Detection of a [3Fe-4S] Cluster Intermediate of Cytosolic Aconitase in Yeast Expressing Iron Regulatory Protein 1 INSIGHTS INTO THE MECHANISM OF Fe-S CLUSTER CYCLING*

Interconversion of iron regulatory protein 1 (IRP1) with cytosolic aconitase (c-aconitase) occurs via assem-bly/disassembly of a [4Fe-4S] cluster. Recent evidence implicates oxidants in cluster disassembly. We investi-gated H 2 O 2 -initiated Fe-S cluster disassembly in c-acon- itase expressed in Saccharomyces cerevisiae . A signal for [3Fe-4S] c-aconitase was detected by whole-cell EPR of aerobically grown, aco1 yeast expressing wild-type IRP1 or a S138A-IRP1 mutant (IRP1 S138A ), providing the first direct evidence of a 3Fe intermediate in vivo . Exposure of yeast to H 2 O 2 increased this 3Fe c-aconitase signal up to 5-fold, coincident with inhibition of c-aconitase activity. Untreated yeast expressing IRP1 S138D or IRP1 S138E , which mimic phosphorylated IRP1, failed to give a 3Fe signal. H 2 O 2 produced a weak 3Fe signal in yeast ex- pressing IRP1 S138D . Yeast expressing IRP1 S138D or a pool of [3Fe-4S] c-aconitase in cells harboring wild-type IRP1 or IRP1 S138A , which was enhanced by exposure of cells to H 2 O 2 . Our results suggest that ROS, such as H 2 O 2 , initiate Fe-S cluster disassembly in c-aconitase in vivo by causing rapid loss of the solvent-exposed Fe a from the [4Fe-4S] cluster. Wild-type IRP1 and IRP1 S138A accumulated as 3Fe c-aconitase in H 2 O 2 -treated cells, whereas the phosphomimetic IRP1 S138D and IRP1 S138E did not. These results support the view that phosphorylation at Ser 138 enhances ROS-mediated conversion of c-aconitase to IRP1 by promoting more rapid and complete loss of the Fe-S cluster.

Iron is an essential cofactor for many enzymes including those of the tricarboxylic acid cycle and electron transport chain. However, iron's participation in Fenton chemistry leads to the production of free radicals and damage to lipids, proteins, and DNA (1). The potential toxicity of iron necessitates that its metabolism be tightly controlled. Iron regulatory proteins 1 and 2 (IRP1 and IRP2) 1 are central regulators of iron homeostasis in animals (2,3). IRPs bind to conserved sequences called iron-responsive elements (IREs), which are found in either the 5Ј-or 3Ј-untranslated region of a diverse range of mRNAs encoding proteins involved in iron storage and transport and intermediary metabolism (4). IRP binding to IREs located in the 5Ј-untranslated region of mRNAs encoding ferritin (5,6), erythroid-5-aminolevulinate synthase (7,8), and mitochondrial aconitase (8) blocks translation initiation (4,9). Binding of IRPs to 3Ј IREs of transferrin receptor mRNA stabilizes the message, presumably by masking a degradation signal (10,11). Potential IREs have also been found in the mRNA encoding MTP1, a basolateral iron transporter in enterocytes (12,13), and one isoform of the mRNAs encoding DMT1, a divalent metal ion transporter (14,15). The involvement of these proteins in iron uptake from the diet and their potential for regulation by IRPs extend this regulatory system to whole body iron regulation (12)(13)(14)16). However, the extent to which the mRNAs encoding MTP1 and DMT1 are regulated by IRPs remains to be elucidated.
Both IRP1 and IRP2 are regulated in response to changes in cellular iron status, but through different mechanisms. At high intracellular iron, IRP2 is degraded (17,18). In contrast, IRP1 responds to high iron by assembling a [4Fe-4S] cluster, which switches its function from IRE binding to that of the cytoplasmic isoform of aconitase (c-aconitase; Refs. 19 -21). With the loss of IRE binding, translational repression of ferritin mRNAs is relieved, allowing for greater iron storage, and transferrin receptor mRNA is destabilized, resulting in reduced iron import (for a review, see Refs. 2-4 and 22). As intracellular iron levels decrease, the [4Fe-4S] cluster in c-aconitase is disassembled, resulting in loss of c-aconitase function and restoration of IRE binding activity. This coordinated regulation of gene expression maintains a balance between iron storage and uptake.
The mechanism by which cluster assembly and disassembly occurs in IRP1 in vivo is not fully understood. In vitro, oxidation of the [4Fe-4S] cluster of c-aconitase causes the rapid loss of the solvent-exposed fourth iron atom (Fe a ), generating a [3Fe-4S] cluster and inactive aconitase (2,20). This and other observations have led to the suggestion that oxidants play a key role in Fe-S cluster disassembly in c-aconitase in vivo (2). Treatment of animal cells with nitric oxide or H 2 O 2 results in an increase in IRE binding activity and decrease of c-aconitase activity (reviewed in Refs. 2, 3, and 22). IRE binding activity is also altered by hypoxia (23)(24)(25). Hypoxia appears to stabilize the Fe-S cluster of c-aconitase, thereby causing a lowering of IRP1's IRE binding activity (23,25). Degradation of IRP2 slows in hypoxic cells, resulting in an increase in its IRE binding activity (25). Whether hypoxia stimulates or inhibits IRE binding appears to depend on the contribution of IRP1 and IRP2 to the IRE binding activity of the cell (25).
Phosphorylation of IRPs appears to be another mode of regulating IRE binding activity. Both IRP1 and IRP2 are phosphorylated by protein kinase C (26). Stimulation of protein kinase C activity increased IRE binding activity and transferrin receptor mRNA levels (26,27). Protein kinase C can phosphorylate IRP1 at serine 138 and serine 711 (27). Mutations at serine 138 to residues that mimic phosphoserine destabilized the Fe-S cluster of c-aconitase, increasing its sensitivity to oxygen (28). Taken together, these results implicate phosphoregulation and reactive oxygen species (ROS) in the process of Fe-S cluster disassembly in c-aconitase.
