine (STS – 4 μM) and ultraviolet (UV) light (2.5 kJ/m2). Apoptosis induction was identical in control (precise excision) and HtrA2Δ1 mutant discs, both without treatment and on treatment (Figure 2a–f). In addition, the development of HtrA2Δ1 animals is completely normal, in particular, showing normal eye morphology, indicating that HtrA2 is dispensable for retinal apoptosis during development (data not shown). This is consistent with findings in HtrA2 knockout mice13 and suggests that HtrA2 is dispensable for developmental and apoptosis induced by multiple stimuli in vivo.

Figure 2
figure 2

Apoptosis is normal in HtrA2 mutants. (a–d) Confocal micrographs of the pouch region of wing imaginal discs stained with anti-cleaved caspase 3 (red) and phalloidin (green) after mock treatment or exposure to 4Gy γ-rays. No difference was detected between precise excision controls (a, b) and HtrA2Δ1 mutants (c, d). Posterior is to the right. Scale bar=50 μm. (e) Quantification of the apoptotic index from wing discs after treatment with γ-rays (see Materials and Methods). n=10 for each genotype. (f) Quantification of the apoptotic index from wing discs after treatment with Staurosporine or UV light (see Materials and Methods). n=8 for each genotype. Error bars represent standard deviations

Full size image

To determine whether HtrA2 plays a role in maintenance of cellular homoeostasis by the PINK1 pathway, we assessed whether HtrA2Δ1 flies share phenotypic similarities with PINK1 mutants. Drosophila PINK1 mutants exhibit locomotor defects in flight and climbing behaviours, widespread degeneration of indirect flight muscles (IFMs) with accompanying mitochondrial morphology defects, male sterility and partial loss of dopaminergic neurons.21, 22, 23 Strikingly, all these phenotypes are phenocopied in Drosophila mutants of another PD-linked gene, parkin,24, 25, 26, 27 which was shown to act genetically downstream of PINK1.21, 22, 23 Thus, we investigated such phenotypes in HtrA2 mutants.

Myopathology was investigated using toluidine blue staining of IFMs in longitudinal sections of adult thoraces (Figure 3a–c). Although marked myopathology akin to that described in parkin and PINK1 mutants was not observed (in 5- or 30–day-old animals), more detailed investigation will be needed to determine whether Htr2A dysfunction leads to overt myodegeneration. However, examination of IFM ultrastructure from aged HtrA2Δ1 mutants revealed increased numbers of defective mitochondria, displaying reduced electron density and open cristae, compared with wild-type and HtrA2Δ1-rescued controls (Figure 3d–g; Supplementary S3). It should be noted that the mitochondrial abnormalities seen in HtrA2Δ1 mutants are not widespread, suggesting this does not reflect a general mitochondrial defect. To confirm this, we measured mitochondrial activity by high-resolution respirometry in whole flies using 30-day-old males. Coupled mitochondria from WT and HtrA2Δ1 showed no difference in oxygen consumption with substrates of complex I and complex II in the presence of ADP (Supplementary Figure S4a, b). This was confirmed by measuring respiratory activities of complex II and complex IV under uncoupled conditions (Supplementary Figure S4c, d). Thus, there is no systemic mitochondrial deficiency in HtrA2Δ1 mutants compared with controls.

Figure 3
figure 3

HtrA2 mutants show mild mitochondrial disruption and progressive locomotor deficits. (a–c) Toluidine blue-stained, whole thorax sections of 30-day-old flies revealed that overall IFM structure is maintained in HtrA2Δ1 mutant flies. (d–f) IFM mitochondria in TEM micrographs exhibit morphological abnormalities, typically reduced electron density and disrupted cristae in HtrA2Δ1 mutant (arrow in e), compared to wild type (d) and HtrA2Δ1 rescue flies (f). (g) Quantification of the mitochondrial phenotype from multiple individuals. Bars show mean and standard error (see Materials and Methods). HtrA2Δ1 mutant flies have an increased proportion of mitochondria with reduced electron density, as determined by a Fisher's exact test (*P<0.01, #=0.058). HtrA2Δ1 mutants show progressive deficits in flight (h) and climbing (i). Five-day-old mutant flies have wild-type flight and climbing ability. Aged (30–day-old) mutants show reduced flight and climbing, which is restored in HtrA2Δ1 rescue flies. Bars show mean and standard error, inset numbers show n. **P<0.001; *P<0.01, by Kruskal–Wallis non-parametric analysis and Dunn's pair-wise comparison

