The Aconitase Function of Iron Regulatory Protein 1

Iron regulatory proteins (IRP) are sequence-specific RNA-binding proteins that mediate iron-responsive gene regulation in animals. IRP1 is also the cytosolic isoform of aconitase (c-aconitase). This latter activity could complement a mitochondrial aconitase mutation (aco1) in Saccharomyces cerevisiae to restore glutamate prototrophy. In yeast, the c-aconitase activity of IRP1 was responsive to iron availability in the growth medium. Although IRP1 expression rescuedaco1 yeast from glutamate auxotrophy, cells remained growth-limited by glutamate, displaying a slow-growth phenotype on glutamate-free media. Second site mutations conferringenhanced cytosolicaconitase-dependent (ECA) growth were recovered. Relative c-aconitase activity was increased in extracts of strains harboring these mutations. One of the ECA mutations was found to be in the gene encoding cytosolic NADP+-dependent isocitrate dehydrogenase (IDP2). This mutation, an insertion of a Ty delta element into the 5′ region of IDP2, markedly elevates expression of Idp2p in glucose media. Our results demonstrate the physiological significance of the aconitase activity of IRP1 and provide insight into the role of c-aconitase with respect to iron and cytoplasmic redox regulation.

The iron regulatory proteins (IRP) 1 are a small family of sequence-specific RNA-binding proteins that mediate gene regulation by binding to iron-responsive elements (IRE) located in either the 5Ј-or 3Ј-untranslated region of a variety of animal cell mRNAs. IRE-containing mRNAs encode proteins of iron storage and transport as well as proteins involved in iron utilization or intermediary metabolism (for review see Refs. 1 and 2). IRPs exert their effect through translational regulation (for ferritin, m-aconitase, and erythroid aminolevulinate synthase (eALAS) mRNAs) or by controlling mRNA stability (for transferrin receptor (TfR) mRNA), depending on the location of the IRE (1-3). Two IRPs have been identified to date, called IRP1 and IRP2 (4 -7). Both bind similar IRE sequences and appear to be capable of mediating iron-responsive gene regulation, although each has a distinct IRE preference (8 -10). IRP1 is a bifunctional protein, having activity as an IREbinding protein (IRE-BP) or as the cytosolic isoform of aconitase (c-aconitase) (1,11). The two activities of IRP1 are mutually exclusive (1,(11)(12)(13). Interconversion of IRP1 between an IRE-BP and c-aconitase is itself regulated by iron, through the assembly/disassembly of a [4Fe-4S] cluster (1,(11)(12)(13). Fe-S cluster assembly occurs under conditions of excess iron, converting IRP1 to c-aconitase and stimulating synthesis of ferritin, m-aconitase, and eALAS, while TfR synthesis is depressed. Iron depletion promotes cluster disassembly, conversion of caconitase to IRE-BP, and repression of ferritin, m-aconitase, and eALAS synthesis, while stimulating TfR expression. In contrast to IRP1, IRP2 has only the IRE binding activity and is regulated by iron through protein degradation via the proteasome pathway (1, 14 -16).
The control of IRP activity is also subject to regulation by other factors. Both IRP1 and IRP2 are phosphorylated in vivo, apparently through the action of protein kinase C (PKC) (17,18). Treatment of cells with PKC stimulators results in an increase in IRE binding activity and TfR mRNA abundance concomitant with an increase in IRP phosphorylation (17,18). The Fe-S cluster may be the target of phosphoregulation of IRP1. Mutation of serine 138 of IRP1, a site of PKC phosphorylation, to amino acids that mimic phosphoserine results in oxygen-dependent Fe-S cluster instability (19). This observation suggests that the Fe-S cluster of phosphorylated c-aconitase is more susceptible to cluster disassembly in response to oxidants, which likely is part of the normal mechanism of cluster turnover in IRP1 (19). Exposure of cells to nitric oxide (NO) or hydrogen peroxide enhances IRE binding activity and inhibits c-aconitase activity in animal cells (reviewed in Ref. 2). Both NO and H 2 O 2 cause Fe-S cluster disruption (20,21). The consequence of these regulatory processes would be to modulate the expression of genes regulated by IRP1 and c-aconitase activity, similar to the effect of iron on IRP1.
