The secondary alcohol metabolite of doxorubicin irreversibly inactivates aconitase/iron regulatory...

$: The secondary alcohol metabolite of doxorubicin irreversibly inactivates aconitase/iron regulatory protein-1 in cytosolic fractions from human myocardium$
5410892-6638/98/0012-0541/$02.25 Q FASEB

/ 3826 0003 Mp 541 Monday Mar 16 03:13 PM LP–FASEB 0003

The secondary alcohol metabolite of doxorubicinirreversibly inactivates aconitase/iron regulatoryprotein-1 in cytosolic fractions from humanmyocardium

GIORGIO MINOTTI,*,1 STEFANIA RECALCATI,† ALVARO MORDENTE,‡ GIOVANNILIBERI,§ ANTONIO MARIA CALAFIORE,§ CESARE MANCUSO,* PAOLO PREZIOSI,* ANDGAETANO CAIRO\

Departments of *Pharmacology and ‡Biochemistry, Catholic University School of Medicine, Rome;†Gastroenterology, University of Milan School of Medicine-IRCCS Ospedale Maggiore; \CNR Centerfor Cell Pathology, Milan; and §Cardiac Surgery, G. D’Annunzio University School of Medicine,Chieti, Italy

ABSTRACT Anticancer therapy with doxorubicin(DOX) is limited by severe cardiotoxicity, presum-ably reflecting the intramyocardial formation of drugmetabolites that alter cell constituents and functions.In a previous study, we showed that NADPH-supple-mented cytosolic fractions from human myocardialsamples can enzymatically reduce a carbonyl groupin the side chain of DOX, yielding a secondary alco-hol metabolite called doxorubicinol (DOXol). Herewe demonstrate that DOXol delocalizes low molecu-lar weight Fe(II) from the [4Fe-4S] cluster of cyto-plasmic aconitase. Iron delocalization proceedsthrough the reoxidation of DOXol to DOX and lib-erates DOX-Fe(II) complexes as ultimate by-prod-ucts. Under physiologic conditions, clusterdisassembly abolishes aconitase activity and forms anapoprotein that binds to mRNAs, coordinately in-creasing the synthesis of transferrin receptor but de-creasing that of ferritin. Aconitase is thus convertedinto an iron regulatory protein-1 (IRP-1) that causesiron uptake to prevail over sequestration, forming apool of free iron that is used for metabolic functions.Conversely, cluster reassembly converts IRP-1 backto aconitase, providing a regulatory mechanism to de-crease free iron when it exceeds metabolic require-ments. In contrast to these physiologic mechanisms,DOXol-dependent iron release and cluster disassem-bly not only abolish aconitase activity, but also affectirreversibly the ability of the apoprotein to functionas IRP-1 or to reincorporate iron within new Fe-S mo-tifs. This damage is mediated by DOX-Fe(II) com-plexes and reflects oxidative modifications of {SHresidues having the dual role to coordinate clusterassembly and facilitate interactions of IRP-1 withmRNAs. Collectively, these findings describe a novelmechanism of cardiotoxicity, suggesting that intra-myocardial formation of DOXol may perturb the ho-meostatic processes associated with cluster assembly

or disassembly and the reversible switch betweenaconitase and IRP-1. These results may also providea guideline to design new drugs that mitigate the car-diotoxicity of DOX.—Minotti, G., Recalcati, S., Mor-dente, A., Liberi, G., Calafiore, A. M., Mancuso, C.,Preziosi, P., Cairo, G. The secondary alcohol metab-olite of doxorubicin irreversibly inactivates aconi-tase/iron regulatory protein-1 in cytosolic fractionsfrom human myocardium. FASEB J. 12, 541–552(1998)

Key words: IRP-1 · cardiotoxicity · doxorubicin · alcohol me-tabolite · DOXol cluster disassembly · iron-responsive elements

DOXORUBICIN (DOX)2 is an anthracycline antibioticwith broad spectrum antitumor activity; however,DOX can also induce cardiotoxicity and congestiveheart failure, especially when the cumulative dose ex-ceeds Ç 550 mg/m2 (1, 2). According to the prevail-ing hypotheses, the cardiotoxicity of DOX ismediated by metabolites that are formed within car-diomyocytes and alter critical cell constituents andfunctions (3, 4). Ethical and practical constraints pre-clude the identification and quantification of thesemetabolites in fine needle myocardial biopsies fromDOX-treated patients. On the other hand, studieswith laboratory animals do not always anticipate theyield and reactivity of a given metabolite in humans,presumably because the enzymes of biotransforma-

1 Correspondence: Department of Pharmacology, CatholicUniversity School of Medicine, Largo F. Vito 1, 00168 Rome,Italy. E-mail: [email protected]

2 Abbreviations: DOX, doxorubicin: DOXol, doxorubicinol;IRP-1, iron regulatory protein-1; IREs, iron-responsive ele-ments; low mol wt, low molecular weight; FAS, ferrous am-monium sulfate; DTT, dithiothreitol; 2-ME, 2-mercapto-ethanol; O2·0, superoxide anion; H2O2, hydrogen peroxide;SOD, superoxide dismutase.

542 Vol. 12 May 1998 The FASEB Journal MINOTTI ET AL.


Scheme I. Structures of DOX and DOXol.

tion and the ultimate target (or targets) of toxicitymay exhibit species-related differences (5). Reconsti-tution of DOX with myocardial fragments disposedduring aorto-coronary bypass grafting provides a con-venient and ethically acceptable strategy to obviatethese problems and characterize the metabolism ofthis drug in the human heart. Using this approach,we have previously demonstrated that cytosolic frac-tions from human myocardium possess NADPH-de-pendent aldo-keto reductases that add two electronsto the C-13 carbonyl group in the side chain of DOX,forming a secondary alcohol metabolite called doxo-rubicinol (DOXol) (cf ref 6 and see Scheme I). Thesestudies have also shown that DOXol releases low mo-lecular weight (low mol wt) Fe(II) from a protein (orproteins) other than ferritin, which per se would ac-count for a major site of iron storage in the cytosolicmilieu of human cardiomyocytes (6, 7). We havetherefore hypothesized that DOXol might becomecardiotoxic by removing iron from alternativesources such as iron-requiring enzymes (6).