Full size image

To determine if, like PINK1 and parkin mutants, HtrA2Δ1 mutant flies present locomotor deficits, HtrA2Δ1 mutants were assayed for climbing and flight ability. HtrA2Δ1 mutants showed wild-type flight and climbing ability at 5 days of age (Figure 3h and i). However, 30-day-old HtrA2Δ1 flies show significantly reduced flight and climbing ability compared with wild-type flies. This defect was completely restored in HtrA2Δ1 rescued flies (Figure 3h and i). Thus, the progressive mitochondrial morphological abnormalities seen in HtrA2Δ1 mutants correlate with progressive motor deficits.

In further similarity to parkin/PINK1 mutants, we also find that HtrA2Δ1 males are sterile, whereas female fertility is unaffected. HtrA2Δ1 males are almost completely sterile (4% fertility compared to 100% fertility in control flies; Figure 4a). Importantly, male fertility is restored to wild-type levels in HtrA2Δ1 rescued lines (Figure 4a). However, although PINK1/parkin mutants exhibit distinct morphological defects during spermatogenesis,21, 22, 27, 28 the ultrastructural analysis of HtrA2Δ1 testes revealed no observable defects (Figure 4b). Indeed, mature individualized sperm are formed properly, however, we noted that HtrA2Δ1 sperm were completely immotile, as was the case for PINK1 mutants (see Supplementary movies S1–3 in Supplementary materials).

Figure 4
figure 4

HtrA2Δ1 have normal spermatogenesis but are male sterile. (a) HtrA2Δ1 males have greatly reduced fertility compared to controls. Normal fertility is regained in HtrA2Δ1 rescued flies. Inset numbers represent n value. (b) TEM micrographs of mature individualized spermatids show no structural defects in the axoneme or nebenkern. Scale bar shows 1 μm and 0.2 μm in upper and lower panels, respectively

Full size image

Recent evidence indicates HtrA2 likely plays a role in cellular stress response mechanisms.18 To test this we determined the effect of loss of HtrA2Δ1 function on longevity and stress resistance. HtrA2Δ1 mutants have a reduced lifespan compared to rescued or precise excision controls (Figure 5a). HtrA2Δ1 flies were also sensitive to paraquat-induced oxidative stress and mitochondrial respiration deficit induced by complex I inhibitor rotenone (Figure 5b and c).

Figure 5
figure 5

HtrA2Δ1 have reduced survival under normal and stressed conditions. (a) Male HtrA2Δ1 flies (red) have reduced lifespan compared with wild-type control (blue), which is restored with a genomic rescue construct (black). n=75 for each genotype. (b) HtrA2Δ1 mutant flies (red) are less resistant to oxidative stress resulting from paraquat ingestion compared to wild-type controls (blue) or rescued flies (black). n=200 for each genotype. (c) HtrA2Δ1 mutant flies are more sensitive to rotenone-induced oxidative stress (red) than wild-type controls (blue) or the HtrA2Δ1 rescued line (black). n=200 for each class. Data are presented as Kaplan–Meier survival distributions, Log-rank tests were performed to determine statistical significance. Individual P-values are shown with the figure

Full size image

Given the evidence linking HtrA2 to PD, we determined whether the dopaminergic neurons are affected in HtrA2Δ1 flies. Although parkin and PINK1 mutants show age-dependent dopaminergic neuron loss,21, 23, 25 we found 30–day-old HtrA2Δ1 flies showed no difference in number of dopaminergic neurons (Figure 6). However, this is also consistent with the weaker phenotypes in other tissues.