In this study, we used wild type IRP1 and IRP1 Ser 138 mutants expressed in aco1 yeast to investigate Fe-S cluster disassembly initiated by H 2 O 2 . EPR of whole yeast cells revealed a pool of [3Fe-4S] c-aconitase in cells harboring wildtype IRP1 or IRP1 S138A , which was enhanced by exposure of cells to H 2 O 2 . Our results suggest that ROS, such as H 2 O 2 , initiate Fe-S cluster disassembly in c-aconitase in vivo by causing rapid loss of the solvent-exposed Fe a from the [4Fe-4S] cluster. Wild-type IRP1 and IRP1 S138A accumulated as 3Fe c-aconitase in H 2 O 2 -treated cells, whereas the phosphomimetic IRP1 S138D and IRP1 S138E did not. These results support the view that phosphorylation at Ser 138 enhances ROS-mediated conversion of c-aconitase to IRP1 by promoting more rapid and complete loss of the Fe-S cluster.
For treatment with H 2 O 2 , cell cultures were grown to midexponential phase. After collecting a presample, H 2 O 2 was added to the remaining culture to the indicated concentrations, and cultures were incubated at 30°C with shaking. At the indicated times, aliquots were removed, and cells were washed twice with an equal volume of sterile water and either prepared for EPR analysis or for making cytoplasmic extracts.
Whole Cell EPR-Yeasts expressing wild-type IRP1 or Ser 138 IRP1 mutants were grown in synthetic complete media without or with H 2 O 2 . Cells were washed twice with a volume of ice-cold water equal to the cell culture volume and once with a volume of ice-cold 10% glycerol equal to one-tenth the culture volume. At this point, cells were quick-frozen in liquid nitrogen for storage until EPR analysis. For EPR analysis, cells were packed into quartz EPR tubes by centrifugation in a low speed tabletop centrifuge at 4°C. Excess cell suspension buffer was removed, and the cell pellet was frozen by slow immersion in liquid nitrogen. EPR spectra were recorded with an X-band Varian E112 century series spectrometer at a temperature of 11-14 K using standard methods. Conditions of spectroscopy were as follows: microwave power, 1 milli-watt; microwave frequency, 9.235 GHz; modulation amplitude and frequency, 5 millitesla and 100 kHz; time constant, 0.064 s; scanning time, 2 min. Each spectrum represents an average of four scans. Subtraction of one spectrum from another using SUMSPEC92 software (obtained from the National Biomedical ESR Center, Milwaukee, WI) was employed in order to eliminate signals from manganese and other signals common to transformed and nontransformed cells. Subtraction of the spectrum of untreated or H 2 O 2 -treated, nontransformed aco1 yeast maximally reduced background signals and revealed a clear [3Fe-4S] c-aconitase signal in the spectra of untreated and H 2 O 2 -treated IRP1transformed aco1 yeast. The spectrum obtained with nontransformed yeast was unchanged by treatment of yeast with H 2 O 2 . Subtraction of the spectrum of untreated or H 2 O 2 -treated nontransformed aco1 yeast from the spectrum of IRP1-transformed yeast yielded identical results.
Cytoplasmic Extracts and Aconitase Assays-Preparation of yeast cytoplasmic extracts and aconitase assays were performed as described elsewhere (28,31). Anaerobic reconstitution of the c-aconitase Fe-S cluster with Fe(II) and Nifsp was performed as described in Ref. 32.
Growth Analysis-Cells from overnight cultures were washed twice with sterile water, suspended into sterile water, and serially diluted. Ten microliters containing from 2 ϫ 10 2 to 2 ϫ 10 5 yeast cells were spotted onto SD-ura-his plates, supplemented with or without glutamate and H 2 O 2 as indicated.
For growth analysis in liquid media, cells from a fresh overnight culture were washed twice with water and suspended in glutamate-free SD-ura media. Media were made iron-rich by adding 1 mM FeSO 4 and 1 mM ascorbate to assist low affinity iron transport (33). Media were made iron-depleted by adding 100 M bathophenanthroline sulfonate (BPS; Ref. 34).
IRE Binding Activity-IRE binding assays were performed as previously described (35). Briefly, 32 P-labeled transcript (8 -10 fmol; 10,000 cpm) containing the first 92 nucleotides of rabbit ferritin L-chain mRNA (including the IRE) was incubated with 200 ng of cytoplasmic extract protein in 25 l at 4°C for 20 min, followed by a 10-min incubation with heparin (2.5 mg/ml). This amount of extract gave subsaturating IRE binding under these conditions. Free transcript was separated from bound by 4% native PAGE (36). Quantitative determination of bound transcript was performed using a Molecular Dynamics PhosphorImager with ImageQuant software.

Analysis of c-aconitase Fe-S Cluster in Yeast by Whole Cell
EPR-Earlier studies suggested that the level of c-aconitase activity in IRP1-transformed aco1 yeast was limited by Fe-S cluster disruption mediated by ROS produced during normal metabolism (28,29). In vitro, exposure of 4Fe c-aconitase to oxidants resulted in its conversion to 3Fe c-aconitase, which can be detected by EPR (2,(37)(38)(39). To investigate whether 3Fe c-aconitase was being generated in aerobically grown yeast, intact IRP1-transformed aco1 yeast were subjected to EPR analysis. Since IRP1 was expressed in yeast lacking mitochondrial aconitase, an EPR signal for [3Fe-4S] c-aconitase, if present, could be detected in IRP1-transformed aco1 yeast when compared with nontransformed yeast. The EPR spectrum of purified 3Fe c-aconitase is in Fig. 1a, showing peaks at g values of 2.015 and 2.033, characteristic of 3Fe c-aconitase (20). In comparison, the spectrum obtained with yeast expressing wildtype IRP1 was much more complex but did show a distinct shoulder at a g value of 2.033 (Fig. 1b), which was absent from the spectrum of nontransformed aco1 yeast (Fig. 1c). Subtraction of the spectrum of nontransformed yeast (Fig. 1c) from that of transformed yeast (Fig. 1b) left a weak [3Fe-4S] signal of oxidized c-aconitase (Fig. 1d). The spectral peak occurred at the same g value and had the same characteristic line shape as that of purified [3Fe-4S] c-aconitase (Fig. 1, compare a with d).
