DIAP1 suppresses ROS-induced apoptosis caused by impairment of the selD/sps1 homolog in Drosophila

Apoptosis is an essential cellular process during the morphogenesis of multicellular organisms and its genetic control has become one of the main topics of study for developmental and cell biologists. However, it is not yet well understood how certain processes such as oxidative stress can trigger specific pathways of apoptosis. Aerobic metabolism uses molecular oxygen as a terminal electron acceptor for mitochondrial respiratory energy production. Reactive oxygen species (ROS) are generated as by-products of this process. These are a variety of oxygen metabolites that have either unpaired electrons (i.e.

⁠, OH·) or the ability to abstract electrons from other molecules (i.e. hydrogen peroxide). Transient fluctuations in ROS serve important regulatory functions, but when present at high and/or sustained levels, they can cause severe damage to DNA, proteins and lipids, which may finally lead to apoptosis (Simon et al., 2000; Curtin et al., 2002).

A number of defense systems have evolved to counteract the persistent state of oxidative siege associated with aerobic life conditions and among them enzymatic intracellular scavengers play an essential role. In this category, mammalian selenoproteins, which contain selenium in the form of selenocysteine, are important for the control of unwanted ROS as most of them catalyze oxidation-reduction reactions or act as antioxidants (Stadtman, 1996; Rayman, 2000; Hatfield and Gladyshev, 2002). The machinery of selenocysteine incorporation is conserved in bacteria, archaea and eukarya and depends on reading the inframe UGA stop codon as the amino acid selenocysteine and on the recognition of a downstream stem loop structure (Low and Berry, 1996; Stadtman, 1996). Four genes (selA-D) of Escherichia coli are essential for selenocysteine synthesis, codon recognition and polypeptide elongation. One of them, the selD gene, encodes for selenophosphate synthetase which catalyses a selenide-dependent ATP hydrolysis reaction to generate monoselenophosphate, the selenium intermediate necessary for the synthesis of selenocysteine. In flies and mammals two highly conserved selD genes have been identified (Low et al., 1995; Guimaraes et al., 1996; Persson et al., 1997; Alsina et al., 1998), sps1 (selD in flies) and sps2. Two effects have been described for a mutation in the Drosophila selD/sps1 gene, perturbation of selenoprotein biosynthesis and accumulation of ROS (Alsina et al., 1999). This mutation, hereafter called selD^ptuf, is a recessive null mutation and homozygous individuals have extremely reduced and abnormal imaginal discs and die as third instar larvae (Alsina et al., 1998; Alsina et al., 1999). Heterozygous selD^ptuf flies are apparently wild type, however, a downregulation of the Ras/MAPK pathway has been observed in transheterozygous combinations of the selD^ptuf mutation and activated members of this signaling pathway (Morey et al., 2001). This downregulation seems to be a consequence of an increase in ROS levels in heterozygous flies (Morey et al., 2001; Morey et al., 2003). Both loss of survival signals and perturbations in the redox balance, among others, are intracellular stimuli that have been shown to trigger apoptosis in mammals (Kroemer and Reed, 2000; Adrain and Martin, 2001). In Drosophila, the cell death genes reaper (rpr), head involution defective (hid) and grim are potent activators of caspase-dependent apoptosis (reviewed by Richardson and Kumar, 2002). Extensive research has been carried out to uncover how distinct death-inducing stimuli converge to activate a common apoptotic program. Both rpr and grim are expressed in cells doomed to die (White et al., 1994; Chen et al., 1996). In contrast, hid is expressed not only in cells that die, but also in living cells (Grether et al., 1995). Thus, these different expression patterns imply that they integrate different signals regulating apoptosis. Survival signals regulate Drosophila apoptosis and several studies have shown the need for Ras/MAPK activity for cell survival in flies (Simon et al., 1991; Diaz-Benjumea and Hafen, 1994; Freeman, 1996; Miller and Cagan, 1998; Sawamoto et al., 1998; Halfar et al., 2001). In the subset of hid-expressing cells prone to die, downregulation of the Ras/MAPK pathway increases hid expression and activity (Bergmann et al., 1998; Kurada and White, 1998). In the case of rpr, besides inducing developmentally programmed cell death, it also triggers apoptosis in response to other stimuli such as aberrant development, steroid hormone signaling and X-irradiation (Asano et al., 1996; Nordstrom et al., 1996; Robinow et al., 1997). rpr is also a transcriptional target of the Drosophila p53 protein, making its expression responsive to genotoxic stress caused by X-irradiation (Brodsky et al., 2000). Because X-irradiation, in addition to direct DNA damage, generates ROS in the aqueous cytoplasm that can also damage DNA, the Dmp53/Rpr pathway is a candidate pathway to be activated in a situation of increased ROS levels such as in the selD^ptuf mutant.