Figure 6
figure 6

Loss of HtrA2 does not cause dopaminergic neurons loss. Number of neurons staining positive with anti-tyrosine hydroxylase antiserum from aged (30-day) precise excision control (black) and HtrA2 mutants (grey). No difference is seen in any of the clusters. Bars show mean and standard error (n⩾10)

Full size image

Thus, HtrA2Δ1 mutant flies share a number of phenotypic similarities with parkin/PINK1 mutants, though generally the HtrA2Δ1 phenotypes are weaker. To address whether HtrA2 functions in a common or divergent pathways to PINK1/Parkin, we generated double mutant combinations and determined the effect on climbing ability. Previously, PINK1 and Parkin have been reported to act in a common pathway, consistent with the observation that PINK1:parkin double mutants do not have a more severe phenotype than either of the single mutants.21, 22, 23 Corroborating these findings we find PINK1:parkin double mutants do not show a difference in climbing ability compared to single mutants of either PINK1 or parkin (Figure 7a). Similarly, we find PINK1:HtrA2 double mutants show no difference in climbing ability compared with PINK1 mutants (Figure 7b), consistent with them acting in a common pathway. In contrast, we find that HtrA2:parkin double mutants show a dramatically enhanced climbing defect in comparison to parkin mutants (Figure 7c), suggesting that HtrA2 might act in a different pathway from Parkin. The PINK1 mutant defects are rescued by overexpression of parkin, leading to the suggestion that Parkin acts downstream of PINK1.21, 22, 23 We therefore generated transgenic animals expressing HtrA2 under GAL4/UAS control and tested its ability to rescue PINK1 phenotypes. Interestingly, ubiquitous expression of the HtrA2 transgene significantly rescued the PINK1 climbing defects (Figure 8). Together, these results suggest that HtrA2 likely plays a role in the PINK1 pathway.

Figure 7
figure 7

HtrA2 mutants enhance the locomotor deficit exhibited by parkin mutants but not PINK1 mutants. (a) Climbing ability of PINK1B9, park25 and PINK1B9:park25 double mutants. (b) Climbing ability of HtrA2Δ1, PINK1B9 and PINK1B9:HtrA2Δ1 double mutants. (c) Climbing ability of HtrA2Δ1, park25 and park25:HtrA2Δ1 double mutants. Bars show mean and standard error (n⩾50). ***P<0.0001

Full size image
Figure 8
figure 8

Overexpression of HtrA2 can rescue PINK1 climbing defect. Compared to heterozygous controls, PINK1 mutants have reduced climbing ability, which is suppressed by overexpression of HtrA2 under the control of the actin5c promoter. Bars show mean and standard error; n>50. Significance was established using a Kruskal–Wallis test and Dunn's comparison (***P<0.0001; *P<0.01)

Full size image

Discussion

We have characterized the Drosophila homologue of the HtrA2 protease and found that HtrA2 mutants share some phenotypic characteristics with previously described Drosophila models of PD. HtrA2 null mutants are viable but exhibit mild mitochondrial defects, loss of flight and climbing ability, male infertility, and sensitivity to oxidative stress and mitochondrial toxins (Figures 3, 4, 5 and 6). In contrast, we find that HtrA2 is not required for developmental or stress-induced apoptosis (Figure 2). These data are consistent with findings in mice and humans, which suggest that, HtrA2 functions primarily to maintain mitochondrial integrity and protect cells against oxidative stress,12, 13 and contradicts previous reports that HtrA2 is a proapoptotic factor.9, 10, 11

Recently, Yun et al.20 have reported a characterization of ethanemethylsulphonate-induced alleles of Drosophila HtrA2/Omi. In accordance with our results, they show that HtrA2 is required for male fertility, normal lifespan and stress resistance. They do not report a mitochondrial defect, though the mild nature of the defect observed in our hands and allelic or environmental differences might explain this discrepancy. Both of our reports suggest that HtrA2 mutants have a markedly weaker phenotype than PINK1, which is consistent with the idea that HtrA2 does not play a critical role in the PINK1-Parkin interaction. However, our genetic interaction studies presented here, in conjunction with our previous evidence,19 support the view that HtrA2 acts downstream of PINK1 in a divergent pathway parallel to Parkin.