The dichotomy between the role of IRP1 in the post-transcriptional regulation of genes involved in iron metabolism and its direct function as an enzyme of intermediary metabolism is intriguing. To gain insight into the significance of c-aconitase activity in vivo, we expressed IRP1 in aconitase-deficient (aco1) Saccharomyces cerevisiae and investigated conditions that affected its ability to provide c-aconitase activity for cell growth. We found that hyperexpression of cytosolic NADP ϩ -dependent isocitrate dehydrogenase (Idp2p) enhanced the ability of IRP1 to provide aconitase function to aco1 yeast and altered the extent of interconversion of IRP1 between IRE-BP and c-acon-itase. These results provide insight into the possible role of c-aconitase in animal cells.
Plasmid Constructs-Constitutive expression of IRP1 in yeast was achieved by placing the rabbit IRP1 cDNA (27) downstream of a minimum yeast alcohol dehydrogenase I (ADHI) promoter, obtained from plasmid pAAH5 (28,29). Plasmid pYADFRP, which carries the chimeric ADHI/IRP1 gene also contains a 2 origin of replication and a URA3 gene for selection in yeast. To express IRP1 from a low copy vector, the ADHI/IRP1 chimeric gene on pYADFRP was excised as a SpeI to XbaI fragment and ligated into the SpeI site of plasmid pRS316 (23) to generate pYLC6. IRP1 mutants were constructed by site-specific mutagenesis, changing the codons encoding cysteines at position 437 or 503 in IRP1 to ones encoding serine using the Altered Sites Mutagenesis kit (Promega Biotech). A BspHI to XbaI fragment containing the mutated cDNA was used to replace the wild-type equivalent fragment in pYADFRP. The authenticity of the resulting constructs was confirmed by DNA sequencing. The IRP1 encoded by these mutants retained IRE binding activity.
To generate a plasmid carrying the yeast mitochondrial aconitase gene, ACO1 (30) was amplified from genomic DNA of a wild-type yeast strain using polymerase chain reaction (PCR) with the Expand Long Template PCR system (Roche Molecular Biochemicals). pGEMACO1 was generated by cloning the 3.1-kbp-amplified ACO1 gene into pGEM-T-Easy using TA cloning (Promega). Restriction enzyme analysis as well as DNA sequencing verified the authenticity of the cloned gene. A BamHI to SacI fragment from pGEMACO1 containing the complete ACO1 gene was cloned into the yeast shuttle vector pRS316 (23) to yield pRSACO1.
This placed the IRE encoding sequence into the 5Ј-untranslated region of the luciferase gene. Luciferase genes with or without the IRE were excised from the pGL3 vector as a SmaI to SalI fragment and cloned into the EcoRV to SalI sites of pRSADC1, placing the genes under the transcriptional direction of the minimal ADHI promoter on pRS313 (23). The plasmid encoding the luciferase gene containing the IRE is called pYab3. The plasmid without the IRE sequence is called pYaa6.
PCR Amplification of IDP2-Genomic DNA for use in PCR analysis of IDP2 gene structure was prepared from 1-ml overnight cultures of the strain of interest. Yeast were washed with 1 ml of sterile water and resuspended in 200 l of water, and 400 l of lysis buffer (10 mM Tris-HCl, pH 8.3; 50 mM KCl; 2.5 mM MgCl 2 ; 0.1 mg/ml gelatin; 0.45% Nonidet P-40; 0.45% Tween 20; 60 g/ml proteinase K) was added to each tube followed by incubation at 55°C for 1 h. Proteinase K was inactivated by heating at 95°C for 5 min, and cell debris was removed by centrifugation at 14,000 rpm for 10 min. The supernatant, which contained the genomic DNA, was used to PCR-amplify a 914-bp region of the IDP2 gene from Ϫ381 to ϩ533 (relative to the AUG translation start codon (31)) using the following primers: primer 1, 5Ј-CCGGGTT-AGCCGGAGAGGGTGACA-3Ј; primer 2, 5Ј-ACCCCACCATGTTCTGG-GTAGTCA-3Ј.
Genomic Library Construction and Screening-Genomic DNA was prepared using Promega's genomic DNA isolation kit according to the manufacturer's protocol. Genomic DNA was partially digested with Sau3AI and size-fractionated on a 10 -40% sucrose gradient by centrifugation at 22,000 rpm for 22 h at 20°C in a SW 50.1 rotor. After fractionation, genomic DNA (average size 12 kbp) was concentrated by ethanol precipitation, ligated into the yeast shuttle vector YCp50 at the BamHI sites, and transformed into E. coli strain XL1-Blue MRFЈ (Stratagene).