We have taken a genetic approach to study how an increase in ROS levels due to impairment of the selD/sps1 function triggers apoptosis in Drosophila imaginal discs. Our results indicate that hid-induced apoptosis may not be the major contributor to the apoptosis observed in selD^ptuf mutant cells and that ROS increase plays an important role in selD^ptuf apoptosis through the activation of the Dmp53/Rpr pathway. We clearly show that this apoptotic pathway is mediated by DRONC and DRICE caspases and that the inhibitor of apoptosis DIAP1 is able to rescue the viability of selD^ptuf cells. This work supports the importance of selD/sps1 in the maintenance of cellular viability and demonstrates that ROS-induced apoptosis triggered by the loss-of-function of selD/sps1 is caspase dependent and activated by Dmp53/Rpr function.

The selD^ptuf line (yw; l(2)k11320/CyO) was obtained from a collection of lethal mutants resulting from PlacW insertions on the second chromosome of Drosophila melanogaster (Torok et al., 1993; Alsina et al., 1998). The following stocks were used: yw; Catⁿ¹/TM3 (Mackay and Bewley, 1989; Griswold et al., 1993); e ftz ry/TM3 P(sev-rasV12) (Fortini et al., 1992); Sco/SM1 P(GMRhid) (Kurada and White, 1998); GMRyan^Act on the second chromosome (Rebay and Rubin, 1995); rprlacZ on the third chromosome (Nordstrom et al., 1996); UAS-DIAP1/TM2 (Lisi et al., 2000); GMR-GAL4 UAS-dronc #80 (GMR-GAL4 UAS-dronc/CyO; UAS-dronc) (Quinn et al., 2000); UAS-Sem on the third chromosome (Martin-Blanco, 1998); +; gl-Dmp53/SM6aTM6B (Ollman et al., 2000) (Exelixis Inc.); hh-GAL4/TM2 (Tanimoto et al., 2000); 2xarm-GAL4/TM3Sb, 69B-GAL4 on the third chromosome (Bloomington Stock Center).

To prepare scanning electron microscopy (SEM) samples, flies were dehydrated in 25, 50, 75 and 100% ethanol for 24 hours each. Flies were critical point dried and coated with gold to be examined in a Leica-360 scanning electron microscope.

selD^ptuf clones in the adult eye were generated using the eyFLP/FRT technique coupled to a cell lethal mutation (cl2R11) (Newsome et al., 2000), which kills the twin clone allowing the growth of more mutant tissue. Adults of the genotype yw eyFLP; FRT42D w+ cl2R11/FRT42D selD^ptuf were examined for mutant clones (two copies of the mini-w marker from the PlacW insertion in the selD locus in a mini-w/ w+ background) in the eye. Histological sections of the eyes were prepared as described previously (Basler and Hafen, 1988). selD^ptuf mitotic clones in imaginal discs were generated using the hsFLP/FRT technique (Xu and Rubin, 1993) in combination with the Minute (M) technique (Morata and Ripoll, 1975), which gives proliferative advantage to the mutant tissue minimizing perdurance of selD product. Larvae of the yw hsFLP; FRT42D arm-lacZ M/FRT42D selD^ptuf genotype were heat shocked at 60 hours after egg laying (AEL) for 30 minutes at 34°C. For rpr transcription in selD^ptuf clones, 60 hours AEL larvae of yw hsFLP; FRT42D πMyc M/FRT42D selD^ptuf; rpr-lacZ/+ genotype were heat shocked for 1 hour at 37°C and πMyc was induced as described previously (Xu and Rubin, 1993). In these experiments the homozygous mutant tissue was marked by the absence of β-galactosidase or πMyc staining.

