Cytosolic Aconitase and Ferritin Are Regulated by Iron inCaenorhabditis elegans *

Iron regulatory protein-1 (IRP-1) is a cytosolic RNA-binding protein that is a regulator of iron homeostasis in mammalian cells. IRP-1 binds to RNA structures, known as iron-responsive elements, located in the untranslated regions of specific mRNAs, and it regulates the translation or stability of these mRNAs. Iron regulates IRP-1 activity by converting it from an RNA-binding apoprotein into a [4Fe-4S] cluster protein exhibiting aconitase activity. IRP-1 is widely found in prokaryotes and eukaryotes. Here, we report the biochemical characterization and regulation of an IRP-1 homolog in Caenorhabditis elegans(GEI-22/ACO-1). GEI-22/ACO-1 is expressed in the cytosol of cells of the hypodermis and the intestine. Like mammalian IRP-1/aconitases, GEI-22/ACO-1 exhibits aconitase activity and is post-translationally regulated by iron. Although GEI-22/ACO-1 shares striking resemblance to mammalian IRP-1, it fails to bind RNA. This is consistent with the lack of iron-responsive elements in the C. elegans ferritin genes, ftn-1 and ftn-2. While mammalian ferritin H and L mRNAs are translationally regulated by iron, the amounts of C. elegans ftn-1 and ftn-2 mRNAs are increased by iron and decreased by iron chelation. Excess iron did not significantly alter worm development but did shorten their life span. These studies indicated that iron homeostasis in C. elegans shares some similarities with those of vertebrates.

Iron is an essential element required for growth and survival of most organisms. The importance of iron is implicit in the role it plays in oxygen transport and heme synthesis as well as its ability to serve as a cofactor for enzymes involved in a variety of biological processes including DNA synthesis, energy production, and neurotransmitter synthesis. Abnormally high concentration of cellular iron is toxic due to its ability to catalyze the generation of free radicals that damage DNA, lipids, and proteins. In humans, the accumulation of excess cellular iron can result in cirrhosis, arthritis, cardiomyopathy, diabetes mellitus, and increased risk of cancer and heart disease. To provide adequate iron for cellular needs yet prevent the accumulation of excess iron, the concentration of iron within cells is tightly controlled.
In vertebrates, the iron regulatory proteins 1 and 2 (IRP-1 and IRP-2) 1 regulate iron homeostasis. IRPs are cytosolic RNAbinding proteins that regulate the translation or the stability of mRNAs encoding proteins involved in iron and energy homeostasis (1)(2)(3)(4). IRPs bind to RNA stem-loop structures, known as iron-responsive elements (IREs), that are located in either the 5Ј-or 3Ј-untranslated regions (UTRs) of specific mRNAs. These mRNAs encode proteins involved in iron storage (ferritin), iron utilization (erythroid aminolevilunate synthase and mitochondrial aconitase), and iron transport (transferrin receptor and divalent metal transporter-1). When iron is scarce, IRP binding to the 5Ј IRE in ferritin mRNA represses translation, whereas IRP binding to the 3Ј IREs in the transferrin receptor mRNA stabilizes this mRNA. When iron is abundant, IRPs lose affinity for the IREs, leading to enhanced ferritin synthesis and to the rapid degradation of transferrin receptor mRNA. By regulating the amount of iron taken up by transferrin receptor and the amount of iron sequestered by ferritin, cellular iron concentration is maintained, and iron toxicity is avoided.
Iron regulates the RNA binding activity of IRP-1 and IRP-2, but the mechanism of regulation differs. In the presence of iron, a [4Fe-4S] cluster assembles in IRP-1, converting it from an RNA-binding protein into a cytosolic aconitase. Aconitases are [4Fe-4S] cluster proteins that are found in the cytosol, mitochondria, and glyoxysomes and catalyze the reversible isomerization of citrate and isocitrate via cis-aconitate (5). Despite information regarding the role of aconitase in mitochondria and in the glyoxylate cycle in microorganisms and plants, the function of cytosolic aconitase in higher eukaryotes is not clear. Unlike IRP-1, IRP-2 lacks a [4Fe-4S] cluster and consequently lacks aconitase activity (6). Rather, iron regulation of IRP-2 involves iron-induced IRP2 degradation by the proteasome (7)(8)(9).
Here, we report on the biochemical properties and regulation of C. elegans cytosolic aconitase (GEI-22/ACO-1). C. elegans is a multicellular organism that shares many basic cellular mechanisms with vertebrates, and consequently, it has been used to study development, neurobiology, stress, and aging. Many of the same genes that are involved in iron and energy homeostasis in vertebrates are conserved in C. elegans, including aconitases, ferritin, divalent metal transporter-1, frataxin, and iron sulfur cluster assembly proteins, suggesting that iron and energy homeostasis are also conserved. These features prompted us to characterize the biochemical properties and the regulation of GEI-22/ACO-1 and mechanisms regulating iron homeostasis in C. elegans.