The genomic library was screened for genes conferring enhanced c-aconitase function by transformation into the slow growing aco1 strain, 1103a (aco1, ura3-52, his3-⌬200, trp1-⌬63, ade2-101), which had been transformed with IRP1 carried on a 2 /HIS3 plasmid. A total of 120,000 transformants were screened. Because of the relatively high background of spontaneous fast-growers generated relative to the frequency with which the mutant gene was expected to occur in transformants, we developed a routine method to eliminate fast-growing transformants that had resulted from spontaneous chromosomal mutations. Transformants were first plated onto media lacking glutamate and histidine, and fast-growing strains were selected. These were then replica-plated onto selective media lacking glutamate and uracil. Here only cells that had taken up a library clone would grow, and only those that had acquired a gene conferring fast growth would grow rapidly. Transformants that fell into this category were then counterselected on media lacking glutamate and containing 5-fluoro-orotic acid (FOA) (32). Cells that required the genomic library clone for fast growth would fail to grow rapidly on FOA media and were likely candidates for having acquired a gene from the library that conferred fast growth. Of the 120,000 transformants, 160 grew rapidly on media lacking glutamate and histidine, but only 4 of these transformants also required the library clone for fast growth. These four transformants were analyzed further.
Preparation of Yeast Cytoplasmic Extracts and Enzyme Assays-The preparation of yeast cytoplasmic extracts and performance of aconitase assays was as described elsewhere (19). Extracts for NADP ϩ -dependent isocitrate dehydrogenase assay were made, and assays were performed as described in Ref. 31.
Protein Immunoblot-Aliquots of cytoplasmic extracts (15 g) were separated on 10% SDS-polyacrylamide gels and transferred to nitrocellulose membranes using a Trans-blot semi-dry transfer unit (Bio-Rad). The membrane was blocked with 5% nonfat milk in phosphate-buffered FIG. 1. Analysis of growth of IRP1-transformed aco1 yeast. aco1 yeast was transformed with a chimeric gene encoding rabbit IRP1 (1150-IRP1) or with the yeast ACO1 gene (1150-ACO1). Cells from a fresh overnight culture, grown in selective medium supplemented with glutamate, were washed twice with sterile water and resuspended into 1 M sorbitol. Cells were serially diluted into 1 M sorbitol to give from 5 ϫ 10 5 to 50 cells in 10 l, which were spotted onto selective medium with or without glutamate as indicated, and without (Ϫ) or with (ϩ) BPS (indicated on the right). Where added, BPS was included at 50 M to reduce available iron in the medium. Shown is growth after 5 days of incubation at 30°C. saline containing 0.05% Tween 20 and incubated overnight with primary antibody to Idp1p/Idp2p (33) followed by incubation with an anti-rabbit secondary antibody. Reacting proteins were visualized using chemiluminescence with ECL (Amersham Pharmacia Biotech). Antibodies to Idp1p/Idp2p were kindly provided by Lee McAlister-Henn of the University of Texas Health Science Center/San Antonio.
Luciferase Assay-Yeast expressing IRP1 from the low copy plasmid pYLC6 were transformed with pYab3 (with IRE) or pYaa6 (without IRE), and three independent pYab3 transformants and two pYaa6 transformants were grown in selective synthetic media to mid-log phase. Cytoplasmic extracts were prepared as for aconitase assays. The luciferase activity in 20 ng of total extract protein was determined using Promega's luciferase assay system. Fold repression was calculated as the ratio of luciferase activity obtained from pYaa6-transformed yeast divided by the activity obtained from pYab3-transformed cells.

RESULTS
Complementation of Yeast aco1 by IRP1-Aconitase deficiency leads to glutamate auxotrophy in yeast (22). The aco1 yeast strain 1150 was transformed with a plasmid designed for constitutive expression of IRP1 and tested for growth on glutamate-free media. Results from a representative transformant clone (called 1150-IRP1) are presented. Expression of IRP1 in this strain restored glutamate prototrophy, although growth in the absence of glutamate was not as vigorous as in glutamatesupplemented media (Fig. 1). In contrast, restoration of wildtype mitochondrial aconitase by transformation with the yeast ACO1 gene resulted in vigorous growth in the presence and absence of glutamate (Fig. 1). To investigate whether it was the c-aconitase activity of IRP1 that was responsible for the rescue from glutamate auxotrophy, we constructed IRP1 mutants C437S and C503S and used them to transform the aco1 yeast. Cys-437 and Cys-503 have been implicated as ligands for the Fe-S cluster in IRP1 and therefore are predicted to be indispensable for aconitase function (34,35). Strain 1150, expressing either of these mutant IRP1s, failed to grow on glutamatefree media (not shown).