To determine the clonal size in wing discs, clones were induced by a 10-minute heat shock pulse at 34°C in yw hsFLP; FRT42D armlacZ/FRT42D selD^ptuf; hh-GAL4/UAS-DIAP1 larvae (4-hourly egg collections) at 55 hours AEL and analyzed at 120 hours AEL.

Imaginal discs from third instar larvae were dissected in PBS and fixed for 20 minutes in 4% paraformaldehyde (PFA) at room temperature (for β-galactosidase staining) or PLP (2% PFA, 75 mM lysine; 10 mM sodium periodate, 37 mM sodium phosphate) at 4°C (for πMyc staining). Following permeabilization, discs were incubated overnight with primary antibodies: rabbit anti-β-galactosidase 1:1000 (Cappel), mouse anti-β-galactosidase 1:250 (Promega) or mouse anti-πMyc 1:1000 (Babco) to mark clones, and mouse anti-Boss (Cagan et al., 1992) 1:2000 (a gift from S. L. Zipursky), mouse anti-Elav (Robinow and White, 1988) 1:100 (from DSHB), mouse anti-Dmp53 (Ollmann et al., 2000) 1:250 (Exelisis Inc.), rabbit anti-active DRICE (Dorstyn et al., 2002) 1:1000 (a gift from B. Hay) and mouse anti-FASIII (Brower and Jaffe, 1989) 1:1000 (a gift from D. Brower). Rhodamine Red- and FITC-conjugated secondary antibodies 1:200 (Jackson ImmunoResearch) were used. YOYO® nuclear marker (1:5000, Molecular Probes, Inc.) was added together with the secondary antibody. Discs were mounted in Slowfade® Light antifade (Molecular Probes, Inc.) and images collected using a Leica TCS 4D confocal laser scanning microscope. Negative controls for these antibodies in selD^ptuf discs and clones were performed in parallel and scanned under the same conditions and showed no staining. All images were processed with Adobe PhotoShop 6.0. and ImageJ 1.29v (National Institute of Health, USA).

After fixation of yw hsFLP; FRT42D arm-lacZ M/FRT42D selD^ptuf discs, apoptotic cells were detected by labeling the 3′-OH ends of DNA with Chromatide BODIPY® Texas Red-14-dUTP (Molecular Probes, Inc.) for 1 hour 30 minutes at 37°C using terminal deoxynucleotidyl transferase (Roche). Primary and secondary antibodies were used to detect clones as described above.

Clones to test for accumulation of ROS were generated in yw hsFLP; FRT42D GFP/FRT42D selD^ptuf larvae after heat shock at 48 hours AEL for 30 minutes at 34°C. Third instar larvae discs were dissected in Schneider's medium. Staining was performed in medium containing 20 μM dihydroethidium (DHE, Molecular Probes, Inc.) for 5 minutes, and a series of washes in Schneider's medium were performed to protect the sample from light before mounting in antifade solution. The samples were examined by confocal microscopy just after mounting.