EXPERIMENTAL PROCEDURES
C. elegans Culture-Wild-type C. elegans (variety Bristol, strain N2) were cultivated on nematode growth medium (NGM) agar plates or in large scale liquid cultures seeded with E. coli strain OP50 at 22°C (23). For large scale cultures, mixed stage worms were placed in complete S-basal medium (500 ml) containing OP50 bacteria (15 g) and grown until the bacteria were gone. S-basal medium was supplemented with 0.003-6.6 mg/ml ferric ammonium citrate (FAC) or 100 M deferoxamine (DF), an iron chelator. Because FAC lowered the pH of S-basal medium, it was adjusted to pH 7.0. Worms were grown for 4 days with gentle shaking at 180 rpm and were collected and washed free of bacteria by sucrose flotation. The toxic concentration of FAC and DF was determined by iron toxicity assays (see below) and by growth of worms in different DF concentrations followed by assaying worm survival.
RNA for Northern blots was made from L4 worms prepared by alkaline hypochlorite of gravid adults (24). Embryos were grown on NGM agar plates (12 ϫ 60-mm plates) supplemented with FAC (0.003-6.6 mg/ml) and adjusted to pH 7.0 or DF (100 M) for 4 days at 22°C. DF-supplemented plates were prepared by seeding NGM plates with 0.1 ml of OP50 bacteria (380 mg/ml) to which 100 M DF was added.
For iron toxicity life span analysis, 10 L4 larval stage worms were placed on NGM agar plates supplemented with FAC (0.03-6.6 mg/ml) at 22°C. Each day, the worms were moved to a new plate and were scored as dead if they failed to move when provoked or lacked pharyngeal pumping. Three distinct experiments were carried out, with 10 worms in each experiment. Statistical comparisons were made using a Cox regression model testing for trend in survival explained by the dose of iron.
Construction of a cDNA Encoding GEI-22/ACO-1 and of a gei-22/aco-1::GFP Reporter Gene-C. elegans gei-22/aco-1 cDNA was synthesized from RNA isolated from mixed stage worms using TRIzol reagent (Invitrogen) and Superscript II reverse transcriptase (Invitrogen). PCR was performed using Pfu polymerase and an upstream primer containing a KpnI site followed by the first 18 nt of the gei-22/ aco-1 coding sequence (5Ј-GCGCGGTACCGCCATGCGTTTCAACAAC-CTT-3Ј) and a downstream primer containing the last 18 nt of gei-22/ aco-1 coding region in frame with a FLAG epitope, a stop codon, and an XbaI site (5Ј-GCGCTCTAGATTACTTGTCATCGTCGTCCTTGTAGTC-TTGGATCAACTTTCTGATCAT-3Ј). These primers were chosen based on the alignment of the GEI-22/ACO-1 amino acid sequence predicted from the ZK455.1 with those of other IRP-1s using the ClustalW multiple sequence alignment program (Fig. 1). The 2.9-kb fragment was cloned into the KpnI-XbaI sites of pcDNA3 (Invitrogen), yielding the mammalian expression construct pACO-1FLAG.
A translational green fluorescent protein (GFP) construct GEI-22/ACO-1::GFP was constructed by digesting ZK455.1 with Eco47III-NruI and inserting the 3,895-bp fragment into the SmaI site of the promoterless GFP reporter vector pPD95.77 (Dr. Andrew Fire). This construct contained 939 nt of 5Ј regulatory sequences and 2.9 kb of gei-22/aco-1 coding and intronic sequences (7 introns) fused in frame to GFP. The construct lacks sequences encoding the last 110 amino acids of GEI-1/ACO-1 and the 3Ј-UTR, resulting in a nonfunctional protein. pGFP-ACO-1 (30 ng/l) was co-injected with pRF4 DNA (30 ng/l) containing a dominant mutant gene for rol-6 allele, and transformed worms were selected by their rolling behavior. Four transgenic worms were generated, and all showed the same GFP expression pattern.