The aconitase activity in extracts of IRP1-transformed aco1 yeast was also examined. Aconitase activity was elevated greater than 20-fold in extracts of aco1 yeast expressing IRP1 than it was in nontransformed cells or yeast transformed with the C437S IRP1 mutant (Table I). To determine the amount of IRP1 that could function as c-aconitase, extracts were subjected to conditions promoting Fe-S cluster reconstitution and assayed for aconitase activity (36). Activity in extracts from 1150-IRP1 was elevated approximately 8-fold by this treatment, whereas the activity in nontransformed 1150 was unaltered (not shown). These results indicate that about 12% of the total amount of IRP1 is present as active c-aconitase in 1150-IRP1.
The addition of iron to yeast culture media did not result in an increase in c-aconitase activity. However, depletion of available iron in the medium by addition of the ferrous iron chelator BPS did cause reduction in aconitase activity (Table I). Subsequent addition of excess iron to iron-depleted medium restored c-aconitase activity. Given the inhibition of c-aconitase by iron depletion, the effect of limiting iron on cell growth was examined. Inclusion of 50 M BPS in growth media supplemented with glutamate did not significantly effect growth of 1150-IRP1 (Fig. 1). In sharp contrast, growth of these yeast on glutamatefree medium was strongly inhibited by 50 M BPS. The addition of this amount of BPS did not inhibit growth of cells expressing m-aconitase (ACO1) with or without glutamate supplementation (Fig. 1). The correspondence between the effects of BPS on growth of 1150-IRP1 and c-aconitase activity (Table  I) suggests that cluster assembly or disassembly in IRP1 is responsive to cellular iron status and that this sensitivity is greater than that of m-aconitase or other iron-requiring activities in yeast.
Isolation of Mutants of 1150-IRP1 Displaying Fast Growth on Glutamate-free Media-To investigate the nature of the growth limitation of 1150-IRP1 in the absence of glutamate, we screened for spontaneous fast-growing mutants on glutamatefree medium. Colonies of 1150-IRP1 normally require at least 7-10 days to appear as very small colonies on media lacking glutamate. A total of 25 colonies appeared in 4 -5 days on lawns of 1150-IRP1 replica-plated onto media lacking glutamate. The  2. Growth characteristics of IRP1-transformed mutant aco1 yeast on media lacking glutamate. Strain 1150 transformed with pYADFRP was grown overnight in YPD medium, and aliquots containing approximately 10 8 cells were spread onto the YPD plates. After a further overnight incubation, cells were replica-plated onto selective media lacking glutamate. Twenty-five colonies appeared as colony outgrowths in a screen of 10 plates. Shown is the growth after 5 days, ϩ and Ϫ glutamate (as indicated), of six of the resulting strains isolated in this screen. These strains represent the range of growth characteristics seen among the isolated strains. approximate frequency at which these faster-growing colonies arose was between 10 Ϫ8 and 10 Ϫ9 . These isolates do not appear to be all independent (see below). An analysis of growth of six of these strains compared with 1150-IRP1 is shown in Fig. 2. All strains grew similarly on medium supplemented with glutamate (Fig. 2, left panels). In contrast, the mutant strains grew significantly faster than 1150-IRP1 on medium lacking glutamate (Fig. 2, right panels). None of these strains grew on glutamate-free media after losing the IRP1 transgene, or when retransformed with either the C437S or C503S IRP1 mutants (not shown). Several of these strains grew nearly as well in the absence of glutamate as in its presence (for example ECA122 and ECA125; Fig. 2). Because the fast growth phenotype of these strains was dependent on the c-aconitase activity of IRP1, we have designated them ECA for enhanced cytosolic aconitase-dependent growth.
Analysis of IRP1-dependent Aconitase Activity in ECA Strains-The aconitase activity in extracts of the most robust ECA strains (ECA121, ECA122, and ECA125) was examined. Aconitase activity measured from each strain was markedly higher than that measured in extracts of the parent, 1150-IRP1 (Fig. 3). We estimate that 20 -40% of IRP1 was active as c-aconitase in extracts from these ECA strains compared with about 12% in extracts of the 1150-IRP1 strain. Glutamate pools were also significantly increased in these ECA strains in comparison to 1150-IRP1, consistent with their improved growth in glutamate-free media and increased aconitase activity (not shown). These results suggest that IRP1 was either converted at a higher rate to c-aconitase or that c-aconitase was more stable in these ECA strains.
