Compartment-specific Protection of Iron-Sulfur Proteins by Superoxide Dismutase*

Iron and oxygen are essential but potentially toxic constituents of most organisms, and their transport is meticulously regulated both at the cellular and systemic levels. Compartmentalization may be a homeostatic mechanism for isolating these biological reactants in cells. To investigate this hypothesis, we have undertaken a genetic analysis of the interaction between iron and oxygen metabolism in Drosophila. We show that Drosophila iron regulatory protein-1 (IRP1) registers cytosolic iron and oxidative stress through its labile iron sulfur cluster by switching between cytosolic aconitase and RNA-binding functions. IRP1 is strongly activated by silencing and genetic mutation of the cytosolic superoxide dismutase (Sod1), but is unaffected by silencing of mitochondrial Sod2. Conversely, mitochondrial aconitase activity is relatively insensitive to loss of Sod1 function, but drops dramatically if Sod2 activity is impaired. This strongly suggests that the mitochondrial boundary limits the range of superoxide reactivity in vivo. We also find that exposure of adults to paraquat converts cytosolic aconitase to IRP1 but has no affect on mitochondrial aconitase, indicating that paraquat generates superoxide in the cytosol but not in mitochondria. Accordingly, we find that transgene-mediated overexpression of Sod2 neither enhances paraquat resistance in Sod1+ flies nor compensates for lack of SOD1 activity in Sod1-null mutants. We conclude that in vivo, superoxide is confined to the subcellular compartment in which it is formed, and that the mitochondrial and cytosolic SODs provide independent protection to compartment-specific protein iron-sulfur clusters against attack by superoxide generated under oxidative stress within those compartments.

Iron and oxygen are indispensable but potentially harmful elements of aerobic life. Individually, their reactivity has been harnessed through association with a variety of proteins and the regulation of iron and oxygen metabolism constitutes one of the major triumphs of molecular evolution (1). Iron sulfur cluster proteins function in electron transport during oxidative phosphorylation and metabolism, but can also serve as iron and oxygen sensors (2). For instance, iron regulatory protein-1 (IRP1) 1 exerts its dual activities through the reciprocal use or dissasembly of its cubane iron sulfur [4Fe-4S] cluster; the holoprotein functions as a cytosolic aconitase, whereas the apoprotein is an RNA-binding translational regulator (1,3). The stability and functionality of IRP1 as a translation regulator is affected not only by iron levels, but also by oxidative stress, which induces IRP1 to bind iron responsive elements (IREs) located on the 5Ј and 3Ј untranslated regions of target genes (4,5). Although it is established that [4Fe-4S] cluster proteins can be specifically inactivated by superoxide (O 2 . ) (6 -8), the ques- have not yet been elucidated (9 -11). Studies in Saccharomyces cerevisiae have suggested an important role for cytosolic and mitochondrial superoxide dismutases (SODs) in iron metabolism (12,13). In addition to its function in the cytosol, SOD1 localizes in the mitochondrial intermembrane space and appears to also contribute to mitochondrial superoxide scavenging (14). Conversely, overexpression of the mitochondrial SOD2 was shown to compensate for lack of the cytosolic enzyme in a set of experiments assessing resistance to freeze-thaw stress (15). Although these results may point to some extent of functional redundancy between the two enzymes, other aspects of the Sod1⌬ phenotype, such as vacuolar fragmentation, cannot be rescued by Sod2 overexpression (16). Moreover, only recombinant bacterial FeSOD that is targeted to yeast mitochondria can rescue Sod2⌬, but it cannot rescue Sod2⌬ if the mitochondrial targeting sequence is omitted and FeSOD is expressed in cytosol (17,18). These results suggest a functional compartmentalization of superoxide metabolism and have broad implications for both physiologic redox signaling and cellular oxidative stress (19). However, the question of whether mitochondrially derived superoxide normally transfers into the cytosol (20,21), or if this is an abnormality associated only with apoptosis (22-24) remains controversial.