This study was aimed at establishing whetherDOXol removes iron from the [4Fe-4S] cluster of cy-toplasmic aconitase, a protein displaying marked ho-mology with the mitochondrial enzyme that convertscitrate to isocitrate via the intermediate cis-aconitatein the Krebs cycle (8). The fourth ferrous iron of thecluster, referred to as Fea, is directly involved in en-zymatic catalysis; limited removal of Fea would there-fore abolish aconitase activity (8). Perhaps moreimportant, dynamic processes of cluster disassemblyand reassembly enable cytoplasmic aconitase to par-ticipate in iron metabolism and cell homeostasis.Cluster disassembly occurs when the process of ironremoval extends beyond Fea and involves the threeother centers, referred to as Feb1–3. This process lib-erates an apoprotein that binds to iron-responsive el-ements (IREs) in mRNAs for transferrin receptor orferritin, increasing stability or decreasing translation,respectively (9–11). In so doing, the apoprotein func-tions as an iron regulatory protein-1 (IRP-1) that co-ordinately up-regulates the synthesis of transferrinreceptor but down-regulates that of ferritin, causingiron uptake to prevail over sequestration and forminga pool of low mol wt Fe that is available for metabolicuse. Under physiologic conditions, the balance be-tween aconitase and IRP-1 is influenced by cell ironstatus. When iron is scarce, both cluster disassemblyand de novo synthesis of the apoprotein shift the bal-ance in favor of IRP-1, promoting the formation oflow mol wt Fe (11). When iron is plentiful, new clus-ters are formed and the balance returns in favor ofaconitase, decreasing low mol wt Fe. Reversible as-sembly or disassembly of [4Fe-4S] clusters can there-fore be viewed as homeostatic mechanisms thatconstantly adapt the concentration of low mol wt Feto the metabolic needs of the cell. Perturbing thesemechanisms with DOXol or any other agent that in-

appropriately induces cluster iron delocalization anddisassembly would predictably affect cell functionsthat require iron as a cofactor.

Our results demonstrate that DOXol delocalizesiron from [4Fe-4S] clusters and causes irreversiblemodifications of both aconitase and IRP-1 activitiesin human myocardium. These findings may help clar-ify how DOX becomes cardiotoxic, setting the stagefor mechanism-oriented interventions that protectcardiomyocytes during anticancer therapy with thisdrug.

MATERIALS AND METHODS

Materials

Lactose-free anthracyclines were obtained through the cour-tesy of Dr. Antonino Suarato (Pharmacia-Upjohn, Milan, It-aly). Ammonium sulfate (ultrapure grade, FeIIIõ0.5 ppm) waspurchased from Schwarz/Mann (Cleveland, Ohio). Cysteine,EDTA, and ferrous ammonium sulfate (FAS) were from Merck(Darmstadt, Germany). d,l-Fluorocitrate (barium salt) was aproduct of Pfaltz and Bauer and was obtained through thecourtesy of Dr. Steven D. Aust (Biotechnology Center, UtahState University, Logan). cis-Aconitate, ferrozine, Sepharose6B, bovine erythrocyte CuZn superoxide dismutase (SOD; EC1.15.1.1), thymol-free bovine liver catalase (EC 1.11.1.6), and

DOXORUBICIN, IRON, AND CARDIOTOXICITY 543


all other chemicals were obtained from Sigma Chemical Co.(St. Louis, Mo.). Unless otherwise indicated, the experimentswere carried out in 0.3 M NaCl, carefully adjusted to pH 7.0just before use. This was done to avoid ligand-catalyzed inter-actions of most common buffers with iron (12). Although un-buffered, the pH of the reaction mixtures did not varythroughout the period of the experiment. All solutions wereprepared with double-distilled water that had been passedthrough a Milli-Q Water System (Millipore; Marlborough,Mass.). Trace metals were eventually removed by ion-exchange chromatography on Chelex 100 (Bio-Rad; Rich-mond, Calif.).

Preparation of cytosols from myocardial biopsies

Small atrial samples (Ç0.1–0.2 g) were taken from 76 patients(45 M, 31 F, aged 61{4 years) undergoing aortocoronary by-pass grafting. All biopsies were collected before cardiopul-monary bypass, using a previously described technique (6).This protocol had been approved by Institutional ReviewBoards, and informed consent was obtained from all patients.After storage at 0807C, pools of 15–20 myocardial biopsieswere processed for cytosol preparation by sequential homog-enization, ultracentrifugation, and overnight stirring of105,000 g supernatants with 65% ammonium sulfate, as de-scribed previously (6). Calcium hydroxyapatite chromatogra-phy (6) was omitted in order to minimize cluster oxidationand other possible modifications of aconitase/IRP-1. After 20min precipitation at 10,000 g, cytosolic proteins were sus-pended in a minimum volume of 0.3 M NaCl and dialyzedfirst against two 1-liter changes of 0.3 M NaCl–1 mM EDTAand then against two 1-liter changes of 0.3 M NaCl (to removeEDTA and EDTA–iron complexes). Previous studies haveshown that EDTA would remove adventitious iron (6, 7) butnot iron ions coordinated within the aconitase cluster (13).Cytosols prepared by this standard procedure were referredto as unmodified. Where indicated, cytosols (3–5 mg protein/ml) were incubated with cysteine and FAS at a final ratio of1100 or 50 nmol/mg protein, respectively. This treatment wasintended to reconstitute the [4Fe-4S] cluster of aconitase, aprocess requiring iron and thiols (13, 14). After 15 min at 47C(13), unreacted FAS and cysteine were removed by gel filtra-tion on a (1.5112 cm) Sepharose 6B column equilibrated andeluted with 0.3 M NaCl, pH 7.0; protein-containing fractionswere then incubated with 65% ammonium sulfate, which alsocontained 0.25 mM cis-aconitate, to protect the cluster againstoxidative inactivation (13). After 4 h stirring, 10,000 g precip-itates were dissolved in a minimum volume of 0.3 M NaCl anddialyzed sequentially against 1-liter changes of 0.3 M NaCl-0.1mM cis-aconitate, EDTA-NaCl, and NaCl. These cytosols werereferred to as iron-loaded. In other experiments, cytosols weredialyzed against two 1-liter changes of 100 mM Tris HCl, pH8.9/40 mM KCl, diluted to 3 mg protein/ml with the samebuffer, and incubated with 100 mM dithiothreitol (DTT) toremove iron from clusters (15). After 15 min at room tem-perature, these cytosols were subjected to Sepharose 6B chro-matography, ammonium sulfate precipitation, and dialysis asdescribed for the preparation of iron-loaded samples, with theexception that cis-aconitate was omitted throughout. Cytosolstreated with DTT/pH 8.9 were referred to as iron-depleted.