Differential Sensitivity of ECA Strains to Iron Depletion-The ECA strains were examined for the sensitivity of aconitasedependent growth to iron depletion caused by addition of BPS to the growth medium. Cytoplasmic aconitase activity and caconitase-dependent growth of 1150-IRP1 was sensitive to iron deprivation ( Fig. 1 and Table I). To examine the effect of iron deprivation on c-aconitase-dependent growth quantitatively, growth in the presence of BPS was measured in liquid culture. Specific effects of iron deprivation on IRP1 function were revealed by examining the ratio of exponential growth rates in the absence and presence of glutamate (relative growth rate) as a function of BPS concentration. The relative growth rate for 1150-IRP1 showed a gradual but steady decline with increasing BPS concentration from 0 to 50 M (Fig. 4). No inhibition of growth rate was observed in yeast grown in glutamate-supplemented media with Յ20 M BPS (not shown). Maximum inhibition of relative growth rate was attained at 50 M BPS. Addition of Ն50 M BPS to the growth medium depressed growth rate in the presence and absence of glutamate, although inhibition was stronger in the absence of glutamate.
The effect of BPS on growth of the ECA strains showed striking differences. The relative growth rate of ECA122 and ECA125 was maximally inhibited at 20 M BPS (Fig. 4). In media supplemented with glutamate, no inhibition of growth of ECA122 or ECA125 was detected at 20 M BPS, whereas growth in the absence of glutamate was inhibited by close to 70%. These results suggest that the ECA phenotype of these strains, and perhaps the enhanced c-aconitase activity itself, also exhibit enhanced iron dependence relative to other cellular activities. In contrast, growth of ECA121 was slightly more resistant to iron depletion than the 1150-IRP1 strain and significantly more resistant than either ECA122 or ECA125. The exponential growth rate () was calculated, and the ratio (ϩ glutamate)/ (Ϫ glutamate) was determined at each BPS concentration. This ratio in the absence of BPS was set at 1.0, and the ratios calculated at each BPS concentration were plotted relative to that value. If BPS did not inhibit growth or did inhibit growth of cells equally in the presence or absence of glutamate, then the relative exponential growth rate would remain at 1.0. A differential effect of BPS on cell growth, dependent on the presence or absence of glutamate, would be revealed by a decrease (or increase) in the relative growth rate and suggests a specific effect of iron availability on IRP1 function.

Identification of a Yeast IDP2 Gene Mutation That Confers
the Fast Growth Phenotype-The ECA125 strain was selected to begin the identification of genes that conferred the fast growth phenotype. We first examined the inheritance of the fast growth phenotype of this strain. ECA125 was cured of the plasmid carrying the IRP1 gene and then crossed with YPH500, an ACO1 strain. The resulting diploid was sporulated, tetrads were dissected, and haploid yeast strains were generated upon germination of the resulting spores. Segregation of aco1 was 2:2 in the haploid progeny. Sixty-eight aco1 strains resulting from this process were transformed with IRP1 and examined for growth on glutamate-free medium. Forty-two of the aco1 strains gave rise to transformants displaying the fast-growth phenotype similar to ECA125, whereas the remaining 26 strains yielded transformants with the slow-growth phenotype similar to the 1150-IRP1 strain. The higher number of fast-growing aco1 strains suggests a linkage between aco1 and the gene mutated to give the fast-growth phenotype. Matings between aco1 haploid strains revealed that the mutation conferring the fast-growth phenotype was dominant.
To identify the mutant gene in this strain, a genomic library was constructed using DNA isolated from one of the spore strains, and the library was used to transform a slow growth, IRP1-transformed aco1 strain to fast-growth (see "Experimental Procedures"). Four fast-growth transformants that were dependent on IRP1 and the resident genomic clone were obtained in a screen of 120,000 transformants. Restriction endonuclease patterns obtained for each of these clones suggested that they were overlapping (not shown).