Expression of GEI-22/ACO-1 in Yeast and Generation of Antibodies-A His 8 tag was cloned onto the N terminus of pACO-1FLAG, and the insert was subcloned into the KpnI-XbaI sites of pYes2 (Invitrogen), yielding the yeast expression construct pYhis-ACO-1FLAG. pYhis-ACO-1FLAG was transformed into yeast strain JEL1 (25). Yeast were grown overnight in synthetic complete medium minus uracil (SC-Ura Ϫ ) containing 2% dextrose. Yeast were then diluted 1:100 into SC-Ura Ϫ containing 3% glycerol, 2.0% lactic acid and grown to an A 600 of 0.8. 2% galactose was added to the culture for 10 -18 h to induce GEI-22/ACO-1 expression. Yeasts were harvested, and the pellet was resuspended in 3 ml/g, wet weight, lysis extraction buffer (0.3 M NaCl, 0.05 M NaH 2 PO 4 , pH 8.0, 10 mM ␤-mercaptoethanol, 0.025% Nonidet P-40, 0.1 mM EDTA) containing a protease inhibitor tablet (Roche Molecular Biochemicals). Cells were disrupted by vortexing with glass beads, and the lysate was centrifuged at 27,000 ϫ g for 30 min. The lysate was mixed with 2.0 ml of equilibrated Ni 2ϩ -nitrilotriacetic acid resin (Qiagen) for 1 h at 4°C. The slurry was poured into a column and was washed with column buffer (0.3 M NaCl, 0.05 M NaH 2 PO 4 , pH 8.0) containing 0.07 M imidazole. GEI-22/ACO-1 was eluted with column buffer containing 0.2 M imidazole, and fractions containing GEI-22/ACO-1 were dialyzed against assay buffer (0.05 M Tris, pH 7.3, 0.1 M NaCl, 1 mM dithiothreitol). For antibody preparation, GEI-22/ACO-1 was further purified by preparative 8% SDS-PAGE. The protein was visualized using a cold solution of 2.5 M KCl, and the band containing GEI-22/ACO-1 was excised and used to inject rabbits. The polyclonal antibodies detected a band on SDS-PAGE with a molecular mass of ϳ100 kDa, which corresponds to the size of the predicted gene product of gei-22/aco-1 mRNA. Preimmune serum did not detect this band.
Expression of GEI-22/ACO-1 in Human Embryonic Kidney 293 Cells-A 10-cm plate of HEK 293 cells was cotransfected with 8 g of pACO-1FLAG and 2 g of pEGFP (Clontech) DNAs. The cells were equally split after 5 h to six 35-mm plates, and duplicate plates received either FAC (50 g/ml) or DF (50 M) or no addition. The cells were incubated for 16 h and then harvested in 0.125 ml of lysis buffer (0.02 M HEPES, pH 7.6, 0.025 M KCl, 1 mM dithiothreitol, 0.25% Nonidet P-40) containing a protease inhibitor tablet (Roche Molecular Biochemicals). The lysate was centrifuged at 15,000 ϫ g for 20 min at 4°C, and the supernatant was assayed for protein using the Coomassie Blue Plus Protein Reagent (Pierce).
Isolation of C. elegans Protein Extracts-Worms were harvested from 500-ml cultures containing 0.33 mg/ml FAC or 100 M DF. The worms were washed free of bacteria by sucrose flotation and suspended in a 2ϫ volume of homogenization buffer (0.05 M Tris-HCl, pH 7.9, 25% glycerol, 0.1 mM EDTA, 0.32 M NH 4 SO 4 ) containing a protease inhibitor tablet (Roche Molecular Biochemicals). The lysates were homogenized on ice using a Brinkmann Instruments Polytron homogenizer at full power for 15 s, repeated seven times. The homogenates were centrifuged at 160,000 ϫ g for 1 h, and the supernatants were assayed for protein using the Coomassie Blue Plus protein reagent (Pierce).
RNA-binding Gel Shift and Aconitase Assays-RNA-binding assays were performed as described (26), using protein (12 g) from yeast, HEK 293 cells, or C. elegans lysates and a 32 P-labeled rat ferritin L-IRE (R. norvegicus fer-l) (25) or a C. elegans ferritin-2 (ftn-2) RNA. The ftn-2 RNA was synthesized from DNA that corresponded to 530 nt 5Ј to the start codon of ftn-2 gene (C. elegans cosmid D1037.3). The forward primer contained a T7 promoter sequence (19 nt) followed by 20 nt of ftn-2 sequence (5Ј-TAATACGACTCACTATAGGGGTTGAAGCATAAT-ACTATTACG-3Ј), and the reverse primer contained the first 20 nt 5Ј to the start site in ftn-2 (5Ј-GGTAGTTTGTGGCTGGTAAG-3Ј). After incubation of the RNA-protein complexes for 20 min, RNase T1 (1 unit) was added to the reaction for 5 min followed by the addition of 1.5 g of heparin (50 mg/ml) for 5 min. A 5% native polyacrylamide gel was used to resolve the RNA-protein complexes.