In vitro RNA transcription and RNA–protein gel retardationassay for IRP-1

The pSPT-fer plasmid containing the IRE of human ferritinH chain (16) was linearized with BAM HI and transcribed invitro with T7 RNA polymerase in the presence of 100 mCi of[a32-P]UTP (800 mCi/mmol; Amersham Co., Arlington

Heights, Ill.). For IRP-1 assay, 2 mg protein samples were in-cubated with a molar excess of IRE probe, digested with RnaseT1, and treated with heparin as described previously (17). Af-ter separation on 6% nondenaturing polyacrylamide gels, IRP-1/IRE complexes were visualized by autoradiography andquantified by computer-assisted densitometry (Imaging Den-sitometer, G5670; Bio-Rad). Binding specificity has beendemonstrated by competition studies (17). In selected exper-iments, iron-loaded cytosols (1.5–3 mg protein/ml) were as-sayed for IRP-1 after 1 h incubation at 377C with cis-aconitateplus or minus DOX(ol) (100 and 4 nmol/mg protein, respec-tively). Preliminary experiments showed that DOX(ol) per sedid not interfere with IRP-1 detection when added to cytosolsjust prior to the gel shift assay at a ratio of 0.5 to 5 nmol/mgprotein; by contrast, ¢ 50 nmol cis-aconitate/mg protein de-creased the formation and/or detection of IRE/IRP-1 com-plexes, in keeping with previous reports demonstrating similareffects of aconitase substrates (18). cis-Aconitate-treated cyto-sols were therefore subjected to Sepharose 6B chromatogra-phy, ammonium sulfate precipitation, and dialysis to ensuresubstrate removal. Where indicated, cytosols were treated with2-mercaptoethanol (2-ME) just before the gel-shift assay to re-duce disulfide bridges and regenerate {SH groups that me-diate RNA binding (19, 20). Based on titration experiments,the concentration of 2-ME was optimized to 0.6% (v:v).

Assay for aconitase

Aconitase activity was determined spectrophotometrically bymonitoring the disappearance of cis-aconitate (e240Å3.6 mM01

cm01) (21). The incubations (1 ml final volume) containedcytosols (80–300 mg protein) and 0.1 mM cis-aconitate in 0.3M NaCl, pH 7.0, 377C; one mU was defined as the amount ofenzyme that consumed 1 nmol cis-aconitate/min (21). Whereindicated, aconitase was measured in cytosols that had beentreated with cis-aconitate plus or minus DOX(ol) and subse-quently made free of substrate as described for IRP-1 assay.

Assays for aldo-ketoreductase activity and DOX metabolites

The anthracycline aldo-ketoreductase activity of human myo-cardium was measured as DOXol formation in 0.5 ml incu-bations containing cytosols (150 mg protein), NADPH, andDOX (both 150 mM) in 0.3 M NaCl, pH 7.0, 377C. After 4 h,the reaction mixtures were extracted and analyzed by 2-di-mensional thin layer chromatography (6). Mobile phases,standard preparations, metabolite assignement, and fluori-metric quantification have been described in detail (6, 22).Control experiments showed that variable amounts of DOXolhydrolyzed nonenzymatically and formed a polar metabolite(e.g., 4-demethyl-deoxy aglycones). Aldo-keto reductase activ-ities were therefore expressed as nmol [DOXol]/mg protein/4 h, where [DOXol] indicates the metabolite per se and sec-ondary products of nonenzymatic hydrolysis.

Assays for iron delocalization

Doxorubicinol-dependent delocalization of low mol wt Fe(II)was studied spectrophotometrically in 1 ml incubations con-taining cytosols (1 mg protein) in 0.3 M NaCl, pH 7.0, 377C.Reactions were started by adding DOXol prepared by NaBH4

reduction of DOX and purified by 2-dimensional thin layerchromatography (6, 22). Delocalized Fe(II) was detected bychelation with 0.25 mM ferrozine, which per se would notremove iron from clusters (13). The formation of ferrozine-Fe(II) complexes [e564 nmÅ27.9 mM01 cm01 (23)] was moni-tored against reference samples lacking the chromophore by



Figure 1. Aconitase and IRP-1 activities in unmodified, iron-loaded, and iron-depleted cytosols. Cytosols were preparedand assayed for aconitase and IRP-1 as described in Materialsand Methods. IRP-1/IRE complexes were quantified densito-metrically and expressed as percent vs. unmodified cytosols.Values given as means { SE of three to five determinations induplicate. *, **P õ 0.05 or P õ 0.01 vs. unmodified cytosols,respectively. The bottom panel shows two representative gelshift assays of unmodified vs. iron-loaded or iron-depleted cy-tosols.

using a Hewlett Packard 8453 diode array spectrophotometerwith computer-assisted corrections for scatter and turbidity.

Other assays

Proteins were determined by the bicinchoninic acid acidmethod (24). The cytosolic content of Fea was determined in1 ml incubations containing cytosols (1–3 mg protein) andammonium persulfate (1 mM), which served to induce slowcluster disassembly (13). Delocalized Fea was detected with fer-rozine (0.25 mM), and readings were corrected for referencesamples that contained 0.25 mM cis-aconitate to protect [4Fe-4S] clusters from oxidative decay (13). The difference be-tween [ferrozine-Fe(II) (0cis-aconitate)] and [ferrozine-Fe(II) (/cis-aconitate)] gave a net measurement ofcluster-delocalized Fe(II). Reactions were allowed to proceeduntil ammonium persulfate decreased aconitase activity toÇ10% of initial values, indicative of near to complete removalof Fea (13).