We have previously shown that a PINK1 overexpression phenotype is partially suppressed by loss of HtrA2, whereas conversely an HtrA2 overexpression phenotype is not suppressed by PINK1 mutations,19 suggesting that HtrA2 functions downstream of PINK1. In a complementary approach using a loss of function phenotype, we show here that overexpression of HtrA2 can partially substitute for loss of PINK1 (Figure 8). These findings are consistent with our previous report and suggest that HtrA2 acts downstream of PINK1. Although the study of Yun et al. did not uncover evidence for HtrA2 acting downstream of PINK1, technical differences, particularly reflecting transgenic expression levels, may account for this discrepancy.

In agreement with Yun et al. we have shown that, similar to PINK1:parkin double mutants, PINK1:HtrA2 double mutants display an identical phenotype to PINK1 mutants alone, suggesting they act in a common pathway (Figure 7). However, we also showed that parkin:HtrA2 double mutants display a stronger phenotype than either mutant alone (Figure 7), suggesting HtrA2 acts in parallel pathway to Parkin. These results are also consistent with our previous findings that loss of either HtrA2 or parkin can partially suppress a PINK1 overexpression phenotype, whereas attenuating both genes together is sufficient to completely suppress the overexpression of PINK1.19 Together these findings support the notion that HtrA2 and Parkin are acting as downstream effectors of PINK1 but in parallel pathways. The relative weakness of the HtrA2 mutant phenotype suggests that it is not the predominant PINK1 effector.

The most pressing questions remain how this pathway is triggered and how its activation leads to protection from mitochondrial stress. By analogy to bacterial DegS, it is possible that the PDZ domain of HtrA2 acts as a sensor for unfolded proteins. In addition, we and others have shown that the mitochondrial protease Rhomboid-7/PARL can process both PINK1 and HtrA2,19, 29 which may represent a regulatory process. Lastly, the p38 SAPK pathway is known to be triggered by reactive oxygen species, which are produced when mitochondrial electron transport is perturbed.30 Mitochondrial dysfunction may therefore modulate HtrA2 activity at several levels. Once activated, HtrA2, like DegS, may cleave unfolded protein and/or elicit a transcriptional response to clear the damaged proteins.4 Further work will be required to fully elucidate the functional relationship between PINK1 and HtrA2, and the involvement of this putative pathway in PD pathogenesis.

Materials and Methods

Drosophila genetics

HtrA2 mutants were created by imprecise excision of P-element G4907 (Genexel Inc.). Breakpoints were mapped by genomic PCR and sequencing. PINK1B9 mutants were a kind gift from J Chung.21 park25 mutants have been described previously.27 All fly experiments were carried out at 25 °C and on the same food batch for each experiment.

For crosses to combine PINK1 mutants with GAL4/UAS transgenes, paternal males with marked X chromosomes (y− or FM6) were used to ensure correct identification of PINK1 mutant progeny.

Fertility assays

Single 1-day-old males of each genotype were individually placed in a vial with three virgin females. The number of vials with larvae was recorded after 5 days.

Flight/climbing assay

Flight/climbing assays were performed as previously described.27 (n⩾30 per genotype, unless otherwise stated within figures.) Statistical significance was calculated by Kruskal–Wallis non-parametric analysis and Dunn's pair-wise comparison.

Longevity assay

A total of 50, 0- to 4-h-old males and females of each genotype were placed into separate vials in groups of 25 for each gender. The flies were transferred to fresh food every 2 days and the number of live flies recorded. Data are presented as Kaplan–Meier survival distributions and significance determined by Log-rank tests.

Oxidative stress assay

The procedure used for longevity assay was repeated for oxidative stress assays using food containing 20 mM paraquat or 20 mM rotenone. Data are presented as Kaplan–Meier survival distributions and significance determined by Log-rank tests.

Molecular biology

RT-PCR

Total RNA was extracted using Trizol reagent (Invitrogen, Carlsbad, CA, USA). RT-PCR was performed using the kit SuperScript III One-step RT-PCR with Platinum Taq (Invitrogen). The 5′ and 3′ sequences used to amplify the dHtrA ORF are ATGGCTTTGCGCGGTTCC and CTAGGGATCTTCTGGCGTA , respectively. Primers used to amplify 5′ and 3′ dPINK1 ORF are ATGTCTGTGAGACTGCTGAC and CTACGCCACCACATTCTGGA , respectively.