Drosophila eye development is a paradigm to understand the cellular mechanisms that coordinately promote cell differentiation and survival. During larval development, cell-to-cell interactions in the eye imaginal disc shape the ommatidia, the units that form the insect compound eye. The Ras/MAPK pathway controls differentiation and survival of ommatidial cells (Freeman, 1996; Halfar et al., 2001), and a wide variety of molecular markers are available to verify these processes. In addition, the role of apoptosis in shaping the eye has been extensively described (Brachmann and Cagan, 2003). We have used the eye imaginal disc as a model to explore how apoptosis is triggered in oxidative stress conditions. We first tested whether selD^ptuf homozygous condition perturbs cell survival and/or normal differentiation in the developing eye. Because the homozygous selD^ptuf individuals are not viable (Alsina et al., 1998; Roch et al., 1998), we have analyzed the mutant condition in genetic mosaics. Clones of selD^ptuf mutant cells were generated in the eye disc and recovered to adulthood. These clones resulted in scarred tissue with poor ommatidial differentiation (Fig. 1A). Tangential sections revealed that no or very few photoreceptors were present in the mutant area, suggesting that extensive cell death had occurred during the development of the mutant sector (Fig. 1B). We also analyzed whether recruitment and differentiation of photoreceptor cells was taking place. Although the R8 photoreceptor-specific marker Boss revealed that R8 differentiation and spacing occurred properly (data not shown), the recruitment of the subsequent photoreceptors was disrupted, as revealed by the pattern of the neuronal marker Elav (Fig. 1C). Mutant ommatidia contained less cells than their neighboring wild-type heterozygous ones. Besides aberrant ommatidial organization, the epithelium of selD^ptuf was severely disrupted. As it has been described that disorganization and loss of polarity in epithelial tissues precedes cell death (Tepass et al., 1990), we checked for apoptosis by TUNEL staining and indeed detected a significant increase in apoptotic cells within selD^ptuf clones. Interestingly, ectopic apoptotic figures were also observed in the neighboring wild-type cells, suggesting a non autonomous effect of selD^ptuf mutation (Fig. 1D).

We next examined whether the impairment of cell differentiation and survival in clones of selD^ptuf homozygous mutant cells was associated with a detectable burst of ROS. We used the ROS-specific nuclear probe dihydroethidium (DHE). This probe specifically detects the superoxide anion (⁠

⁠), a primary oxygen free radical produced by mitochondria rapidly removed by conversion to hydrogen peroxide. Upon an in vivo burst of

\(\mathrm{O}_{2}^{-}\)

⁠, DHE oxidizes to ethidium and incorporates into the nuclear DNA emitting fluorescence (Molecular Probes, Inc.). Certainly, some cells in selD^ptuf clones accumulated ROS, as shown by nuclear DHE staining. In addition to that, DHE-positive cells were detected in wild-type cells adjacent to the clone (Fig. 1E). This could explain the presence of apoptotic cells bordering the clone. ROS are very small molecules that can freely diffuse through cell membranes and after an accumulation in the clone the surrounding heterozygous tissue acts as a sink diluting this accumulation. To confirm that ROS accumulation was the cause and not a consequence of apoptosis, we performed the DHE assay on discs in which apoptosis was ectopically induced by constitutive expression of an inductor of apoptosis (GMRhid-expressing discs, see below). In these discs no DHE labeling was observed (data not shown) reinforcing the role of ROS in inducing apoptosis.