The major underlying signal was that of the tyrosyl radical of ribonucleotide reductase (40), which partially contributes to the peak created by [3Fe-4S] c-aconitase. The effect of the ribonucleotide reductase signal was to diminish the trough seen on the shoulder of the 3Fe c-aconitase peak (Fig. 1, compare a with d). Nonetheless, the [3Fe-4S] c-aconitase signal was readily apparent in the subtracted spectrum (Fig. 1d). The existence of a pool of 3Fe c-aconitase in exponentially growing cells suggests that Fe-S cluster assembly and disassembly in IRP1 is a dynamic process in yeast.
The results presented above do not distinguish between the 3Fe c-aconitase detected in aerobically growing aco1 yeast being an intermediate of [4Fe-4S] cluster disassembly or of Fe-S cluster assembly. To investigate this further, IRP1-transformed yeast were treated with H 2 O 2 to attempt to induce cluster disassembly. Treatment of animal cells with H 2 O 2 results in an increase in IRE binding activity, suggesting complete Fe-S cluster disassembly caused by this oxidant in vivo (41,42). The unsubtracted EPR spectrum of yeast transformed with wild-type IRP1 and treated with 6 mM H 2 O 2 is shown in Fig. 1e. The major effect of H 2 O 2 on the unsubtracted EPR spectrum was to increase the signal at g value 2.033, which could now be seen as a distinct peak compared with the shoulder in the unsubtracted spectrum of untreated IRP1-transformed yeast (Fig. 1, compare e with b). The spectrum obtained with H 2 O 2 -treated nontransformed yeast was identical to the spectrum of untreated nontransformed yeast (not shown). Subtraction of the nontransformed cell spectrum from the transformed cell spectrum left a more intense [3Fe-4S] signal from the treated cells (Fig. 1f). The intensity of this signal was 5 times that of untreated yeast expressing wild-type IRP1 (Fig. 1, This indicates that there is at least 5 times more 4Fe c-aconitase than 3Fe c-aconitase in IRP1-transformed aco1 yeast growing aerobically, assuming that the increase in intensity of the 3Fe signal induced by H 2 O 2 is due solely to conversion of the [4Fe-4S] cluster to a [3Fe-4S] cluster. Enhancement of the 3Fe c-aconitase signal in IRP1-transformed yeast was H 2 O 2 dose-dependent at concentrations ranging from 15 M to 6 mM, and was seen as early as 15 min after exposure to H 2 O 2 (not shown).
Effect of Mutation at Ser 138 of IRP1 on the EPR Spectrum of IRP1-transformed aco1 Yeast-Mutation of serine 138 of IRP1 to aspartate or glutamate alters c-aconitase by making it more susceptible to inactivation by oxidants (28). The 3Fe c-aconitase detected by EPR in aerobically grown aco1 yeast transformed with wild-type IRP1 is probably generated from 4Fe c-aconitase by reaction with metabolically produced oxidants. Therefore, it was of interest to determine whether the Ser 138 c-aconitase mutants gave a similar 3Fe signal in vivo and whether the proportion of c-aconitase in the 3Fe form differed from that seen for wild-type c-aconitase. The S138A c-aconitase mutant displays sensitivity to inactivation by oxidants in vitro and functions as aconitase in vivo similarly to wild-type caconitase (28). Consistent with this, the EPR spectrum of untreated IRP1 S138A -transformed yeast was similar to that obtained with aco1 yeast transformed with wild-type IRP1, with a distinct shoulder at g ϭ 2.033 (Fig. 2a). A clear signal for a [3Fe-4S] cluster was left when the spectrum of nontransformed aco1 yeast was subtracted from the spectrum of IRP1 S138Atransformed cells (Fig. 2b). The intensity of this 3Fe signal also increased by about 5-fold upon treatment of cells expressing IRP1 S138A with 6 mM H 2 O 2 (Fig. 2, c and d).
Although both IRP1 S138E and IRP1 S138D support aconitasedependent growth of aco1 yeast, indicating that they are converted to c-aconitase in vivo ( Fig. 4; see also Ref. 28), no signal for [3Fe-4S] c-aconitase was detected with untreated yeast expressing these IRP1 mutants (not shown). A weak signal for FIG. 1. EPR spectra of yeast expressing wild-type IRP1. Yeast cells were grown in SD-ura-his medium without or with H 2 O 2 and prepared for EPR analysis as described under "Experimental Procedures." Scans b, c, and e are the spectra that resulted from EPR with intact yeast cells. Each spectrum represents the average of four scans.
Graphs d and f show the signal remaining after subtraction of the nontransformed cell spectrum from the spectra of transformed yeast. a, EPR spectrum of purified 3Fe c-aconitase; b, wild-type IRP1-transformed yeast; c, nontransformed yeast; d, subtracted spectrum of wildtype IRP1-transformed yeast; e, spectrum of wild-type IRP1-transformed yeast treated for 1 h with 6 mM H 2 O 2 ; f, subtracted spectrum of wild-type IRP1-transformed yeast treated with 6 mM H 2 O 2 for 1 h. The spectrum of nontransformed yeast was unchanged after treatment of cells with H 2 O 2 (not shown).