To address these questions, we used the cytosolic (IRP1) and mitochondrial aconitases as compartment-specific markers of O 2 . reactivity in conjunction with genetic modulation of superoxide dismutase levels in the cytosolic and mitochondrial compartments. We present evidence that, in Drosophila, these compartments define and limit the range of O 2 . reactivity.

EXPERIMENTAL PROCEDURES
Drosophila Stocks-Drosophila was cultured at 25°C on standard cornmeal agar medium. Sod1 n108 and Sod1 x39 represent null-activity alleles of the Sod1 gene (25,26). For RNA interference studies, UAS-* This work was supported by grants from the Natural Sciences and Engineering Research Council of Canada (NSERC) (to J. P. P. and A. J. H.) and by the Intramural program of the National Institute of Child Health and Human Development (NICHD). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Drosophila Cell Culture-Schneider II cells were maintained at 25°C in Schneider's Drosophila medium (Invitrogen). Sixteen hours prior harvesting, either 20 and 50 M deferoxamine or 20 and 100 g/ml ferric ammonium citrate (Sigma) were added to the medium.
RNA Mobility Shift Assay-Whole flies or cell pellets were homogenized directly in band shift buffer (40 mM KCl, 25 mM Tris HCl, pH 7.5) containing 1% Triton X-100, 5 mM dithiothreitol, and protease inhibitors. Extracts were centrifuged twice at Ͼ16000 ϫ g on a bench-top centrifuge (4°C), and the supernatant was immediately used for further analysis. Protein concentrations were determined using the Bradford reagent (Bio-Rad, Hercules, CA). Ten g of total protein were added to a final volume of 12.5 l of band shift buffer with or without 2% 2-mercaptoethanol, which activates IRP1 in vitro. The samples were incubated for 5 min at room temperature with 12.5 l of a reaction mixture containing 20% glycerol, 0.2 units/l Super RNAsine, 2 g/l yeast tRNA, 200 M dithiothreitol, and 32 P-labeled IRE from the human ferritin H chain gene (2000 counts/l). Twenty l of the reaction mixture were loaded onto a 10% acrylamide/Tris borate EDTA (89 mM Tris, 89 mM boric acid, 10 mM EDTA, pH 8.0 gel, run at 200 V for 2 h, then the gel was dried and exposed for autoradiography. A single band was observed, which could be competed out by addition of 10-fold excess of cold IRE probe (data not shown).
SOD2 Activity Assay-Adult males, 24 -72 h old, were homogenized in buffer-1 (50 mM sodium phosphate, pH 7.4, 0.1 mM EDTA). The extract was sonicated for 10 s to rupture mitochondria and centrifuged at 13,000 ϫ g. The supernatant was incubated in 60 mM diethyldithiocarbamic acid for 1 h at room temperature to inactivate SOD1. SOD2 activity was then determined spectrophotometrically by monitoring the autooxidation of 6-hydroxydopamine (6-HD) at 490 nm and 37°C in 500 l buffer-1 containing 0.1 mM 6-HD (32). Protein concentration was determined using the Bradford reagent (Bio-Rad).
Paraquat Exposure-w 1 or w 1 ;UAS-Sod2/ϩ;da G32 Gal4/ϩ adult  5). Radioactive human ferritin H chain IRE was used as a probe; each lane was loaded with 8 g of total protein. Addition of 2-mercaptoethanol (2-ME) after lysis reveals the total amount of activable IRP1 (lanes 6 -10, respectively). Note that a robust activation of IRE-binding activity of IRP1 follows depletion of iron by deferoxamine treatment, whereas addition of iron to the medium diminishes this activity. Thus, Drosophila IRP1 registers change in iron levels in an analogous manner to IRP1 of mammals.