Unless otherwise indicated, all data are expressed as thearithmetic mean { SE. Statistical analyses were performed bypaired and unpaired Student’s t tests, and differences wereconsidered significant when P õ 0.05. Other conditions areindicated in the legends to the figures and tables.

RESULTS

Aconitase and IRP-1 in human myocardium cytosols

Both aconitase and IRP-1 were detected in unmodifiedcytosols from human myocardial biopsies, showingthat these samples contained [4Fe-4S] holoproteinswith enzymatic function as well as cluster-free apo-proteins with RNA binding activity (Fig. 1). Treat-ment with cysteine and FAS yielded iron-loadedcytosols bearing more [4Fe-4S] clusters; therefore,aconitase increased and IRP-1 decreased. By contrast,treatment with DTT/pH 8.9 yielded iron-depleted cy-tosol bearing fewer [4Fe-4S] clusters; hence, aconi-tase decreased whereas IRP-1 increased (see also Fig.1). These findings showed that iron incorporation orremoval and cluster assembly or disassembly madeaconitase and IRP-1 activities mutually exclusive.

In addition to IRP-1, cells may contain variableamounts of an iron regulatory protein-2, which bindsto IREs but lacks the ability to form the [4Fe-4S] clus-ter required for aconitase activity (25). Currentlyavailable techniques do not permit gel shift separa-tion and detection of IRP-2 vs. IRP-1 in human tis-sues. Nonetheless, the observation that iron loadingor unloading procedures were accompanied by co-ordinate but divergent changes of aconitase activityvs. IRE binding suggests that IRP-1 accounted for amajor part of the biochemical features of our sam-ples.

Doxorubicinol-dependent cluster irondelocalization: requirement for cis-aconitate

NADPH-supplemented cytosols were found to reduceDOX to DOXol, in keeping with the presence of

aldo-keto reductases in human myocardium (6). Spe-cific activities averaged 3.4 { 0.4, 3.5 { 0.7, and 3.9{ 0.4 nmol DOXol/mg protein/4 h in unmod-ified, iron-loaded, and iron-depleted cytosols, re-spectively (nÅ3; Pú0.05). Therefore, subsequent ex-periments to assess the metabolic fate of DOXol andits ability to release iron from aconitase were per-formed by reconstituting the purified metabolite withcytosols at a final ratio of 4 nmol/mg protein, repro-ducing the aldo-keto reductase activity of these sam-ples. These experiments were performed in thepresence or absence of cis-aconitate as a substrate in-termediate of aconitase. This was done because sub-strates are known to modulate the response ofcytoplasmic aconitase to redox-active agents that tar-get its [4Fe-4S] cluster (20, 26). cis-Aconitate is mostsuitable for these studies (26) and readily isomerizesto both citrate and isocitrate within [4Fe-4S] clusters(8), reproducing physiologic equilibration of thesethree isomers during the aconitase reaction. Asshown in Fig. 2A, incubation of DOXol with iron-loaded cytosols accompanied the recovery of DOX,indicative of a two-equivalent reoxidation of the sec-ondary alcohol moiety at C-13. Formation of DOXwas enhanced by cis-aconitate in a concentration-de-pendent manner, suggesting that DOXol oxidationwas mediated by reactions with aconitase. In the ex-periments with 100 mM cis-aconitate, Ç2 nmol



Figure 2. cis-Aconitate- and DOXol-dependent Fe(II) mobili-zation and DOX formation in different cytosol samples. In-cubations (1 ml final volume) contained cytosols (1 mgprotein), purified DOXol (4 M), and increasing concentra-tions of cis-aconitate in 0.3 M NaCl, pH 7.0, 377C. A) Incuba-tions were extracted and assayed for DOX by 2-dimensionalthin layer chromatography. B) The mobilization of Fe(II) wasmonitored by chelation with 0.25 mM ferrozine. Values arethose determined at 1 h and are means { SE of three to fiveexperiments.

Figure 3. Effects of fluorocitrate on aconitase activity and DOXformation or Fe(II) mobilization. Iron-loaded cytosols wereassayed for aconitase as described in Materials and Methods,using 100 mM cis-aconitate plus or minus 10–200 mM fluoroci-trate. Cytosols were also incubated with DOXol and assayedfor DOX, 4-demethyl-deoxy DOXol aglycone, and low mol wtFe(II) as described in Materials and Methods and the legendto Fig. 2, with the exception that cis-aconitate was held con-stant at 100 mM and added in conjunction with 10–200 mMfluorocitrate. Values are taken from a representative experi-ment and are expressed as percent to permit direct compari-sons. In the absence of fluorocitrate, aconitase activity was 2.4mU/mg protein, whereas 4-demethyl deoxy-DOXol aglycone,DOX, and low mol wt Fe(II) were 1.6, 1.8, and 3.4 nmol/mgprotein, respectively. DOXol ag, 4-demethyl-deoxy DOXolaglycone.

DOXol converted to DOX; the remaining was recov-ered as 4-demethyl-deoxy DOXol aglycone, reflectingnonenzymatic hydrolysis and removal of the amino-sugar bound to the anthracycline (cf Scheme I).However, the formation of this aglycone did not in-volve reactions with aconitase inasmuch as the rela-tive yield was the same in the absence as in thepresence of 100 mM cis-aconitate (1.8{0.2 vs. 1.9{0.3nmol/mg protein; nÅ3).