FIG. 2. EPR spectra of yeast expressing Ser 138 IRP1 mutants.
Yeast transformed with Ser 138 IRP1 mutants were grown and prepared for EPR analysis as described for Fig. 1. Scans a, c, and e are the spectra that resulted from EPR with intact yeast cells. Each spectrum represents the average of four scans. The graphs shown in b, d, and f were obtained by subtracting the spectrum obtained with nontransformed cells from the spectra shown in a, c, and e, respectively. a, IRP1 S138Atransformed yeast; b, subtracted spectrum of IRP1 S138A -transformed yeast; c, IRP1 S138A -transformed yeast treated for 1 h with 6 mM H 2 O 2 . d, subtracted spectrum of IRP1 S138A -transformed yeast treated for 1 h with 6 mM H 2 O 2 . e, IRP1 S138D -transformed yeast treated with 0.6 mM H 2 O 2 . f, subtracted spectrum of IRP1 S138D -transformed yeast treated with 0.6 mM H 2 O 2 . The doublet signal in f is mostly a signal from the tyrosyl radical and attributed to ribonucleotide reductase (40). 3Fe c-aconitase, masked by a more intense signal for the tyrosyl radical of ribonucleotide reductase, was detected in cells expressing IRP1 S138D after treatment with 0.6 mM H 2 O 2 (Fig.  2, e and f). The 3Fe c-aconitase signal from IRP1 S138D -transformed cells was evident as a shoulder superimposed on the tyrosyl radical signal from ribonucleotide reductase. The line shape of the signal in Fig. 2f is dominated by the tyrosyl radical with the typical doublet split by 20G (40). The 3Fe signal was absent from the spectrum of IRP1 S138D -transformed cells that had been exposed to 6 mM H 2 O 2 but was detected in spectra of cells exposed to concentrations of H 2 O 2 lower than 0.6 mM (not shown). This suggests that exposure to 6 mM H 2 O 2 caused cluster disassembly beyond the 3Fe stage in the S138D caconitase mutant. No signal was detected in cells expressing IRP1 S138E treated with 0.6 mM H 2 O 2 or at any other concentration of H 2 O 2 tested (not shown).
Analysis of the Fe-S Cluster of Oxidized Ser 138 c-aconitase Mutants in Vitro-Experiments were performed on cellular extracts in attempts to gain information on the relative concentrations of active c-aconitase in yeast transformed with wild-type IRP1 and each of the Ser 138 IRP1 mutants. As reported by Brown et al. (28), when cytoplasmic extracts were prepared from IRP1-transformed aco1 yeast, only extracts from yeast expressing wild-type IRP1 or IRP1 S138A retained substantial aconitase activity (see also Fig. 2 in Ref. 28). IRP1 S138D and IRP1 S138E had only 7% and Ͻ2% of the wild-type activity, respectively. These results are consistent with earlier in vitro studies, which showed that the aconitase activities of purified IRP1 S138D and IRP1 S138E were more sensitive to oxygen than that of wild-type IRP1 (28).
To determine the relative amount of each IRP1 that was available to become aconitase in these extracts, Fe(II), dithiothreitol, and Nifsp were added to each extract, followed by incubation under anaerobic conditions in order to reconstitute the c-aconitase Fe-S cluster. Aconitase activities were then determined. Aconitase in extracts of aco1 yeast expressing wild-type IRP1 was 0.99 M after anaerobic reconstitution. This level of c-aconitase represents about an 8-fold increase over its level in the unreconstituted extract. After anaerobic reconstitution, c-aconitase in extracts of yeast expressing IRP1 S138A , IRP1 S138D , or IRP1 S138E was 0.93, 0.73, and 0.59 M, respectively. The level of IRP1 protein expressed from the mutant genes versus the wild-type gene differed by Յ20% (not shown). Thus, a comparable level of IRP1 was converted to c-aconitase upon anaerobic Fe-S cluster reconstitution of mutant and wild-type IRP1s. The relative level of c-aconitase activity obtained from wild-type and mutant IRP1s after anaerobic reconstitution (Table I) is consistent with results of earlier studies using purified proteins (28).
The effect of oxidation on the Fe-S cluster in the reconstituted wild-type and Ser 138 IRP1 mutants was next investigated. Extracts, which had been subjected to anaerobic reconstitution, were concentrated ϳ5-fold, oxidatively inactivated with ferricyanide and analyzed by EPR to determine the relative 3Fe signal remaining. Ferricyanide was carefully titrated into each sample to inhibit c-aconitase activity by ϳ95%. A strong signal for 3Fe c-aconitase was detected in extracts containing reconstituted wild-type c-aconitase or the S138A mutant (Fig. 3). A much weaker and distorted signal was detected in extracts of cells expressing IRP1 S138D . No 3Fe c-aconitase signal was detected in extracts containing IRP1 S138E , although spectra from IRP1 S138E -transformed and nontransformed aco1 yeast contained a weak radical type signal (Fig. 3). Taking the wild-type 3Fe signal as 100%, the intensity of the signal from the S138A mutant was 67% of the wild-type signal. Remarkably, the signal from the S138D mutant was only 12% as intense as wild-type (Table I). By comparison, c-aconitase from the S138D mutant prior to ferricyanide oxidation was 74% of wild-type (Table I). Since c-aconitase had been inhibited by ϳ95% by ferricyanide oxidation in each case, the weak 3Fe signal from S138D c-aconitase and the lack of a signal from the S138E mutant indicate that most of the Fe-S cluster in the TABLE I Reconstitution of the c-aconitase Fe-S cluster in IRP1-transformed aco1 yeast extracts Aconitase-deficient yeast transformed with the indicated IRP1 were grown to midexponential phase, harvested, and extracts were prepared and subjected to anaerobic reconstitution as described under "Experimental Procedures." The concentration of c-aconitase after reconstitution was determined from enzymatic activity based on the relationship 3 units/nmol of C-aconitase. Extracts that had been subjected to anaerobic reconstitution were concentrated approximately 5-fold prior to ferricyanide oxidation and EPR analysis. Aconitase activity and 3Fe signal intensity are given relative to that in reconstituted extracts of aco1 yeast transformed with wild-type IRP1. ND, none detected.