Aconitase assays were performed using lysates obtained from worms and from HEK 293 cells transfected with pcACO-1FLAG. Aconitase activity was assayed by the addition of cis-aconitate (0.2 mM final concentration) to 50 g of protein in 0.5 ml of aconitase buffer (0.05 M Tris-HCl, pH 7.5, 0.1 M NaCl) (27). The disappearance of cis-aconitate was measured at 240 nm over time. For aconitase assays in worms, four separate experiments were carried out. The differences between FACand DF-treated worms were determined by paired Student's t test, and p Ͻ 0.05 was considered significant.
The expression pattern of GEI-22/ACO-1 in living worms was determined using GFP-reporter fusions constructs. A gei-22/aco-1::GFP translational fusion gene was made that con- tained ϳ1 kb of 5Ј regulatory sequences and 2.9 kb of coding and intronic sequences of gei-22/aco-1 fused in-frame to gfp ( Fig. 2A). This construct lacks 3Ј-UTR sequences and those encoding the last 110 amino acids GEI-22/ACO-1. Transgenic embryos, L2 larval stage, and adult animals carrying the reporter construct showed high levels of cytosolic GFP expression in the hypodermal seam cells and in the intestine (Fig. 2, B-D). No significant GFP expression was observed in muscle cells or in neurons.
GEI-22/ACO-1 Does Not Bind RNA-Mammalian IRP-1 exhibits the mutually exclusive activities of RNA binding and aconitase. Iron causes IRP-1 to switch from an RNA-binding apoprotein form to a non-RNA binding aconitase form containing a [4Fe-4S] cluster. The switch between these forms occurs without changes in IRP-1 protein levels. Since GEI-22/ACO-1 shows significant amino acid identity with mammalian IRP-1, we questioned whether GEI-22/ACO-1 binds RNA. The 5Ј sequences flanking the C. elegans genes encoding ferritin-1 (C. elegans FTN-1) and ferritin-2 (C. elegans FTN-2), succinate dehydrogenase, and ACO-2 were inspected for IREs, since 5Ј IREs are found in homologous genes in other organisms (19, 34 -36). No consensus IREs was identified in these genes. It was possible, however, that nonconsensus IREs might be present in the 5Ј-UTRs of these genes. To test for this, IRE binding activity was measured in extracts of worms grown in FAC or the iron chelator DF using 32 P-labeled RNAs corresponding to sequences located ϳ500 nt upstream of the start site of the ftn-1, ftn-2, aco-2, and succinate dehydrogenase genes. Since C. elegans exons and introns are generally shorter than in vertebrates (37), we reasoned that if these genes harbor 5Ј-IREs, then they should be present within these sequences. Worms were grown in FAC-or DF-supplemental medium, because previous studies showed that FAC and DF can decrease and increase, respectively, IRP-1 RNA binding activity (1, 2, 4). The concentration of FAC and DF was chosen based on testing the growth of worms in different concentrations of FAC and DF (see "Experimental Procedures"). Fig. 3B shows that no specific RNA binding activity was detected in worm extracts by RNAband shift gels with the 32 P-labeled ftn-2 RNA or with other 32 P-labeled C. elegans RNAs (data not shown), although GEI-22/ACO-1 was detected in these extracts by immunoblotting using an anti-GEI-22/ACO-1 antibody (Fig. 3C). As controls, HEK 293 cells were treated with FAC or DF, and IRP RNA binding activity was measured using the 32 P-labeled R. norvegicus fer-l IRE (Fig. 3A). DF increased IRP RNA binding activity as expected, whereas FAC had little effect on IRP RNA binding activity due to the high iron concentration in these cells. No RNA-protein complexes were formed with a 32 Plabeled R. norvegicus fer-l IRE in worm extracts (Fig. 3A). C. elegans extracts spiked with R. norvegicus IRP-1 showed that 32 P-labeled R. norvegicus fer-l IRE bound to R. norvegicus IRP-1, indicating that an inhibitor of RNA binding activity was not present in these extracts (data not shown).
To confirm that GEI-22/ACO-1 lacks RNA binding activity, we expressed His/FLAG-tagged GEI-22/ACO-1 in yeast using a galactose-inducible promoter. The advantage of yeast is that they do not express endogenous IRPs, which might interfere with the detection of small amounts of RNA binding activity. As controls, yeast strains expressing His-tagged R. norvegicus IRP-1 and IRP-2 were also assayed for RNA binding activity. Galactose induced the expression of GEI-22/ACO-1, R. norvegicus IRP-1, and R. norvegicus IRP-2, but only R. norvegicus IRP-1 and R. norvegicus IRP-2 bound to the R. norvegicus fer-l IRE (Fig. 4, A and B). Taken together, these data indicate that GEI-22/ACO-1 does not bind to a mammalian consensus IRE or to C. elegans RNAs that might be expected to harbor functional IREs.