Lifespan Measurements-Adult males were collected 1 day after eclosion, kept in groups of 10 per vial, and transferred to fresh food every second or third day.
der-II cells cultured with varying concentrations of iron. Fig. 1 shows that chelation of iron by deferoxamine greatly enhances binding of IRP1 to radiolabelled IRE. In contrast, addition of ferric ammonium citrate, which reconstitutes the [4Fe-4S] cluster, converts IRP1 to its aconitase form. This assay does not discriminate between the two Drosophila IRP1 proteins. Taken together, these results suggest that the [4Fe-4S] cluster switch that mediates the functional interconversion between a translational regulator and a cytosolic aconitase in response to iron levels is evolutionarily conserved between Drosophila and mammals.
IRP1 . that originates in mitochondria will react with IRP1 localized in the cytoplasm. As for other eukaryotes, Drosophila possess a form of SOD (SOD2) that is confined to mitochondria (30). However, because no Sod2 mutant has been reported that could be used to create a mitochondrial environment of high oxidative stress, we utilized transgenic strains developed earlier in this laboratory that silence Sod2 by means of RNAi (27). For controls, we generated strains with transgenically silenced Sod1. 3 If O 2 .
does not normally cross membrane boundaries, as suggested by its anionic properties and as supported by experiments in mammalian cells (8) (see "Discussion"), IRP1 is predicted to be unaffected when Sod2 expression is silenced. Fig. 2B shows that IRP1 binding to IREs is indeed unaffected when Sod2 is silenced. In contrast, silencing of Sod1 by the same mechanism leads to robust activation of IRP1 (Fig. 2C), in keeping with the activation of IRP1 seen in the Sod1 Ϫ/Ϫ mutant ( Fig. 2A).
Mitochondrial and Cytosolic Aconitases Are Selectively Inactivated by Loss of SOD2 and SOD1, Respectively-Having established that IRP1 activity is unaffected by mitochondrial O 2 . , we asked if silencing Sod2 would exhibit any corresponding effect on cytosolic aconitase activity. As shown in Fig. 3B, (see also Ref. 27) silencing of mitochondrial Sod2 has no discernible affect on cytosolic aconitase activity. In contrast, loss of SOD1 activity by mutation or RNAi-mediated silencing strongly inactivates cytosolic, but not mitochondrial, aconitase (Fig. 3, A  and C). Collectively, the results presented in Figs. 2 and 3 argue strongly for a compartmentalized redox environment in which the reactivity of O 2 . generated within either the mitochondrial or cytosolic compartment is limited to protein ironsulfur substrates residing within the same respective compartment.
Resistance to Paraquat Toxicity Is Unaffected by Augmentation of SOD2-In view of the compartmentalized effects conferred by selective diminution of SOD1 and SOD2 in the cytosolic and mitochondrial compartments, respectively, we asked if the two enzymes function differentially in the context of whole organism biology as was previously shown for the thioredoxin antioxidant defense system of Drosophila (37,38). To address this question, we produced UAS-Sod2 transformants that ectopically express Sod2 through use of the UAS/Gal4 system (29). Using the da G32 Gal4 driver to broadly overexpress Sod2 produces a 3-fold elevation of mitochondrial SOD2 activity (Fig. 4A). We then assessed the protective effect of Sod2 overexpression against the toxicity of paraquat, a widely used O 2 . -generating agent (25). In sharp contrast to the robust paraquat-resistance conferred by augmentation of SOD1 (31,39,40), augmentation of SOD2 activity provided no increased protection against the toxicity of paraquat (Fig. 4B). This result was also seen with augmentation of SOD2 using motorneuron or muscle Gal4 drivers (data not shown) and is consistent with previously published results on the failure of Sod2 overexpression to provide resistance to hyperoxia, heat stress, and starvation (41).