IRP1
Relative aconitase activity Relative 3Fe signal Wild type 1.00 1.00 S138A 0.94 0.67 S138D 0.74 0.12 S138E 0.60 ND FIG. 3. EPR analysis of in vitro reconstituted Ser 138 c-aconitase mutants. Cytoplasmic extracts of yeast transformed with the indicated IRP1 were prepared, and c-aconitase was activated by anaerobic Fe-S cluster reconstitution with Fe(II) and nifS as described under "Experimental Procedures." For EPR spectroscopy, anaerobically reconstituted extracts were concentrated ϳ5-fold and subjected to ferricyanide oxidation. phosphomimetic c-aconitase mutants had been disrupted beyond the 3Fe stage. These results are consistent with the view that oxidant attack on the Fe-S cluster of the phosphomimetic c-aconitase mutants in vivo would cause more complete cluster disassembly than would be seen for wild-type IRP1.
Effect of H 2 O 2 on Growth of IRP1-transformed aco1 Yeast-Growth of IRP1-transformed aco1 yeast on media lacking glutamate is a measure of c-aconitase function in vivo (28,29). To further assess the effects of oxidative stress on the Ser 138 IRP1 mutants, aco1 yeast transformed with wild-type IRP1 or each of the IRP1 mutants were grown on media lacking glutamate and supplemented with H 2 O 2 . The addition of 6 mM H 2 O 2 to growth media strongly inhibited growth of these yeast with or without glutamate supplementation (not shown). The addition of 0.6 mM H 2 O 2 had only a modest effect on growth of yeast expressing wild-type IRP1 or IRP1 S138A on glutamate-supplemented media (compare growth in top two panels of Fig. 4). In comparison, 0.6 mM H 2 O 2 inhibited growth of IRP1-transformed yeast more severely on media lacking glutamate (Fig.  4). This stronger effect of H 2 O 2 on aconitase-dependent growth suggests that c-aconitase itself was more sensitive to H 2 O 2 than other essential cellular functions. The addition of 0.6 mM H 2 O 2 had a stronger inhibitory effect on aconitase-dependent growth of IRP1 S138D -and IRP1 S138E -transformed aco1 yeast than on yeast expressing wild-type IRP1 or IRP1 S138A (Fig. 4). For example, IRP1 S138D -transformed yeast grew as well on glutamate-free medium in absence of H 2 O 2 as yeast expressing wild-type IRP1 but grew significantly less well on this medium when H 2 O 2 was added (Fig. 4, compare bottom panels). IRP1 S138E -transformed cells grew poorly on glutamate-free medium, but their growth was undetectable when H 2 O 2 was added to the medium (Fig. 4). Curiously, 0.6 mM H 2 O 2 also inhibited aconitase-independent growth of yeast expressing IRP1 S138D and IRPI S138E more severely than yeast expressing wild-type IRP1 (Fig. 4). The reason for this effect is not yet clear. One possibility is that IRP1 itself has a protective effect against oxidative stress by contributing to production of NADPH in combination with cytosolic isocitrate dehydrogenase (29,43). NADPH is a key molecule for protection against oxidative stress (44 -46). Alternatively, the release of iron from c-aconitase could enhance H 2 O 2 toxicity (47). Nonetheless, the greater inhibitory effect of H 2 O 2 on aconitase-dependent growth of aco1 yeast expressing phosphomimetic IRP1 mutants reflects the greater sensitivity of these mutant c-aconitases to oxidant-mediated Fe-S cluster disassembly.
Effect of H 2 O 2 on Aconitase Activity-Since aco1 yeast trans-formed with wild-type IRP1 showed aconitase-dependent growth in the presence of 0.6 mM H 2 O 2 , it was of interest to determine the status of aconitase activity in these cells. IRP1transformed aco1 yeast were incubated in liquid media supplemented with 0.6 mM H 2 O 2 , and cells were collected at various times after oxidant addition and analyzed for aconitase activity and by EPR. At 1 h after H 2 O 2 addition, aconitase activity had decreased by ϳ80%, and the 3Fe EPR signal had increased by 2.7-fold (Fig. 5a). Thus, 0.6 mM H 2 O 2 caused strong inhibition of aconitase activity concomitant with an increase in the amount of 3Fe c-aconitase present in the cell, although the magnitude of change in aconitase activity and 3Fe EPR signal was less than that seen after exposure of yeast to 6 mM H 2 O 2 . By 3 h after the addition of 0.6 mM H 2 O 2 , aconitase activity had rebounded to nearly 50% of the original untreated level (Fig. 5a). Interestingly, the 3Fe c-aconitase EPR signal did not decrease at 3 h post-H 2 O 2 addition as aconitase activity increased (Fig. 5a). Yeast transformed with wild-type IRP1 were grown in SD-urahis medium to midexponential phase, at which time 0.6 mM H 2 O 2 was added. a, at the indicated times after the addition of H 2 O 2 , cells contained in an aliquot of the culture were harvested and analyzed by EPR and for aconitase activity. The intensity of the 3Fe signal, given as peak height, was normalized to cell mass. b, IRP1-transformed yeast were treated with 0.6 mM H 2 O 2 for 1 h, at which time cells were removed from this medium by centrifugation, washed once with ice-cold sterile deionized water, and placed into fresh medium without (f, Ⅺ) or with 100 M bipyridyl (q, E). At the indicated times, aliquots were removed and analyzed by EPR (open symbols) and for aconitase activity (closed symbols). The zero time point represents an aliquot taken at the end of the 1-h incubation with H 2 O 2 without the subsequent addition of fresh media.