GEI-22/ACO-1 Is an Aconitase-The data indicate that GEI-22/ACO-1 lacks RNA binding activity. The 24 active sites in mitochondrial aconitases (29 -31) are present in GEI-22/ ACO-1, suggesting that GEI-22/ACO-1 is an aconitase. To determine whether GEI-22/ACO-1 exhibits aconitase activity and whether it is regulated by iron, total aconitase activity was measured in HEK 293 cells transfected with FLAG-tagged GEI-22/ACO-1 or pcDNA3 control. Some cells were treated with either FAC or DF for 16 h before assaying aconitase activity. Cells transfected with FLAG-tagged GEI-22/ACO-1 showed a ϳ2-fold increase in total aconitase activity compared with pcDNA3-transfected cells (Fig. 5A). When cells were treated with FAC, total aconitase activity increased ϳ4-fold in GEI-22/ACO-1-transfected cells compared with pcDNA3transfected cells. Endogenous aconitase activity did not signif- icantly increase in pcDNA3-transfected cells treated with FAC, indicating that these cells are iron-sufficient, and that the majority of IRP-1 is in the aconitase form. In contrast, DF dramatically reduced aconitase activity in GEI-22/ACO-1-and pcDNA3-transfected cells. The decreased aconitase activity observed in pcDNA3-transfected cells is due to decreased endogenous cytosolic and mitochondrial aconitase activities. GEI-22/ACO-1 was expressed at similar levels in all extracts, suggesting that the change in aconitase activity was due to the post-translational conversion of the apoprotein form to the [4Fe-4S] form (Fig. 5B). The fact that iron caused a ϳ3-fold increase in aconitase activity in GEI-22/ACO-1-transfected cells compared with GEI-22/ACO-1-untreated cells suggested that iron is limiting under conditions of overexpression. Taken together, these data showed that GEI-22/ACO-1 is an aconitase and that this activity can be regulated by iron.
Iron Regulates Aconitase Activity in C. elegans-To determine whether aconitases are regulated by iron in worms, aconitase activity was measured in lysates from mixed stage worms grown for 4 days in liquid cultures supplemented with either 0.33 mg/ml FAC or 100 M DF. We compared FAC and DF treatments because these conditions represent the extremes of iron concentrations. In these experiments, total aconitase activity (GEI-22/ACO-1 and ACO-2) was measured, because the cytosol was routinely contaminated with mitochondrial proteins. Total aconitase activity from four separate experiments showed base-line variability; however, in all experiments a significant reduction (12-16%) was observed in DFtreated worms compared with FAC-treated worms (p Ͻ 0.003, paired Student's t test) (Fig. 6A). A representative immunoblot showed that the amount of GEI-22/ACO-1 did not change in DF-or FAC-treated worms. Although we cannot attribute the change in aconitase activity specifically to GEI-22/ACO-1 because of ACO-2 contamination, these data indicated that aconitase activity can be regulated by iron in worms. The data also showed that the decrease in aconitase activity in worms grown in DF is not as great as in HEK 293 cells transfected with GEI-22/ACO-1 (compare Figs. 5A and 6A). This might be due to inefficient DF uptake into worms grown in liquid culture compared with DF uptake into cultured cells.
Iron Regulates C. elegans ftn-1 and ftn-2 mRNAs-The lack of discernible IREs in the C. elegans genome and the absence of GEI-22/ACO-1 RNA binding activity suggested that C. elegans ferritin genes might be iron-regulated by mechanisms other than translation. C. elegans expresses two ferritin-like proteins, which we designated C. elegans FTN-1 and FTN-2. C. elegans FTN-1 and FTN-2 are more homologous to R. norvegicus FER-H (54 and 61%) than to R. norvegicus FER-L (45 and 50%), and both proteins contain ferroxidase active-site residues that are characteristic of R. norvegicus FER-H subunits (38).
To determine whether ftn-1 and ftn-2 mRNAs are regulated by iron, worms were grown on agar plates supplemented with FAC (0.003-6.6 mg/ml) or with DF (100 M) for 4 days, and ftn-1 and ftn-2 mRNAs were measured by Northern blots. Fig.   FIG. 4. Recombinant GEI-22/ACO-1 expressed in yeast does not bind RNA. A, extracts (20 g) from yeast expressing His/FLAG-tagged GEI-22/ACO-1 and His-tagged R. norvegicus IRP-1 and R. norvegicus IRP-2 were assayed by immunoblots using anti-FLAG, anti-rat IRP-1, or anti-rat IRP-2 antibodies (Ab). Galactose (ϩ) was added to some cultures to induce IRP expression. Note that there is some IRP expression in yeast grown in the absence of galactose due to leaky expression. B, the same extracts (12 g) from A were incubated with a 32 P-labeled R. norvegicus fer-l IRE followed by separation of the RNA-protein complexes by 5% native PAGE.