Paraquat Exposure Converts Cytosolic Aconitase to IRP1 but Has No Effect on Mitochondrial Aconitase-We then examined the compartment-specific impact of paraquat exposure on the cytosolic and mitochondrial aconitases. Wild type flies (Sod1 ϩ , Sod2 ϩ ) were exposed to the same concentrations of paraquat as in Fig. 4B, but for a shorter, non-lethal time (16 as opposed to 48 h), and IRP1 binding and aconitase activities were assayed. A decrease in cytosolic aconitase activity and a reciprocal increase in IRP1 binding activity confirmed that, as expected, paraquat exposure causes an increased flux of O 2 . in the cytosol but not in mitochondria (Fig. 4, C and D). The failure of Sod2 overexpression to afford additional protection against paraquat toxicity (Fig. 4B)

compartments in
Drosophila. To answer this question, we overexpressed Sod2 in the Sod1 x39 genetic background using ubiquitous, motorneuron or muscle-specific Gal4 drivers and determined the lifespan of the resulting adults (Fig. 5). The results clearly demonstrate that, in striking contrast to Sod1, ectopic expression of Sod2 confers no discernible effect on the severely truncated lifespan of Sod1 x39 adults (Fig. 5). Neither does it rectify the unusually high pupal mortality of Sod1 x39 (data not shown). We thus conclude that the role of SOD2 in mitigating cytosolic oxidative stress in general and scavenging cytosolic O 2 . in particular is negligible.

DISCUSSION
These studies represent the first in vivo analysis in a higher eukaryote of the consequences of genetic modulation of cytosolic and mitochondrial SODs on the function of iron-sulfur proteins that reside within those respective compartments. These iron-sulfur proteins are both important cellular targets and sensitive indicators of superoxide-mediated oxidative stress. The distinct subcellular partitioning of the cytosolic and mitochondrial aconitases and the availability of methods to assay these two activities concurrently in the same extract provided the opportunity to investigate whether superoxide generated in the mitochondrial compartment can affect an iron-sulfur protein in the cytosol and vice versa. Using a combination of genetic and pharmacological methods to generate compartment-specific oxidative stress, we examined the responses of the cytosolic and mitochondrial aconitases. The results are consistent with the interpretation that in Drosophila, cytosolic aconitase is unaffected by superoxide generated in mitochondria and conversely, that mitochondrial aconitase is unaffected by superoxide generated in the cytosol. We interpret these results to mean that containment of O 2 . within the compartment of its origin along with the SOD specific to that compartment is a central feature of reactive oxygen homeostasis in Drosophila. How this homeostatic mechanism might relate to the severe debilitating phenotypes that characterize Sod1 Ϫ/Ϫ mutants of Drosophila (25,26) as compared with the relatively benign phenotype of the corresponding Sod1 Ϫ/Ϫ mutant in mice (42) is unclear at present. In principle, the liberation of intracellular iron resulting from the reaction of unscavenged O 2 . with [4Fe-4S] clusters (12,13,43), could also contribute to some of the complex phenotypes of Sod1 Ϫ/Ϫ mutants (25). In this regard, it is interesting to note that mutations in iron-sulfur cluster assembly genes have been shown to suppress the Sod1 deficiency phenotype of S. cerevisiae (44), suggesting that reduced iron sulfur protein expression can abrogate O 2 . toxicity.
Whether mutations in homologous genes would have similar effects on Sod1 Ϫ/Ϫ phenotypes in Drosophila remains an open question. The results described here contrast with a recent in vitro study of isolated mammalian mitochondria. Han et al. (21) present evidence that superoxide can be released from rat heart mitochondria cultured in vitro through voltage-dependent anion channels. Biological differences between Drosophila and rat heart mitochondria, the use of different methods used to detect and modulate superoxide flux, and in vivo versus in vitro conditions are all potential contributors to these differences. Further experiments will be required to clarify this matter.
These studies set the stage for investigating of the role of Sod1, Sod2, and oxidative stress in iron metabolism in Drosophila. The results presented here underscore the importance of compartment boundaries in maintaining reactive oxygen homeostasis in Drosophila and could have therapeutic implications in relation to the oxidative stress component of inflammatory disease.