Several explanations can be given for the persistence of the 3Fe EPR signal at a time when cells were recovering aconitase activity. Since H 2 O 2 remained in the medium throughout the experiment shown in Fig. 5a, it is possible that its continued presence caused a balance between formation and removal of 3Fe c-aconitase. Second, conversion of 3Fe c-aconitase to apo-IRP1 or back to 4Fe c-aconitase could be a slow process in yeast, particularly in the presence of H 2 O 2 . Third, 3Fe c-aconitase could be a "dead end" form in yeast. To investigate these various possibilities, IRP1-transformed aco1 yeast were treated with H 2 O 2 , and then the oxidant was washed out, and cells were analyzed for aconitase activity and by EPR after various periods in fresh media. Aconitase activity showed a steady increase upon removal of H 2 O 2 , reaching ϳ70% of the pretreated level by 3 h after oxidant removal (Fig. 5b). Inclusion of cycloheximide in the fresh media did not prevent recovery of aconitase activity, indicating that new synthesis was not required for recovery of aconitase activity in H 2 O 2 -treated yeast (not shown). Bipyridyl, a cell-permeant ferrous iron chelator, did inhibit recovery of aconitase activity, suggesting that chelatable iron was important to this process (Fig. 5b). The 3Fe c-aconitase signal showed a reciprocal decrease after the removal of H 2 O 2 , although this appeared to be delayed somewhat relative to the gain in aconitase activity (Fig. 5b). The 3Fe EPR signal remained high for at least 15 min after H 2 O 2 removal but had decreased by ϳ40% by 3 h after oxidant removal. The inclusion of bipyridyl did not alter the rate or extent to which the 3Fe EPR signal decreased after oxidant removal (Fig. 5b). Therefore, although iron availability appeared to be important for recovery of aconitase activity after exposure of yeast to H 2 O 2 , the reciprocal decrease in 3Fe c-aconitase was less dependent on chelatable iron. These results suggest that 3Fe c-aconitase is converted to apo-IRP1 and/or 3Fe c-aconitase is converted back to 4Fe c-aconitase using iron that is not readily chelated by bipyridyl.
Effect of H 2 O 2 on IRE Binding Activity in IRP1-transformed aco1 Yeast-We were interested to determine whether the loss of c-aconitase function upon exposure of IRP1-transformed yeast to H 2 O 2 was accompanied by an increase in IRE binding activity. IRP1-transformed yeast were treated with H 2 O 2 , and extracts were prepared and subjected to IRE binding analysis as indicated under "Experimental Procedures." IRE binding was not altered by treatment of IRP1-transformed yeast with H 2 O 2 (Table II). The amount of IRP1 with c-aconitase activity was 2-4-fold higher than the amount with IRE binding activity in log phase yeast expressing wild-type IRP1 or the S138A IRP1 mutant. 2 Therefore, if c-aconitase had been converted to the IRE binding form by H 2 O 2 , an increase in IRE binding activity should have been seen. The lack of an effect on IRE binding activity is consistent with the interpretation from the EPR analysis that H 2 O 2 treatment of yeast expressing wildtype IRP1 or IRP1 S138A converted the [4Fe-4S] cluster of caconitase to a [3Fe-4S] cluster and that conversion of the 3Fe form to apo-IRP1 was slow. The fact that H 2 O 2 did not cause an increase in IRE binding activity in yeast expressing IRP1 S138D and IRP1 S138E was more surprising (Table II). Disassembly of the Fe-S cluster in these c-aconitase mutants was more complete upon treatment of yeast with H 2 O 2 . One interpretation of these results is that conversion of c-aconitase to an IRE-binding protein after cluster disassembly requires additional steps that occur slowly in yeast. It is also possible that this process required additional factors that are absent in yeast.
Effect of Media Iron Status on Aconitase-dependent Growth of IRP1-transformed aco1 Yeast-The results presented above are consistent with continuous assembly and disassembly of an Fe-S cluster in IRP1 in yeast, thus establishing a level of c-aconitase in cells based on the ratio of assembly/disassembly. Since oxidant-initiated Fe-S cluster disassembly in the S138D and S138E c-aconitase mutants is more rapid (28), it seemed likely that aconitase activity in aco1 yeast transformed with IRP1 S138D or IRP1 S138E would have a heightened response to changes in cellular iron status. Alteration in cellular iron should affect Fe-S cluster assembly in IRP1. While we were unable to measure aconitase activity directly for these IRP1 mutants when isolated from yeast, growth of aco1 yeast expressing IRP1 is a measure of the relative aconitase activity in the cell (28,29). Growth of IRP1-transformed aco1 yeast was followed after inoculation into iron-rich media (supplemented with 1 mM FeSO 4 plus 1 mM ascorbic acid) or iron-poor media (supplemented with 100 M BPS). Excess iron did not alter the growth of any of the yeast in glutamate-supplemented media, indicating that iron was not limiting for aconitase-independent growth of yeast in minimal medium (not shown). The addition of excess iron to glutamate-free media also did not alter growth of aco1 yeast transformed with wild-type IRP1, IRP1 S138A , or IRP1 S138D (Fig. 6). However, aco1 yeast transformed with these IRP1s grew in glutamate-free media nearly as well as they grew in glutamate-supplemented media ( Fig. 4 and Ref. 28). Failure to observe stimulation of growth suggests that iron was not limiting for aconitase-dependent growth of these yeast as well. In contrast, growth of IRP1 S138E -transformed cells, which are severely growth-limited on glutamate-free media, was significantly stimulated by the addition of iron to glutamate-free media (Fig. 6). Its exponential growth rate was increased by 30% relative to control conditions, indicating that aconitase-dependent growth of IRP1 S138E -transformed cells was iron-limited. These results are consistent with the notion that excess iron promoted Fe-S cluster assembly and/or repair in IRP1 S138E , increasing the proportion of this IRP1 that was active as c-aconitase.