FIG. 5. GEI-22/ACO-1 is an aconitase.
A, HEK 293 cells were transfected with FLAG-tagged pGEI-22/ACO-1 or pcDNA3 (empty vector). Cells were treated with either FAC (100 g/ml), DF (50 M), or no treatment (NT) for the last 12 h of the transfection. After 36 h, cell lysates were prepared, and total aconitase activity was assayed by using a spectroscopic assay that measures the disappearance of cisaconitate at 240 nm (27). Aconitase activity is plotted as arbitrary activity units relative to pcDNA3-NT control. Data are presented as means Ϯ S.E. (n ϭ 4); asterisks denote significant differences using a paired Student's t test (p Ͻ 0.01). B, samples from two experiments in A were analyzed by 8% SDS-PAGE, and immunoblots were carried out using anti-GEI-22/ACO-1 antibody. 6. Aconitase activity is regulated by iron in worms. A, mixed stage worms were grown in liquid cultures containing FAC (0.33 mg/ml) or DF (100 M) for 4 days. Worm lysates were prepared, and total aconitase was measured as in Fig. 5A. Aconitase activity is plotted as arbitrary activity units. Four independent experiments (represented by different symbols) are shown with each line representing worms grown in FAC or DF. The differences in aconitase activity between FAC and DF groups were compared using a paired Student's t test (p Ͻ 0.003). B, protein (20 g) in one of the experiments in A was analyzed by 8% SDS-PAGE, and immunoblots were carried out using anti-GEI-22/ACO-1 antibody (Ab). 7 shows that ftn-1 mRNA steadily increases ϳ2-fold as FAC concentration increases from 0.33 to 6.6 mg/ml, whereas a smaller, but consistent, ϳ0.5-fold increase is observed for ftn-2 mRNA. In worms grown in the presence of DF, ftn-1 and ftn-2 mRNAs decrease ϳ10and ϳ2-fold, respectively. To quantify changes in ftn-1 and ftn-2 mRNAs, band intensity was normalized to act-1 mRNA (Fig. 7B). These data indicated that ftn-1 and ftn-2 mRNAs are regulated by iron but that ftn-1 is more responsive to changes in iron concentration than ftn-2. gei-22/ aco-1 mRNA does not significantly change with FAC or DF treatment (Fig. 7, A and B), which is consistent with data indicating the amounts of GEI-22/ACO-1 protein are not affected by iron (Figs. 3C and 6B).
Effect of Iron on the Development and Life Span of C. elegans-Since little is known about iron homeostasis in worms, we determined the concentration of iron that causes toxicity in worms. First, we tested whether the development of worms was affected by iron. Embryos were grown on FACsupplemented (0.033-6.6 mg/ml) NGM plates. All worms grew into adults within 4 days, indicating that iron did not affect worm development (data not shown). We then tested whether the life span of worms was altered when grown on iron plates. L4 larval stage worms were placed on agar plates supplemented with FAC (0.033-6.6 mg/ml), and worms were counted each day until all were dead. Fig. 8 shows that the mean life spans Ϯ S.E. of worms grown on 3.3 and 6.6 mg/ml FAC plates were 12.9 Ϯ 2.0 and 11.1 Ϯ 1.2 days, respectively, compared with 15.8 Ϯ 0.70 days for worms grown on control plates. These data indicated that excess iron does not affect the development of worms but is toxic when worms are exposed to iron during their lifetime. DISCUSSION Here, we investigate the iron-dependent regulation of GEI-22/ACO-1 and the regulation of iron homeostasis in C. elegans. We show that GEI-22/ACO-1 is similar to other vertebrate IRP-1 proteins in that it exhibits aconitase activity and is post-translationally regulated by iron. Unlike vertebrate IRP-1 proteins, GEI-22/ACO-1 does not bind RNA. Our data show that GEI-22/ACO-1 is expressed in intestine and hypodermis, which are organs that absorb and store nutrients, and is consistent with its role in energy homeostasis. We show that high concentrations of iron reduce the life span of worms. Finally, we demonstrate that iron regulates the amounts of ftn-1 and ftn-2 mRNAs, unlike in vertebrates, where ferritin H and L mRNAs are primarily regulated by translation by the IRP-IRE system.