Having determined that DOXol oxidized with acon-itase, we performed experiments to evaluate whetherit also delocalized iron from the [4Fe-4S] cluster ofthis enzyme. As shown in Fig. 2B, reconstitution ofDOXol with iron-loaded cytosols did accompany themobilization of low mol wt Fe(II). This process wasenhanced by cis-aconitate in a concentration-depen-dent manner, similar to the oxidation of DOXol toDOX, pointing to the release of iron ions associatedwith aconitase. It is important that the release of Fe(II)and the formation of DOX were both abolished byreplacing iron-loaded cytosols with unmodified oriron-depleted samples having less or no [4Fe-4S] clus-

ters, irrespective of the presence or absence ofincreasing concentrations of cis-aconitate (Fig. 2A–B).Collectively, these experiments suggested that DOXand low mol wt Fe(II) were formed upon reaction ofDOXol with the [4Fe-4S] cluster of aconitase. More-over, the requirement for cis-aconitate gave prelimi-nary evidence that DOXol–aconitase interactionswere favored by the presence of physiologic sub-strates/intermediates within [4Fe-4S] clusters. To testthis hypothesis, iron-loaded cytosols were coincubatedwith cis-aconitate and increasing concentrations offluorocitrate, which competes for the cluster and in-hibits aconitase (27). As shown in Fig. 3, fluorocitrateconcentration-dependently suppressed aconitase ac-tivity and similarly decreased DOX formation andFe(II) mobilization when cytosols were exposed toDOXol. These experiments gave more direct evidencethat: 1) cytoplasmic aconitase was the source ofDOXol-releasable iron; and 2) both DOXol oxidationand iron delocalization occurred when aconitase wasinvolved in catalysis with physiologic substrate inter-mediates. Figure 3 also shows that fluorocitrate did notmodify the yield of 4-demethyl-deoxy DOXol agly-cone, confirming that this metabolite was formed byaconitase-independent hydrolysis of DOXol.

Mechanisms and stoichiometry of cluster irondelocalization

cis-Aconitate and DOXol-dependent iron release wasfurther characterized from mechanistic and stoichi-



TABLE 1. cis-Aconitate- and/or DOX(ol)-dependent mobilizationof low mol wt Fe(II) in iron-loaded cytosolsa

Addition nmol Fe(II)/mg protein

cis-Aconitate NDDOXol 0.3cis-Aconitate / DOXol 3.9cis-Aconitate / DOX 0.4

a Incubations were prepared and assayed for Fe(II) mobilizationas described in the legend to Fig. 2B. cis-Aconitate and DOX(ol) con-centrations were 100 and 4 mM, respectively. Values are taken from arepresentative experiment.

Figure 4. Effects of DOXol on aconitase activity. Iron-loadedcytosols were assayed for aconitase before and after 1 h incu-bation with cis-aconitate and/or DOX(ol), as described in Ma-terials and Methods. Where indicated, the incubationscontained SOD (100 U/ml) and/or catalase (400 U/ml). Val-ues are means{ SE of three determinations. cis-acon, cis-acon-itate; CAT, catalase.

ometric viewpoints. A series of control experimentsshowed that neither cis-aconitate nor DOXol per secould release iron; a combination of cis-aconitate withDOX was similarly ineffective (Table 1). Thus, ironrelease was mediated by a concerted action of cis-aconitate with the secondary alcohol moiety ofDOXol. In a subsequent set of experiments, we char-acterized whether the simultaneous formation ofDOX and low mol wt Fe(II) reflected a redox cou-pling of DOXol with oxygen, yielding superoxide an-ion (O2·0) and hydrogen peroxide (H2O2) as theultimate mediators of iron delocalization. Duplicatedeterminations with 80–92% experimental agree-ment showed that cis-aconitate and DOXol-depen-dent release of 3.9 nmol Fe(II) was not appreciablyaffected by SOD (100 U/ml), catalase (400 U/ml) ora combination of the two enzymes, yielding 3.6, 3.8,and 4.1 nmol Fe(II)/mg protein, respectively. Redoxcoupling of DOXol with oxygen was therefore irrel-evant to the mechanisms of iron release, which in-stead reflected direct reactions of DOXol withclusters. Not surprisingly, the data in Fig. 2A–B hadshown that replacing iron-loaded cytosols with un-modified or iron-depleted samples was sufficient toabolish the formation of DOX, implying that DOXoloxidized with cluster iron rather than with oxygen.

From a stoichiometric viewpoint, we thought it wasimportant to establish whether iron delocalizationwas limited to Fea or extended to Feb1–3. For this pur-pose, cytosols were assayed for Fea by limited clusterdisassembly with ammonium persulfate (13). Usingthis method, we could calculate that unmodified cy-tosols contained 0.4 { 0.1 nmol Fea/mg protein,which increased to 1.1 { 0.3 in iron-loaded cytosolsand decreased down to õ0.2 in iron-depleted sam-ples (nÅ3). Changes in Fea were therefore consistentwith the procedures that had been used to promotecluster assembly or disassembly. With particular ref-erence to iron-loaded samples, a value of Ç1 nmolFea/mg protein was opposed to the ability ofcis-aconitate and DOXol to release up to Ç4 nmolFe(II) (cf Fig. 2B). This comparison strongly sug-gested that cis-aconitate and DOXol-dependent irondelocalization exceeded limited removal of Fea andinvolved Feb1–3, setting the premises for cluster dis-

assembly and conversion of aconitase to the cor-responding apoprotein/IRP-1. After this char-acterization, it also became evident that the oxidationof two DOXol was sufficient to mobilize the four ironcenters of aconitase, as attested by the ability of 100mM cis-aconitate to couple the formation ofÇ2 nmolDOX with the mobilization ofÇ4 nmol Fe(II) (cf Fig.2A–B). This 1:2 stoichiometry was in keeping with anearlier suggestion that the two reducing equivalentsof a secondary alcohol moiety can be returned to twoindependent iron ions (6).

Effects of DOXol on aconitase and IRP-1 activities

Having demonstrated that cis-aconitate coupledDOXol oxidation with cluster iron delocalization, weperformed experiments to evaluate functional con-sequences on the enzymatic or RNA binding activitiesof this protein. As shown in Fig. 4A, treatment of iron-loaded cytosols with DOXol and cis-aconitate almost



Figure 5. Recovery of aconitase activity after cluster iron re-moval with DTT/pH 8.9 or cis-aconitate plus DOXol. Iron-loaded cytosols were assayed for aconitase activity before andafter cluster iron removal by DTT/pH 8.9 (A) or cis-aconitateplus DOXol (B), as described in Materials and Methods. Afterthese treatments, cytosols were exposed to cysteine and FASand assayed for recovery of aconitase activity. Values are means{ SE of three determinations. cys, cysteine; FAS, ferrous am-monium sulfate; cis-acon, cis-aconitate.