Ras/MAPK downregulation as a result of ROS increase has been observed in selD^ptuf heterozygotes (Morey et al., 2001), and so it is possible that apoptosis through hid could account for cell death in selD^ptuf homozygous mutants. To address this question, we checked for the enhancement of ectopic hid phenotype under the control of the eye-specific glass multimer reporter [pGMR (Hay et al., 1994)] after removing one copy of selD. Flies carrying one copy of GMRhid have severely reduced eyes devoid of most normal ommatidial morphology because of massive apoptosis in the eye disc (Fig. 2B) (Grether et al., 1995). The removal of one dose of selD, or of catalase (as a selD/sps1-independent source of ROS) did not enhance the GMRhid phenotype (Fig. 2C). Even the combination of selD^ptuf and Catⁿ¹ mutations in transheterozygotes could not enhance the GMRhid phenotype (Fig. 2D), whereas that combination is able to suppress a strong constitutively activated rasV12 construct (Fig. 2I,J) (Fortini et al., 1992). As it is likely that the GMRhid phenotype is too strong to detect an enhancement, we generated GMRhid sev-rasV12 flies to get a weaker apoptotic phenotype (Fig. 2E) (Bergmann et al., 1998; Kurada and White, 1998). Again, neither reduction of one dose of selD, nor of catalase, enhanced the rough eye phenotype of the GMRhid sev-rasV12 background (Fig. 2F). In addition, we performed a genetic interaction with an activated form of yan (GMRyan^Act), a negative regulator of the Ras/MAPK pathway. Overexpression of yan in the developing eye induces apoptosis (Fig. 2G) (Rebay and Rubin, 1995), and increases hid mRNA levels (Kurada and White, 1998). GMRyan^Act eyes display a milder phenotype than the GMRhid ones, but no enhancement was observed by reduction of one dose of selD (Fig. 2H). In addition to that, ectopic expression of a gain-of-function allele of the MAPK rolled locus (rl^sem) under different drivers (69B-GAL4 and arm-GAL4) does not rescue imaginal disc morphology in homozygous mutants at any time during larval development (data not shown). Together these results indicate that the impairment of the Ras/MAPK pathway may not be the main cause of selD^ptuf apoptosis and suggested that other apoptotic inducers could be involved.

Several signaling pathways converge to activate a cell-death program triggered by the pro-apoptotic gene reaper (White et al., 1994; Asano et al., 1996; Nordstrom et al., 1996; Brodsky et al., 2000). It is known that in mammals ROS can induce apoptosis (Buttke and Sandstrom, 1994; Simon et al., 2000) thus we tested if the increase in ROS caused by selD^ptuf mutation would trigger apoptosis through rpr. Indeed, we found rpr transcription in selD^ptuf homozygous discs using a rpr-lacZ reporter. Widespread β-galactosidase staining was observed throughout the mutant disc (Fig. 3A). Moreover, rpr expression was observed in clones of selD^ptuf homozygous mutant cells in both the wing and eye imaginal discs (Fig. 3B,C). In cell cultures ROS can activate the tumor suppressor protein p53, a sensor of genotoxic stress (Yin et al., 1998; Kitamura et al., 1999; Buschmann et al., 2000). As a transcription factor, one of the critical roles of p53 is to regulate the expression of genes involved in eliminating damaged cells via apoptosis. Activation of p53 occurs largely through posttranslational mechanisms that enhance its stability and DNA binding activity, making possible the transcription of its targets (Burns and El-Deiry, 1999; Sionov and Haupt, 1999). Drosophila melanogaster p53 protein (Dmp53) (Brodsky et al., 2000; Jin et al., 2000; Ollman et al., 2000) targets a radiation responsive enhancer at the rpr locus (Brodsky et al., 2000). It has been shown that targeted transcription of rpr rapidly causes widespread ectopic apoptosis (White et al., 1996). In fact, overexpression of Dmp53 in the eye gives rise to viable adults that have small rough eyes as a result of massive apoptosis (Brodsky et al., 2000; Jin et al., 2000; Ollman et al., 2000). Nuclear stabilization of Dmp53, in cells of the eye disc overexpressing Dmp53 under the glass (gl)-responsive enhancer elements (Moses and Rubin, 1991), can be detected with anti-Dmp53 antibody (Fig. 4A,B) (Ollman et al., 2000). Accordingly, we detected rpr transcription in gl-Dmp53-expressing cells, as shown by double staining with anti-Dmp53 and rprlacZ (Fig. 3D). We wondered whether rpr transcription observed in selD^ptuf mutant cells was associated with Dmp53. In selD^ptuf homozygous discs, widespread Dmp53 stabilization was detected with anti-Dmp53. This accumulation is characterized by a punctate pattern throughout the whole imaginal disc (Fig. 4C) and localized in the nuclei (Fig. 4D), which is in agreement with the function of p53 as a transcription factor.