The ends of the insert in the genomic clone containing the smallest insert were sequenced and compared with the yeast genome. One end of the genomic fragment showed identity to the 3Ј-end of the gene encoding an isoform of NADP ϩ -dependent, isocitrate dehydrogenase (IDP2), the cytosolic isoform of isocitrate dehydrogenase (31). Surprisingly, the sequence at the other end of the 2-kbp genomic clone showed homology with Ty delta elements (37), and not to the 5Ј-end of the IDP2 gene. The fact that neither Ty nor delta elements have been found in the yeast genome at the position predicted by this clone suggests that an insertion of a Ty element in the 5Ј region of the IDP2 gene occurred in the ECA125 strain.
PCR primers were generated to probe the 5Ј region of the IDP2 gene in ECA125 and other ECA strains. Amplification of parental 1150 DNA or DNA of a wild type yeast strain (YPH500) yielded the predicted 914-bp-long fragment, nucleotides Ϫ381 to ϩ533 relative to the IDP2 translation start codon (Fig. 5). The PCR product from ECA125 was larger, approximately 1.3 kbp long, indicating the presence of an approximately 400-bp insert (Fig. 5). Amplification of DNA from ECA122 indicated a similar insert, whereas strain ECA121 gave the 914-bp fragment seen with DNA from 1150 and wildtype yeast (Fig. 5). Examination of the other ECA strains for the presence of this insert in the IDP2 gene showed that most of the other ECA strains gave the 914-bp product seen with strain 1150 DNA. However, seven of these ECA strains gave the 1.3-kbp-long amplified product, indicating the presence of the ϳ400-bp insert in the 5Ј region of the IDP2 gene (not shown) and suggesting that these isolates may not be independent.
Analysis of Expression of Idp2p in ECA Strains Carrying the IDP2 Insertion Mutation-It has been documented that insertion of the yeast retrotransposon Ty into promoter regions can alter gene expression (37). To investigate the effect of the Ty insertion mutation on Idp2p expression, the relative level of Idp2p was determined by protein immunoblot (see "Experimental Procedures"). A clear band corresponding to Idp2p was seen in extracts of strains harboring the mutant IDP2 gene, such as ECA122 and ECA125 (Fig. 6A). This protein band was undetectable in extracts of ECA121 or the parental 1150 strain (Fig.  6A). Idp2p was undetectable in other ECA strains that lacked the 400-bp insert, whereas all strains that carried the insertion mutation in IDP2 showed variable but highly elevated expression of Idp2p (not shown).
We measured NADP ϩ -dependent isocitrate dehydrogenase activity in unfractionated yeast extracts. A basal level of activity in all strains reflects the constitutive expression of Idp1p, the mitochondrial isoform of NADP ϩ -dependent isocitrate dehydrogenase (38) (Fig. 6A). ECA strains harboring a normal IDP2 gene (such as ECA121) gave enzymatic activity levels that were similar to 1150-IRP1 (Fig. 6B). In contrast, NADP ϩdependent isocitrate dehydrogenase activity was elevated significantly in strains in which Idp2p was up-regulated. For example, this activity was nearly 4.5-fold higher in extracts of ECA125 and more than 2-fold higher in ECA122 extracts when compared with this activity in extracts of the parental 1150-IRP1 strain (Fig. 6B). These results confirm that the insertion mutation in the IDP2 gene causes an increase in Idp2p expression and activity and suggest that elevated Idp2p activity pro- FIG. 5. PCR analysis of the promoter region of IDP2 in aco1 yeast. PCR amplification was performed on genomic DNA using primers upstream and within the coding regions of IDP2 (see "Experimental Procedures"). The amplification products obtained with ECA121, ECA122, ECA125, the parental strain (1150), and a wild-type strain (YPH500) are shown. PCR products were separated on a 0.8% agarose gel and detected by staining with ethidium bromide. Shown is a reverse image of the gel that has been overexposed to more clearly show the products amplified with DNA from strains ECA122 and ECA125, which do not amplify as efficiently as the smaller products obtained with DNA from the other strains. The arrows mark the position of the wild-type fragment (lower) and the fragment containing the insert (upper). DNA size markers are shown on the right. motes enhanced IRP1-dependent c-aconitase activity.
The IDP2 Mutation Segregates with the Fast-growth Phenotype in aco1 Yeast-Haploid strains obtained from the cross of ECA125 with YPH500 were analyzed for the segregation of the IDP2 mutation and elevated expression. Fig. 7A shows a protein immunoblot of Idp2p in extracts from the four spores of one tetrad. Idp2p was detected in extracts of spore a and spore d strains, whereas it was undetectable in extracts of spore b and spore c strains (Fig. 7A). The delta element insertion in the IDP2 gene also segregated 2:2 and cosegregated with the elevated expression of Idp2p seen in Fig. 7 (data not shown). The two aco1 spores of the tetrad were analyzed for IRP1-dependent growth. Spore d, which harbors the mutation in IDP2, grew much faster than spore b, which lacks the mutation (data not shown).