Aconitases are conserved [4Fe-4S] cluster proteins found in prokaryotes and eukaryotes that catalyze the reversible isomerization of citrate and isocitrate in the Krebs and glyoxylate cycles. Aconitases can be subdivided into several groups (5). One group is represented by IRP-1, which shares Ͼ47% identity with mammalian IRP-1, whereas the other group is represented by mitochondrial aconitase, which shares only ϳ24% identity with IRP-1. Since GEI-22/ACO-1 and other IRP members are so similar to vertebrate IRP-1s, which are RNAbinding proteins, it was expected that GEI-22/ACO-1 also binds RNA. Our data show that GEI-22/ACO-1 lacks RNA binding activity, which is consistent with the lack of discernible IREs in the C. elegans genome. Other IRP-1s, such as T. brucei (17), A. thaliana (13), and Nicotiana tabacum (15), were shown to be aconitases, but RNA binding activity was not reported, and it is probable that these proteins also do not bind RNA. However, some nonvertebrate IRP-1s bind RNA. For example, P. falciparum IRP-1 was reported to bind a mammalian consensus IRE (22), and B. subtilis aconitase binds not only to a mammalian consensus IRE but also to IRE-like stem-loop structures present in bacterial genes (11). Whether these interactions are functional in vivo is not clear. Overall, these studies show that RNA binding is not a requisite feature of the IRP-1 family.
Iron post-translationally regulates GEI-22/ACO-1 activity. In GEI-22/ACO-1-transfected HEK 293 cells, FAC and DF increased and decreased aconitase activity, respectively. In worms, total aconitase activity was 14% lower in worms grown in DF compared with FAC. Because FAC and DF have no effect on the amounts of GEI-22/ACO-1 protein or mRNA, this sug-

FIG. 7. C. elegans ftn-1 and ftn-2 mRNAs are regulated by iron.
A, embryos were grown on NGM plates containing FAC (0.003-6.6 mg/ml), DF (100 M), or no treatment (NT) for 4 days until they reached the L4 larval stage. Total RNA (25 g) was analyzed from each condition by a 1.2% formaldehyde-agarose gel. The RNA was transferred to a membrane that was sequentially probed with 32 P-labeled DNAs for ftn-1, act-1 (actin), ftn-2, and gei-22/aco-1. B, the data in A were quantified by PhosphorImager analysis and normalized to act-1, and band intensity was plotted relative to the untreated control. Two independent DF experiments were carried out. Data for FAC groups are presented as means Ϯ S.E. (n ϭ 3). The asterisks denote significant differences between untreated control and FAC groups using a paired Student's t test (p Ͻ 0.05). Ceftn-1 and 2, C. elegans ftn-1 and -2; Ceact-1, C. elegans act-1. gested that iron regulates GEI-22/ACO-1 activity by the assembly and disassembly of the [4Fe-4S] cluster. These studies also show that the [4Fe-4S] cluster machinery in human cells is efficient in assembling a cluster in GEI-22/ACO-1. Although we cannot attribute the changes in aconitase activity solely to GEI-22/ACO-1 because of contamination with mitochondrial ACO-2, the data indicated that altering the iron concentration in worms could regulate aconitase activity. Our data show that in worms grown in the presence of DF, the decrease in aconitase activity was much less than in HEK 293 cells transfected with GEI-22/ACO-1. The reasons for these differences are not clear, but we suspect that DF uptake in worms grown in liquid cultures might not be as efficient as DF uptake into cultured cells.
Despite information regarding the role of aconitase in mitochondria and in the glyoxylate cycle, the function of cytosolic aconitase in higher eukaryotes is not clear. The glyoxylate pathway takes place in plants and microorganisms, allowing these organisms to convert lipids to gluconeogenic precursors for the synthesis of carbohydrates. In plants, IRP-1 participates in the glyoxylate pathway (14,39), where its expression increases with the growth of seedlings and coincides with the increase in activity of other glyoxylate enzymes (14). A recent study shows that aconitase along with other glyoxylate and Krebs cycle enzymes are increased in A. thaliana during infection with Pseudomonas syringae, suggesting a role for aconitase in plant defense against pathogens (40). In T. brucei, IRP-1 expression is developmentally regulated, increasing in procyclic stage when Krebs cycle enzymes also increase (17). Whether IRP-1 functions in a glyoxylate pathway in T. brucei is unclear. In mammals, the role of IRP-1 was hypothesized to be important in maintaining redox control during periods of iron excess (41). Finally, some studies have hypothesized that the glyoxylate pathway may be operative in mammals during periods of stress, such as during starvation (42) and alcohol consumption (43); however, this has yet to be tested.
Because the glyoxylate cycle is operative in C. elegans (44,45), it is plausible that GEI-22/ACO-1 is involved in this pathway. In C. elegans, the activities and expression of the glyoxylate cycle protein increase when carbohydrates are low and fatty acids are elevated, which would occur in embryos and in larva during fasting (46 -48). We examined the expression of GEI-22/ACO-1 in L1 to L4 larval stages and in adults and found no differences in amounts of GEI-22/ACO-1. 2 It is possible, however, that GEI-22/ACO-1 activity might be altered during these conditions. Other functions for GEI-22/ACO-1 are suggested by our data showing that the expression patterns of GEI-22/ACO-1 are not identical with glyoxylate cycle protein.