In most cases, apoptotic cell death culminates in the activation of the caspase family of cysteine proteases, leading to the orderly dismantling and elimination of the cell. Caspases are central components of the apoptotic machinery induced by the pro-apoptotic genes. The Drosophila initiator caspase DRONC (Dorstyn et al., 1999) has been shown to be necessary in rpr-induced apoptosis (Hawkins et al., 2000; Meier et al., 2000; Quinn et al., 2000). Likewise, DRONC induces cell death in a dose-dependent manner when ectopically expressed in the developing eye under the GMR promoter (Meier et al., 2000; Hawkins et al., 2000; Quinn et al., 2000). Adult flies have slightly rough and mottled eyes, because of ablation of photoreceptors and pigment cells (Fig. 5A) (Meier et al., 2000). To test the involvement of this caspase in ROS-induced apoptosis, we performed a series of crosses using the selD^ptuf mutant as well as a mutation in the catalase gene (Catⁿ¹) as an example of ROS production in a selD/sps1-independent mode. Reduction of one dose of selD enhanced the mottled-eye phenotype as deduced from the presence of wider areas of white tissue (Fig. 5B). The same result was obtained when one dose of catalase was removed, reinforcing the idea that an increase in ROS levels might be responsible for selD^ptuf apoptosis (Fig. 5C). Moreover, when removing one dose of both catalase and selD this phenotype became even more severe (Fig. 5D).

DRICE is an effector caspase essential for apoptosis in Drosophila cells activated by rpr overexpression (Fraser and Evan, 1997; Fraser et al., 1997). It has been shown to physically interact with, and to be processed by, the initiator caspase DRONC, thus being one of its downstream targets (Meier et al., 2000). We assessed the involvement of DRICE caspase in selD^ptuf apoptosis using the processed DRICE-specific antibody (Dorstyn et al., 2002) as a marker for active DRICE. In selD^ptuf homozygous mutant imaginal discs we found active DRICE in several cells (Fig. 5E). Accordingly, in selD^ptuf clones in the wing and eye imaginal discs we found DRICE in some of the mutant cells (Fig. 5F). Interestingly, we also found non autonomous labeling of DRICE, as shown by non-mutant cells adjacent to the mutant clone. This observation is in agreement with the previous observation that ROS and apoptotic cells can also be non autonomous. Importantly, we have also detected active DRICE in gl-Dmp53 discs (Fig. 5G), which reinforces the idea that selD^ptuf apoptosis is mediated through a Dmp53/Rpr caspase-dependent pathway.

Since caspases promote and amplify proteolysis cascades, the inhibitor of apoptosis proteins (IAPs) provide a critical barrier to impede apoptosis through direct binding and inhibition of caspases (reviewed by Hay, 2000). Thus, apoptotic stimuli could induce cell death through degradation of IAPs. It has been shown that rpr can negatively regulate the levels of the Drosophila IAP1 (DIAP1) to trigger apoptosis (Goyal et al., 2000; Holley et al., 2002; Yoo et al., 2002). The inhibition of DRONC and DRICE caspases by DIAP1 has also been extensively demonstrated (Kaiser et al., 1998; Meier et al., 2000; Hawkins et al., 2000; Quinn et al., 2000). To confirm that selD^ptuf apoptosis is caspase dependent, we misexpressed the Drosophila IAP1 (DIAP1) to block caspase activity and test for viability rescue. The analysis was performed in the wing imaginal disc since divisions occur throughout the whole disc, whereas in the eye disc the pattern of cell divisions changes along the anterior-posterior axis which renders a statistical analysis difficult. By using hh-Gal4 as a driver, DIAP1 can be ectopically expressed in the posterior compartment, and the anterior compartment can act as a control in the same imaginal disc. We generated twin clones (i.e. wild-type control selD⁺ clone and selD^ptuf mutant clone produced in the same recombination event) in both the anterior and posterior compartment of these wing discs. We next measured the size of mutant selD^ptuf clones and defined them as the ratio of their own area relative to the area of their twin control selD⁺ clones and also compared the size of mutant clones in the anterior compartment to the size of mutant clones in the posterior compartment. We found that the size of selD^ptuf mutant clones in the posterior compartment is almost double that in the anterior compartment (Fig. 6). While the mean area of anterior mutant clones is 34% of the area of their wild-type twin clones, the mean area of posterior mutant clones is 58% of the area of their wild-type twin clones. Ectopic expression of DIAP1 robustly rescues viability and doubles the size of selD^ptuf mutant clones, which strongly supports the idea that selD^ptuf ROS-induced apoptosis is caspase dependent.