The dual activities of IRP1 are mutually exclusive and so an enhancement of c-aconitase is expected to coincide with a decrease in IRE binding activity (1,2). To investigate how the hyperexpression of Idp2p affected IRP1 function as an IREbinding protein in vivo, we examined IRP1-mediated translational repression in strains generated from spores b and d shown in Fig. 7A. These strains, called 0615b and 0615d, respectively, were transformed with a firefly luciferase gene en-coding mRNA either containing or lacking an IRE at the 5Ј end (see "Experimental Procedures"). IRP1 inhibited synthesis of luciferase from the IRE containing mRNA in 0615b much more strongly than in the 0615d strain (Fig. 7B). This suggests that a lower amount of IRP1 was active as an IRE-binding protein in 0615d than in 0615b. These results are consistent with the conclusion that hyperexpression of Idp2p increases the proportion of IRP1 that gets converted to, or remains as c-aconitase in yeast.

DISCUSSION
The role of IRP1 as c-aconitase in animal cells has been overshadowed by the focus on its function as a regulator of gene expression. In fact, IRP1 exists predominantly as c-aconitase in some tissues, particularly liver (11,39). Reconstituted in yeast, the ability of IRP1 to function physiologically to provide caconitase activity is evident. Here we show that a mutation that leads to hyperexpression of cytosolic isocitrate dehydrogenase enhances the ability of yeast to utilize IRP1 as c-aconitase. Because animal cells also have a cytosolic isocitrate de-FIG. 6. Analysis of Idp2p expression in aco1 yeast. The effect of the insertion mutation on IDP2 expression was investigated by determining Idp2p levels by protein immunoblot (A) and by measuring NADP ϩ -dependent isocitrate dehydrogenase activity (B). A, aliquots of extracts from the parent and the mutant strains, as indicated, were separated on a 10% SDS-polyacrylamide gel, and proteins were transferred to nitrocellulose filters as described under "Experimental Procedures." Anti-yeast Idp1p/Idp2p antibodies (33), were used to probe for the presence of Idp2p. Idp1p, which is expressed constitutively, is also shown. B, the total NADP ϩ -dependent isocitrate dehydrogenase activity was measured by following the production of NADPH at 340 nm as described under "Experimental Procedures." This assay measures activity from Idp1p as well as Idp2p. Error bars, standard deviation.
FIG. 7. Analysis of Idp2p expression and IRP1 function in haploid, meiotic progeny of an ECA125/wt diploid. ECA125 was crossed with YPH500, a wild-type strain, and the resulting diploid was sporulated and tetrads were dissected. A, analysis of Idp2p levels in extracts of haploid progeny of one of the tetrads, called 0615 (spores a-d), by protein immunoblot. The immunoblot was performed as described for Fig. 6A. B, analysis of the translational repression function of IRP1 in normal and Idp2p-hyperexpressing aco1 yeast. Yeast strains 0615b (normal Idp2p expression) and 0615d (hyperexpression of Idp2p) were transformed with the IRP1 gene on pYLC6, and then with a luciferase gene on pYab3 (with an IRE) or pYaa6 (without an IRE; see "Experimental Procedures"). Transformed cells were grown to mid-log phase, and extracts were prepared and luciferase assays were performed as described under "Experimental Procedures." The strength of translational repression by IRP1 is indicated by the repression ratio, which is the ratio of luciferase activity obtained from the gene lacking the IRE divided by the activity obtained from the gene containing the IRE.
hydrogenase, c-aconitase in animal cells may well contribute normally to glutamate biosynthesis and other metabolic processes such as fatty acid metabolism (40 -43).