Figure 6. Effects of cis-aconitate and/or DOX(ol) onIRP-1 activity. A) IRP-1 was assayed in unmodifiedcytosols (lanes a, d), iron-loaded cytosols (lanes b,e), and cis-aconitate/DOXol-treated, iron-loaded cy-tosols (lanes c, f). Where indicated, samples weremixed with 0.6 % 2-ME just prior to the assay. cis-Aconitate/DOXol treatments were performed as de-scribed in Materials and Methods and the gel isrepresentative of three similar experiments. B) Iron-loaded cytosols (1 mg protein/ml) were incubatedwith 0.5–5 mM purified DOXol in 0.3 M NaCl, pH7.0, 377C. After 1 h, aliquots were taken and assayedfor IRP-1. C) Unmodified cytosols (lane a) weremade iron loaded with cysteine/FAS (lane b) andsubsequently treated with cis-aconitate (lane c) or cis-aconitate plus DOX (lane d). 2-ME, 2-mercapto-ethanol.

abolished aconitase activity, consistent with the re-moval of cluster iron that occurred under these con-ditions. Doxorubicinol and cis-aconitate per se didnot decrease aconitase, nor was the enzyme activity

affected by cis-aconitate plus DOX (see also Fig. 4A).This confirmed that cluster iron delocalization andconsequent loss of enzyme activity reflected a con-certed action of cis-aconitate with the secondary al-cohol moiety of DOXol. On the other hand,cis-aconitate and DOXol could abolish aconitase ac-tivity irrespective of the presence or absence of SODand/or catalase, as one would expect if O2·0 and H2O2

were not involved in cluster disassembly (Fig. 4B).After these characterizations, cytosols were as-

sessed for their ability to undergo reversible disassem-bly and reassembly of [4Fe-4S] clusters. For thispurpose, iron loaded cytosols were treated withDTT/pH 8.9 or cis-aconitate plus DOXol to removecluster iron and abolish aconitase activity; both sam-ples were then reconstituted with cysteine and FASto assist the formation of new Fe-S motifs. As shownin Fig. 5, cytosols that had been treated with DTT/pH 8.9 responded to FAS and cysteine by formingnew [4Fe-4S] clusters, as evidenced by excellent re-covery of aconitase activity. In contrast, cytosolstreated with cis-aconitate and DOXol failed to recoverenzymatic function. These comparative experimentssuggested that cis-aconitate- and DOXol-dependentiron removal was accompanied by apoprotein modi-fications that precluded iron reincorporation andcluster reassembly.

Under the same conditions as described in Fig. 1,iron-loaded cytosols consistently exhibited less IRP-1activity than did unmodified cytosols, in agreementwith the inverse relationship between cluster iron sat-uration and RNA binding (Fig. 6, lanes a, b). How-ever, subsequent treatment of iron-loaded cytosolswith DOXol and cis-aconitate, while releasing ironfrom the cluster, did not increase but actually sup-pressed the formation of IRE/IRP-1 complexes (Fig.6A, lane c). To better characterize these findings, cy-tosols were treated with 2-ME just prior to the gel shiftassay in order to reduce disulfide bridges and regen-erate {SH groups that facilitate RNA binding (19,20). As also shown in Fig. 6A (lanes d, e), 2-MEslightly increased IRP-1 activity in both unmodified



Figure 7. Mechanisms of cis-aconitate- and DOXol-dependentIRP-1 inactivation. A) Iron-loaded cytosols were assayed forIRP-1 after 1 h incubation with cis-aconitate plus DOXol, asdescribed in Materials and Methods. Where indicated, incu-bations were carried out in the presence of SOD (100 U/ml)and/or catalase (400 U/ml). B) Iron-depleted cytosols (1 mgprotein/ml) were incubated with DOX (2 mM) and/or FeSO4

(4 mM), plus or minus SOD and catalase, in 0.3 M NaCl, pH7.0, 377C. After 1 h, aliquots were taken and assayed for IRP-1. Values are taken from a representative experiment and areexpressed as percent to permit comparisons. cis-acon, cis-acon-itate; CAT, catalase.

or iron-loaded cytosols; therefore, IRP-1 remainedÇ50% lower in iron-loaded cytosols. 2-Mercapto-ethanol had a more visible effect on iron-loaded cy-tosols that had been treated with cis-aconitate andDOXol; nonetheless, 2-ME could not restore IRP-1activity to the same levels as observed in cytosols thathad not been treated with these compounds (Fig. 6A,lane f). Collectively, these results suggested that: 1)the different RNA binding activity of unmodified andiron-loaded cytosols reflected the degree of clusteriron saturation more than the availability of {SHgroups; and 2) treatment of iron-loaded cytosols withcis-aconitate and DOXol converted sulphydryl groupsto disulfides, but also to more complex species thatcould not be reduced by 2-ME, causing the irrevers-ible inactivation of IRP-1. Two lines of evidence con-firmed that IRP-1 was inactivated by a concertedaction of cis-aconitate and the alcohol moiety ofDOXol. First, increasing concentrations of DOXolper se had no effect on RNA binding (Fig. 6B). Sec-ond, neither cis-aconitate nor cis-aconitate plus DOXcould modify IRP-1 activity (Fig. 6C).