For example, glyoxylate cycle protein is expressed in body wall muscle and intestine (47,48), whereas GEI-22/ACO-1 is expressed in hypodermal seam cells and intestine but not in body wall muscle. Furthermore, a recent study showed that GEI-22/ ACO-1 interacted with GEX-3 in a yeast two-hybrid screen (49). GEX-3 is thought to be a Rac-interacting protein, which is essential for proper cell migration and cell morphology, and in its absence, hypodermal cells fail to become organized (50). The localization of GEI-22/ACO-1 to hypodermal cells and its interaction with a GEX-3 suggest that GEI-22/ACO-1 might have a distinct function in development in hypodermis. However, when gei-22/aco-1 expression was inhibited by RNA interference, mutant worms showed no developmental abnormalities (49). These data suggested that gei-22/aco-1 may have other roles in hypodermal cells or that aco-2 can substitute for gei-22/aco-1 function.
To understand how C. elegans controls iron homeostasis, the regulation of C. elegans ftn-1 and ftn-2 mRNAs was studied. Ferritins are conserved iron storage proteins found in bacteria, plants, and animals that can store up to 4,500 iron atoms and, by so doing, prevent iron toxicity (38,51). Vertebrate ferritins are generally composed of H and L subunits (FER-H and FER-L), whereas those of plants and bacteria contain mainly H subunits. FER-H contains amino acids that are required for ferroxidase activity, whereas the FER-L functions in enhancing iron nucleation and long term iron storage. In mammalian cells, FER-H and FER-L are regulated at the translational level by the IRP-IRE system. Iron-dependent transcriptional regulation of ferritin has been described in mammalian cells (52), but this plays a minor role in iron regulation.
Since GEI-22/ACO-1 does not bind RNA, and C. elegans lacks obvious IREs, we hypothesized that ftn-1 and ftn-2 mRNAs might be regulated by iron through processes other than translation. Our data showed that ftn-1 mRNA increased ϳ2-fold in worms grown in the presence of iron, whereas ftn-2 mRNA showed a small but consistent ϳ0.5-fold increase. In contrast, DF treatment decreased ftn-1 and ftn-2 mRNAs ϳ10-fold and ϳ2-fold, respectively. One explanation for the ϳ2-fold increase in ftn-1 mRNA by iron is that worms are grown in iron-replete medium and fed E. coli and are iron-sufficient. The data also show that ftn-1 mRNA is more responsive to iron concentration than ftn-2 mRNA. A similar situation occurs in A. thaliana, where the fer-1 and fer-3 mRNAs are more responsive to iron than fer-2 or fer-4 (53), and in maize, where fer-1, but not fer-2, is iron-regulated (54 -56). Although our data do not conclusively show that ftn-1 and ftn-2 genes are transcriptionally regulated by iron in worms, this mechanism would be consistent with the observed transcriptional regulation of ferritin in plants (53,56,57) and in insects (58).
The response of C. elegans to heavy metals, such as copper, arsenate, and cadmium (59 -64), has been investigated, but less is known about iron regulation and toxicity in worms. Our first experiment was to determine the effect of iron on the development of worms. No significant change was observed in the timing of development of worms grown on FAC-supplemented plates. High concentrations of iron, however, did decrease the life span of worms. The mean life span of worms grown on 3.3 and 6.6 mg/ml FAC plates was 12.9 Ϯ 2.04 and 11.1 Ϯ 1.2 days, respectively, compared with worms grown on control plates (15.8 Ϯ 0.70 days). Although the mechanism responsible for iron-mediated decrease in life span is not known, it is possible that worms maintained on high iron would accumulate iron as they age, leading to increased iron-catalyzed free radical formation. An elevation in the concentrations of free radicals would overwhelm the antioxidant defense machinery, leading to cellular damage and eventual cell death. Studies show that when reactive free radicals are decreased (65,66) or antioxidant enzymes, such as catalase and superoxide dismutase, are increased (67)(68)(69), the life span of the organism can be increased. The toxicity of iron in worms may be akin to iron overload diseases in humans, where the accumulation of iron over a lifetime can result in cirrhosis, cardiomyopathy, diabetes mellitus, neurodegeneration, and increased risk of cancer (70). These studies show that C. elegans may serve as a model system for the study of iron homeostasis in mammals.
Imaging Core Facility for assistance in worm imaging and Dr. Aniko Szabo at the Huntsman Cancer Institute Biostatistics Shared Resource Center for statistical analysis.