The main conclusion of the work presented here is that selD^ptuf cells generate a ROS oxidative stress that triggers a Dmp53/Rpr-mediated apoptosis. The increased levels of ROS, Dmp53 stabilization, rpr transcription and activation of caspases that can be inhibited by DIAP1, strongly hint at an important contribution of the Dmp53/Rpr caspase pathway in selD^ptuf apoptosis. However, when a cell is committed to die, several pro-apoptotic molecules may be activated to irreversibly execute an apoptotic program. For example, the Ras/MAPK down-regulation induced by selD (Morey et al., 2001) could contribute, to a lesser extend and beyond our detection levels, to apoptosis (Fig. 7).

Our results could be explained by an activation of the apoptotic machinery resulting from oxidative stress triggered by the absence of selenoproteins in the selD^ptuf cells. However, there are some observations that tempt us to propose that the fly and mammalian selD/sps1 could have derived a novel function. There are only three selenoprotein genes described so far in the fly genome: selenophosphate synthetase type 2 (sps2) (Hirosawa-Takamori et al., 2000), and dselM and dselG, both of unknown function (Castellano et al., 2001; Martin-Romero et al., 2001). The presence of non-selenoprotein paralogs of dselM and dselG, makes it hard to account for the lethality of selD^ptuf by simply removing the selenoproteins. Every known selenoprotein in vertebrates that acts as a detoxifyer enzyme (carrying a UGA-coded selenocysteine in the active center) is not a selenoprotein in the fly (i.e. the selenocysteine is substituted by another amino acid). Taken together, it is unclear whether Drosophila selenoproteins are part of the oxidative stress defense system. In addition, the enzymatic activity of the two highly conserved eukarya selD gene products differs. The selenocysteine amino acid residue of Sps2 is essential for its selenophosphate synthetase activity in mammals (Kim et al., 1997), whereas the mammalian Sps1 only weakly complements an E. coli selD mutation (Low et al., 1995). Similarly, the fly Sps2 contains a selenocysteine in its catalitic domain (Hirosawa-Takamori et al., 2000), whereas the Drosophila selD/sps1 does not complement a bacterial selD mutation (Persson et al., 1997). It is, therefore, possible that selD/sps1 has a dual function: one mediated by selenoproteins, or involved in their biosynthesis as basal selenophosphate synthetase activity (Guimaraes et al., 1996), and a second ROS-related function, independent of selenoproteins, conserved in flies and mammals.

We would like to thank E. Hafen, K. White, J. Abrams, T. Laverty, H. Richardson, J. Mahaffey, G. Struhl and E. Martín-Blanco for fly stocks. We also thank K. White and I. Patten for critically reading the manuscript. We thank the Developmental Studies Hybridoma Bank, University of Iowa, for providing anti-Elav. gl-Dmp53 flies and anti-Dmp53 antibody were provided by Exelixis Inc. through their p53 material transfer agreement. This work was supported by grant 01/0906 from FIS, Ministerio de Sanidad y Consumo and BMC2000-076 from Ministerio de Ciencia y Tecnología, Spain. M.M. is a fellow of CIRIT, Generalitat de Catalunya.