Experiments were performed to establish how cis-aconitate and DOXol failed to increase but actuallysuppressed IRP-1 activity, although they effectively re-moved iron from the cluster of aconitase. In a firstset, neither SOD nor catalase or a combination of thetwo enzymes was found to prevent the decline of IRP-1 in iron-loaded cytosols exposed to cis-aconitate andDOXol (Fig. 7A). This showed that O2·0 and H2O2

were not involved as causative agents of IRP-1 dys-function, in keeping with previous findings that theywere similarly unable to delocalize cluster iron. In asecond set, we attempted to establish whether IRP-1was affected by the two specific products of cis-acon-itate- and DOXol-dependent cluster disassembly, i.e.,Ç2 nmol DOX and Ç4 nmol low mol wt Fe(II) (cfFig. 2A–B). For this purpose, we prepared iron-de-pleted cytosols having high levels of IRP-1; these sam-ples were then exposed to DOX, Fe(II) or a mixtureof DOX with Fe(II). As reported in Fig. 7B, neitherDOX nor Fe(II) affected IRP-1 activity; however, asignificant decrease was observed when DOX andFe(II) were both included. A series of parallel deter-minations showed that a mixture of 2 nmol DOX with4 nmol Fe(II) changed the spectral characteristics ofthe drug, causing a decrease in absorbance at 477 nmand an increase at 600 nm (not shown). These spec-tral changes reflected the formation of DOX-Fe(II)complexes that converted to DOX-Fe(III) by oxidiz-ing with oxygen (28). Comparative analyses againstknown amounts of DOX-Fe(III) also allowed us tocalculate that ligand interactions of 4 nmol Fe(II)with 2 nmol DOX approached completion inÇ5 minand yielded 0.9{ 0.1 nmol Fe(III) (nÅ3). Within thesame experimental period, unchelated Fe(II) did notappreciably oxidize with oxygen, as judged by a com-plete recovery after chelation with ferrozine

(3.9{0.05 nmol; nÅ3). Taking these findings into ac-count, we concluded that IRP-1 was targeted by DOX-Fe(II) complexes in which iron had become redoxunstable. Superoxide dismutase and catalase werenonetheless unable to prevent the inactivation ofIRP-1 by DOX-Fe(II), showing that drug–iron com-plexes acted by redox intermediates other than O2·0and H2O2 (see also Fig. 7B).

DISCUSSION

Iron is an important cofactor of several enzymes andcellular functions, varying from respiration to DNA



synthesis (11). To maintain homeostasis, cells haveevolved cytoplasmic aconitase/IRP-1, a bifunctionalprotein that modulates iron uptake vs. sequestrationand regulates the concentration of metabolicallyavailable low mol wt Fe. In this study, we have shownthat an anticancer drug like DOX may become car-diotoxic through the action of DOXol, a secondaryalcohol metabolite that irreversibly inactivates acon-itase/IRP1. Inasmuch as these processes cannot bestudied in cardiac biopsies from cancer patients, wehave used cytosolic fractions from myocardial sam-ples obtained during surgery. Moreover, the experi-ments have been performed with concentrations ofDOXol that reproduce those formed by the aldo-ketoreductases of human myocardium. Results obtainedunder these conditions may therefore be relevant topathologic conditions in vivo.

The aconitase reaction relies on [4Fe-4S] clustersin which Fea and Feb1 have distinct ferrous propertiesand Feb2–3 tends to stabilize as Fe(II)-Fe(III) dimers(29). Previous evidence for a possible reaction ofDOXol with iron was obtained in cytosols that hadbeen chromatographed on calcium hydroxyapatiteor subjected to iron loading in the absence of reduc-ing agents (6). Both of these manipulations may havecaused cluster iron oxidation, forming protein–ironcomplexes that contained predominantly Fe(III) (6).Under those conditions, DOXol was found to reduceFe(III) and mobilize Fe(II) by virtue of reactionmechanisms that did not require additional cofactors(6). In the present study, hydroxyapatite chromatog-raphy has been omitted and iron loading has beenperformed in the presence of cysteine (see Materialsand Methods). These modifications have allowed usto stabilize some iron in a ferrous form attributableto the Fea of properly assembled [4Fe-4S] clusters.Under these newly defined conditions, DOXol inac-tivates cytoplasmic aconitase by delocalizing both Fea

and Feb1–3 in a low mol wt Fe(II) form. Iron mobili-zation proceeds through the reoxidation of DOXolto DOX and is accompanied by irreversible modifi-cations that preclude cluster reassembly and recoveryof aconitase activity (cf Fig. 4 and Fig. 5).

Although cis-aconitate is often shown to protect Fe-S motifs from chemical agents (13, 26), our studiesdemonstrate that it represents an absolute require-ment and cofactor for DOXol oxidation and iron re-lease to occur (cf Fig. 2A–B and Table 1). Thisfinding is pathophysiologically relevant, as [4Fe-4S]clusters would almost certainly contain some sub-strate in vivo (20). To achieve some insight into therole of cis-aconitate, one must remember that theaconitase reaction accompanies sterical and redoxtransients, reflecting different modes of Fea ligationby citrate or isocitrate and reversible intensificationof the ferrous character of Fea as opposed to the un-changed Fe(II)/Fe(III) behavior of Feb1–3 (29). cis-Aconitate is an obligatory intermediate of these

processes, inasmuch as it isomerizes to both citrateand isocitrate and can bind to Fea either mode byrotating 7180 (29). Sterical and redox transients at ornear Fea might thus enable DOXol to introduce re-ducing equivalents into the cluster, explaining howfour iron atoms with relatively different characteris-tics can eventually be delocalized in a ferrozine-che-latable form having net Fe(II) properties. In keepingwith this possible mechanism, the effects of cis-acon-itate on DOXol oxidation and Fe(II) release decreaseremarkably when aconitase is exposed to fluoroci-trate, forming by-products that bind tightly to Fea anddisplace the physiologic substrate intermediatesotherwise involved in sterical and redox transients (cfFig. 3 and ref 27). Under the same experimental con-ditions, fluorocitrate would not affect drug metabo-lites that are formed by aconitase-independentmechanisms, e.g., 4-demethyl-deoxy DOXol aglycone(cf Fig. 3).