Depletion of available iron in the growth medium had an inhibitory effect on aconitase-dependent growth of all of the IRP1-transformed aco1 yeast. The exponential growth rate of yeast transformed with wild-type IRP1 or IRP1 S138A was reduced by ϳ20% in medium supplemented with BPS compared with control medium (Fig. 6). Growth of IRP1 S138A -transformed aco1 yeast was affected sooner after the BPS addition than wild-type IRP1-transformed cells (Fig. 6). BPS did not inhibit growth of yeast in glutamate-supplemented media (not shown). (Incubation of cells in media containing 100 M BPS for longer than 15 h resulted in more severe inhibition of growth as cellular iron stores were depleted (29).) Reduction in available iron had an even stronger effect on growth of IRP1 S138D -and IRP1 S138E -transformed aco1 yeast (Fig. 6). The exponential growth rate of these yeasts in glutamate-free media was inhibited by 40 and 70%, respectively, by iron depletion. BPS is an impermeant ferrous iron chelator, which inhibits iron uptake. Cells grown in media containing BPS are therefore likely to have a reduced rate of Fe-S cluster assembly and/or repair because of lower iron availability. The observation that aco1 yeast transformed with the phosphomimetic IRP1 mutants were more strongly growth-inhibited by the addition of BPS to glutamate-free media than was growth of wild-type IRP1transformed yeast is consistent with Fe-S cluster disassembly being more rapid in the phosphomimetic c-aconitase mutants in aerobically growing yeast.
Since the realization that IRP1 is the cytoplasmic isoform of aconitase (20,21,48), a great deal of effort has gone into defining the mechanism by which IRP1 interconverts with c-aconitase. Active c-aconitase contains a [4Fe-4S] cluster and can be inactivated by loss of only a single iron atom (20,21,31). However, conversion to IRP1 and activation of IRE binding requires complete disassembly of the Fe-S cluster (2). Exposure of c-aconitase to oxidants such as ferricyanide or oxygen results in its inactivation (2,28). In vitro studies indicate that oxidantmediated inactivation results from Fe-S cluster oxidation and disassembly. Cluster disassembly mediated by several oxidants occurs in stages, with the solvent-exposed iron (Fe a ) dissociating first, followed by the complete removal of the [3Fe-4S] cluster with prolonged exposure to oxidants or at higher concentrations of oxidant (20,21). Subsequent activation of IRE binding requires reduction of critical cysteine thiols in oxidized apo-IRP1 (39). To date, it is not clear how cluster assembly/disassembly occurs in vivo, which of the various intermediate states between c-aconitase and IRP1 observed in vitro are a part of the natural process in vivo, or how IRE binding activity is regenerated. Here, we employed a yeast model system to gain insight into the Fe-S cluster disassembly process in c-aconitase in vivo. Using EPR analysis of intact yeast cells, we show for the first time that the [3Fe-4S] form of c-aconitase exists in vivo. The fact that 3Fe c-aconitase was detected in cells under standard growth conditions suggests that it is an intermediate along the path for normal disassembly of the [4Fe-4S] cluster of c-aconitase in yeast. It is also possible that the 3Fe c-aconitase detected by EPR of untreated IRP1-transformed yeast could be an intermediate in cluster assembly. We think that this is unlikely, since we were unable to detect a 3Fe signal with untreated cells expressing IRP1 S138D or IRP1 S138E . If the 3Fe signal detected in IRP1-transformed aco1 yeast was a Fe-S cluster assembly intermediate, we would expect to detect this signal in cells expressing these IRP1 mutants, since both are converted to c-aconitase in aco1 yeast, as shown by their ability to support aconitase-dependent growth. Moreover, a weak 3Fe signal was detected in cells expressing IRP1 S138D only after exposure to low concentrations of H 2 O 2 . The fact that exposure of cells to H 2 O 2 increased the EPR signal for 3Fe c-aconitase and inhibited aconitase activity is strong evidence that cluster disassembly initiated by this oxidant goes through a 3Fe intermediate in vivo. In either event, detection of 3Fe c-aconitase is consistent with assembly and disassembly of a Fe-S cluster in IRP1 being a dynamic process in aerobically growing yeast, even in the absence of significant changes in iron status.
Gardner et al. (49,50) proposed that aconitases continuously undergo a 4Fe/3Fe/4Fe cycle in vivo. In Gardner's scheme, metabolically generated superoxide radical mediates Fe-S cluster oxidation and conversion of [4Fe-4S] aconitase to [3Fe-4S] aconitase. This 3Fe aconitase would undergo repair and reactivation by the insertion of Fe(II) from the labile iron pool, giving the 4Fe/3Fe/4Fe cycle. Interconversion of IRP1 and caconitase requires that the protein cycle between the apo-and 4Fe forms, however. At this time, it is not clear to what extent the 3Fe form of wild-type c-aconitase undergoes further cluster disassembly and conversion to apo-IRP1 versus repair of the [3Fe-4S] cluster to a [4Fe-4S] cluster in yeast. The 3Fe EPR signal decreased only after the removal of H 2 O 2 . The addition of bipyridyl, a cell-permeant ferrous iron chelator, did not prevent the decrease in the 3Fe signal upon H 2 O 2 removal, although it did inhibit recovery of aconitase activity. This is consistent with cluster disassembly in wild-type c-aconitase proceeding beyond the 3Fe stage in the absence of available iron for cluster repair. However, it is possible that reduction in the labile iron pool caused by bipyridyl preferentially affects de novo Fe-S cluster assembly and not cluster repair. Aconitase containing a [3Fe-4S] cluster has been observed to acquire a fourth iron atom in vitro in presence of chelators such as EDTA and to even scavenge iron from other 3Fe aconitase molecules (31). Whether this occurs in vivo and explains our results is not yet known.