Depletion of available iron in yeast growth media inhibited c-aconitase activity and growth of IRP1-transformed aco1 yeast on glutamate-free media. That iron may regulate pathways involving c-aconitase in animal cells raises questions regarding the role and significance of iron in regulating metabolic pathways involving c-aconitase. In addition to converting isocitrate to ␣-ketoglutarate, which is a precursor in glutamate biosynthesis (38), the reaction catalyzed by Idp2p also produces NADPH. Idp2p is an important source of this cofactor in the cytosol (40,(42)(43)(44). NADPH is a key cofactor in cellular defenses against oxidative stress, particularly through its involvement in the thioredoxin and glutathione redox cycles (44 -46). We propose a model whereby regulation of IRP1/caconitase by iron coordinates NADPH levels with iron uptake, utilization, and storage. This would provide the cell with the reducing power to deal with the increased oxidative stress brought on by higher intracellular iron and an additional source of NADPH as a cofactor for ferric reductase (47,48). Ferric reductase has been shown to be an important component of iron transport systems in eukaryotes (49 -54). The increase in NADPH predicted to accompany the rise in c-aconitase activity in iron replete cells would provide additional reducing equivalents, allowing animal cells to maintain the redox balance in the cytoplasm during intensive iron transport. Moreover, the effect of iron on c-aconitase activity and downstream steps catalyzed by Idp2p would provide animal cells with a means to modulate NADPH production specifically. Increased c-aconitase-driven NADPH levels also would favor the Fe(II) state and thereby promote ferritin-mediated iron biomineralization and iron incorporation into heme (55,56). The evolution of c-aconitase as an iron-responsive regulator of ferritin synthesis may have been prompted by this redox-dependent regulation of iron storage and utilization.
It is not surprising that a mutation in the gene encoding the cytosolic isoform of isocitrate dehydrogenase alters glutamate synthesis in cells utilizing IRP1 as c-aconitase. Idp2p most likely drives the reactions toward glutamate by mass action in Idp2p-hyperexpressing cells. On the other hand, we would have predicted that hyperexpression of Idp2p would not effect the interconversion of IRP1 between the IRE-BP and c-aconitase. However, a higher percentage of IRP1 was converted to caconitase in strains that hyperexpressed Idp2p. This suggests that the increase in Idp2p activity either enhanced conversion of IRP1 to c-aconitase or stabilized c-aconitase once it was formed. At present, we cannot distinguish between these possibilities. The proportion of IRP1 that exists as c-aconitase is affected by oxidants produced during normal, aerobic metabolism in yeast (19). Therefore, it is possible that the anti-oxidant effects of elevated NADPH may have protected c-aconitase in these yeast. Alternatively, conversion of apo-IRP1 to c-aconitase could have been enhanced in these strains. Elevated NADPH would effect levels of reduced thioredoxin, which has been shown to reduce oxidized apo-IRP1 generated upon Fe-S cluster removal (57). Protein thiol reduction appears to be a necessary step in the assembly of Fe-S clusters in aconitases (36). Increased NADPH may also enhance Fe-S cluster assembly by increasing availability of Fe(II).
In animal cells, interconversion of IRP1 and c-aconitase and iron-responsive gene regulation respond to chelatable iron levels (58). In yeast, the transcription factor Aft1p responds to chelatable iron levels by tightly regulating iron uptake by controlling the expression of genes encoding the components of the high affinity iron transport system (59 -61). Therefore, we might expect that overexpression of IRP1, which could deplete the chelatable iron pool, would cause a net increase in iron accumulation in yeast. We did not observe a consistent increase in iron accumulation in strains overexpressing IRP1, suggesting that iron was not the limiting factor for cluster assembly in IRP1 in yeast (not shown). On the other hand, assembly of an Fe-S cluster in IRP1 in the cytosol of yeast could be limiting. We cannot rule out this possibility at the present time, but other cytosolic Fe-S proteins do exist in yeast, and so it is expected that the machinery for assembling Fe-S clusters in cytosolic proteins is present (62).
Growth of aco1 strains that hyperexpress Idp2p on glutamate-free media was very sensitive to the level of c-aconitase activity. This was most evident when iron availability was reduced in the growth media, a condition that significantly inhibited c-aconitase activity (Table I). These yeast strains were hypersensitive to iron depletion, in fact, suggesting that growth of these strains became limited by c-aconitase activity in low iron media. We have also observed effects on growth of an Idp2p hyperexpressing strain when expressing IRP1 mutants that have defects in c-aconitase function (19). Mutations in IRP1 that decreased c-aconitase activity in vivo strongly reduced growth of these yeast strains on glutamate-free media. Moreover, reduction in c-aconitase activity in vivo by lowering IRP1 expression in these strains also led to a much slower growth rate. 2 The responsiveness of the strains that hyperexpress Idp2p to the level of c-aconitase activity makes them very useful for the study of factors and conditions that affect IRP1 function in vivo.