To detect transcripts by RT-PCR in mutants, 5′ primer GTTGTGCCGCTGTCGTCTAT and 3′ primer CGTCGTGCGAAACGTACG were used to amplify a product of 1 kb between exon 2 and exon 3, respectively.

Genomic rescue constructs

The predicted full-length HtrA2 gene, 700 bp of upstream promoter region (including the complete 3′ UTR), was amplified by PCR using 5′ primer CAACTCGAGGAAGTACATTGGGCGGGTC and 3′ primer GGGACTAGTGGGTTTGTCAGCGATTTC , sequenced and subcloned into pCaSpeR-HS using XhoI and SpeI restriction sites. mRPL11 was constructed by including 463 bp of upstream promoter region, the entire predicted gene and 108 bp of downstream genomic region using 5′ primer GGGTCTAGAGCAGCTGATTTCAATTTGGC with XbaI restriction site and 3′ primer GGGGAATTCCACTGGAAAACTCAACAAGC with EcoRI restriction site, sequenced and subcloned into pCaSpeR-HS.

pUASp construct

The dHtrA2 ORF was amplified using 5′ primer TCACGCGGCCGCATGGCTTTGCGCGGTTCC containing NotI restriction site, and 3′ primer ACAGCTCGAGCTAGGGATCTTCTGGCGTA containing XhoI restriction site. The sequenced, digested product was subcloned into pUASp. Western blotting was performed by standard procedures. The anti-HtrA2 antibody (used at 1:1000) was a kind gift from Dr. E. Alnemri (Kimmel Cancer Institute, USA).

Cell biology

Tissue sectioning

Thoraces and testes were prepared from 30- and 2-day-old adult flies, respectively, and fixed overnight in 2% paraformaldehyde, 2.5% gluteraldehyde in sodium cacodylate (0.1 M). Thoraces were rinsed in sodium cacodylate (0.1 M) and 1% tannic acid and postfixed in 1:1 2% OsO4 and sodium cacodylate (0.2 M) for 1 h. After rinsing, thoraces were dehydrated in an ethanol series and embedded using Spurr's resin. Semithin sections were then taken and stained with toluidine blue, whereas ultrathin sections were examined using TEM. Quantification of the mitochondrial phenotype was carried out under blinded conditions for both imaging and analysis. Ten randomly selected fields of view were captured per individual and the number of mitochondria exhibiting a disrupted morphology phenotype and those with normal mitochondria were counted. Approximately 1500 mitochondria were examined per genotype. Pair-wise differences in proportion of mitochondria exhibiting the phenotype were made utilizing the conservative Fisher's exact test.

Dopaminergic neuron counts

Brains were dissected from 30-day-old flies and fixed in 4% paraformaldehyde, rinsed in PBS and incubated with anti-TH antibody (Immunostar Inc., Hudson, WI, USA) in PBS-T BSA (0.5%) overnight (4 °C). After washing with PBS-T, brains were incubated (2 h) with Alexafluor 488 secondary antibody (Invitrogen) in PBS-T BSA (0.5%). After washing, brains were mounted and TH-positive neurons imaged by confocal microscopy. In all cases, TH-positive neurons were counted with the investigator blinded to the genotypes.

Apoptosis assay

Third instar wing imaginal discs were exposed to 4 Gy γ-rays, fixed and stained with anti-caspase 3 after 2 h (Cell Signaling Technologies, Danvers, MA, USA) and FTIC-phalloidin (Sigma, Highland, IL, USA) as previously described.31 For STS (4 μM) and UV treatment (2.5 kJ/m2), larvae were dissected and discs were incubated in Schneider's medium (Invitrogen), then fixed and stained after 4 h. Images were acquired on a Zeiss LSM 510 confocal microscope. The apoptotic index corresponds to the percentage of caspase 3-positive pixels relative to the total area, and was determined as described using Image J.31 For consistency, the analysis was performed on the pouch region of the wing discs.