Exposure of animal cells to H 2 O 2 causes a rapid increase in IRE binding activity concomitant with a decrease in c-aconitase activity. This is consistent with H 2 O 2 -mediated conversion of c-aconitase to IRP1 as a result of complete cluster disassembly and subsequent reduction of cysteine thiols in oxidized IRP1 FIG. 6. Effect of media iron status on aconitase-dependent growth of IRP1-transformed aco1 yeast. Yeasts were inoculated into glutamate-free media containing no additions (control; Ⅺ), 1 mM FeSO 4 plus 1 mM ascorbate (ϩFe 2ϩ ; छ), or 100 M BPS (ϩBPS; E). Cell growth was monitored by measuring optical density at 600 nm over the indicated time. The particular IRP1 being expressed is indicated in the upper left corner of each panel. wt, wild-type IRP1; S138E, IRP1 S138E ; S138A, IRP1 S138A ; S138D, IRP1 S138D . (41,42). We found that inhibition of c-aconitase activity and generation of 3Fe c-aconitase occurred very rapidly in aco1 yeast exposed to H 2 O 2 . However, a reciprocal increase in IRE binding was not observed. Wild-type c-aconitase and the S138A c-aconitase mutant accumulated as [3Fe-4S] c-aconitase after exposure of cells to H 2 O 2 . The failure to see an increase in IRE binding activity in yeast expressing wild-type IRP1 or IRP1 S138A was therefore not surprising, since 3Fe c-aconitase cannot bind IRE with high affinity (2). Brazzolotto et al. (38) reported that exposure of purified c-aconitase to H 2 O 2 in vitro similarly generated [3Fe-4S] c-aconitase without conversion to apo-IRP1. H 2 O 2 appeared to cause more complete Fe-S cluster disassembly in the S138D and S138E c-aconitase mutants but also did not cause a detectable increase in IRE binding activity under the conditions examined. When expressed in mammalian cells, these phosphomimetic IRP1 mutants preferentially accumulated in the RNA binding form. 3 It has been suggested that activation of IRE binding in response to H 2 O 2 in animal cells occurs through a signal transduction pathway (51-53). Thus, it is possible that such a pathway for H 2 O 2 -induced conversion of c-aconitase to IRP1 does not exist in yeast. Alternatively, steps required for activation of IRE binding after cluster disassembly, such as reduction of critical cysteine thiols (39), may occur slowly in IRP1 in H 2 O 2 -treated yeast.
We were unable to detect an EPR signal for [3Fe-4S] caconitase in cells expressing IRP1 S138E . The Fe-S cluster in this IRP1 mutant is very unstable; aconitase activity of the purified enzyme has a half-life of 4 min in the presence of oxygen (28). Mild oxidation of reconstituted S138E c-aconitase with ferricyanide in vitro inhibited aconitase activity without yielding a signal for a [3Fe-4S] cluster. Similar treatment of wild-type c-aconitase or the S138A mutant yielded a strong 3Fe signal. This suggests that even the 3Fe form of the S138E c-aconitase mutant is unstable and that oxidant-initiated cluster disassembly in this mutant rapidly progresses beyond the 3Fe stage. Alternatively, disassembly of the Fe-S cluster in S138E c-aconitase may bypass the 3Fe stage altogether. The Fe-S cluster of S138D c-aconitase was also unstable to ferricyanide oxidation in vitro, although a 3Fe EPR signal was detected. The intensity of this signal was much less than that of wild-type c-aconitase, suggesting that most of the cluster had been completely disassembled upon ferricyanide oxidation. Similarly, only a weak signal for a [3Fe-4S] cluster was ever detected in yeast expressing the IRP1 S138D , and only upon treatment with low concentrations of H 2 O 2 . These results argue strongly that oxidantinitiated Fe-S cluster disassembly in the phosphomimetic c-aconitase mutants rapidly progresses beyond the [3Fe-4S] stage, potentially enhancing their conversion to the RNA binding form. Consistent with this notion, a higher fraction of IRP1 S138E exists in the RNA binding form in aerobically grown aco1 yeast compared with the other Ser 138 IRP1 mutants, which are more stable as c-aconitase in these yeast. 2 Media iron availability also appeared to alter Fe-S cluster status in c-aconitase in yeast. Depletion of available iron in growth media by the addition of the cell-impermeant Fe(II) chelator BPS preferentially inhibited aconitase-dependent growth (see also Ref. 29). Inhibition was greatest for aco1 yeast expressing IRP1 S138D or IRP1 S138E , with the larger effect on the S138E mutant. The addition of iron to growth media preferentially stimulated aconitase-dependent growth of IRP1 S138E -transformed aco1 yeast. These results support a model of dynamic Fe-S cluster assembly/disassembly in IRP1 in yeast, with the proportion of IRP1 existing as c-aconitase being determined by the balance of assembly to disassembly.
The enhanced sensitivity of aconitase-dependent growth of aco1 yeast expressing IRP1 S138D or IRP1 S138E to media iron status suggests that the instability of the Fe-S cluster in caconitase derived from these IRP1 mutants under aerobic growth conditions increased their iron responsiveness. However, our results do not preclude the possibility that phosphomimetic mutations at serine 138 influenced 3Fe to 4Fe cycling as well. Substitution of glutamic acid or aspartic acid for serine residues is commonly employed to mimic phosphorylation. Therefore, these data point to phosphorylation of serine 138 as part of a mechanism that allows for rapid and complete Fe-S cluster disassembly in conversion of c-aconitase to IRP1.