Experiments with unmodified vs. iron-loaded oriron-depleted cytosols indicate that cluster assemblyor disassembly generates an inverse relationship be-tween aconitase and IRP-1 in human myocardium (cfFig. 1). This regulatory mechanism is lost when cis-aconitate and DOXol remove cluster iron, formingan apoprotein that lacks aconitase activity but alsofails to function as IRP-1 (cf Fig. 6A). Under theseparticular conditions, the physiologic switch of acon-itase to IRP-1 is precluded by oxidative modificationsof cysteine residues—most notably Cys437—that me-diate interactions of the apoprotein with IREs (19,20, 26). In fact, cis-aconitate and DOXol-treated cy-tosols can visibly recover IRP-1 activity after exposureto 2-ME, a reducing agent that converts disulfides to{SH groups (see also Fig. 6A). Treatment with 2-ME is nonetheless insufficient to achieve a completereactivation of IRP-1, suggesting that some oxidativemodifications have proceeded beyond the formationof disulfides, yielding sulfenic, sulfinic, or sulfonicmoieties that cannot be reverted to {SH by 2-ME(30). A series of reconstitution experiments indicatesthat IRP-1 is inactivated by the two specific productsof DOXol-induced cluster disassembly, i.e., DOX andFe(II), forming redox-active DOX–Fe(II) complexes(cf section on ‘‘Effects of DOXol on aconitase andIRP-1 activities’’ and Fig. 7B). Previous studies byother investigators have similarly suggested a possiblerole for DOX-iron complexes in protein thiol oxida-tion (31) and cardiotoxicity (32–34). Our presentfindings extend those reports, showing that cytoplas-mic aconitase/IRP-1 is both a source of DOXol-re-leasable iron and a target of DOX–iron complexesthat react in situ with critical residues of the apopro-tein. Cys437 can also contribute along with Cys503 andCys506 to the process of cluster iron coordination,conferring aconitase activity on the apoprotein (19,20). The dual role of cysteines in RNA binding andcluster assembly would adequately explain how cis-



aconitate and DOXol treatments preclude iron rein-corporation and recovery of aconitase activity (cfFig. 5).

The observation that DOXol can irreversibly inac-tivate IRP-1 anticipates several pathologic conse-quences, the most obvious being a disturbance in theregulation of transferrin iron uptake vs. ferritin ironsequestration. Cardiomyocytes would thus becomeunable to sense iron levels and coordinate iron move-ments between the different sites of deposition ormetabolic use. In addition to these changes, theremay be other and more subtle consequences ofDOXol-cluster interactions. For example, the simul-taneous loss of aconitase activity should increase thecytosolic levels of citrate, altering the putative role ofthis compound as a chelator and transporter of in-tracellular iron (35). Recent studies have also shownthat IRP-1 can bind to IREs in mRNA for mitochon-drial aconitase, decreasing the synthesis of this en-zyme (36, 37). The biologic significance of thisregulatory connection is unknown, but these findingsreinforce the idea that a reversible switch betweenaconitase and IRP-1 modulates several homeostaticprocesses. Interrupting this switch with DOXolshould predictably expose cardiomyocytes to a mul-tifactorial damage, especially if one appreciates howsensitive these cells can be to anomalous changes iniron and energy metabolism (38). De novo synthesisof IRP-1 would probably fail to compensate for thesedisorders, inasmuch as the apoprotein is itself a targetof DOX-iron complexes. Similar reasonings mightperhaps be applied to IRP-2, which is also redox-re-gulatable at critical {SH residues (25, 26).

The mechanisms described above must be con-trasted with other prevailing hypotheses of cardiotox-icity, mostly involving a quinone group placed in thetetracycline ring of DOX (cf Scheme I). Several stud-ies have shown that this moiety is liable to one-elec-tron addition by NAD(P)H oxidoreductases ofintracellular organelles, yielding a semiquinone freeradical that readily regenerates the parent quinoneby reducing oxygen to O2·0 and H2O2 (3, 4). Duringthis redox cycling, O2·0 can reductively release someiron from ferritin (39, 40). Inasmuch as cardiomyo-cytes are relatively deficient in oxyradical-detoxifyingenzymes such as SOD, catalase, or glutathioneperoxidase, one-electron redox cycling of DOX is of-ten said to facilitate intracardiac accumulation andreaction of O2·0 and H2O2 with delocalized Fe(II),forming hydroxyl radicals or iron–oxygen complexesthat eventually promote lipid peroxidation (3, 4). At-tempts to validate this biochemical hypothesis in hu-man subjects have produced conflicting results. Wehave recently demonstrated that DOX infusions donot increase but actually suppress cardiac lipid per-oxidation in cancer patients, as evidenced by adecrease of lipid-conjugated dienes and hydro-peroxides in blood samples collected from the coro-

nary sinus (7). These findings cast doubt on the im-portance of lipid peroxidation and explain othernegative reports showing that antioxidants like a-to-copherol or N-acetylcysteine would not protect can-cer patients from DOX toxicity (41, 42). A significantdegree of protection has nonetheless been observedwith dexrazoxane, a bis-keto-piperazinedione that hy-drolyzes intracellularly and liberates a diacid diamidethat chelates iron (1, 2, 33, 34). Keeping this in mind,one cannot escape the conclusion that dexrazoxanemitigates cardiotoxic events other than iron-depen-dent lipid damage. The experiments described in ourpresent study may help reconcile these inconsisten-cies. For example, dexrazoxane might reequilibrateiron between ferritin, enzymes, or putative transport-ers like citrate, thereby relieving the disturbancesotherwise associated with the inactivation of aconi-tase/IRP-1 and consequent changes in iron uptakevs. sequestration or metabolic use. Alternatively, de-xrazoxane might compete with DOX for iron andprevent formation of those drug–iron complexesthat interrupt the physiologic interconversion be-tween aconitase and IRP-1. In either case, dexrazox-ane would prevent homeostatic disorders rather thanlipid peroxidation or equivalent free radical reac-tions, making it possible to rationalize the protectiveeffects of chelation therapy as opposed to unsuccess-ful interventions with antioxidants.

A role for O2·0 and H2O2 might be maintained ifthese species mediate the deleterious effects ofDOXol on aconitase/IRP-1. Our studies indicate thatthis is not the case. In fact, neither SOD nor catalaseor a combination of the two enzymes can preventDOXol-dependent cluster disassembly, in keepingwith the earlier suggestion that DOXol interacts withiron by direct electron transfer mechanisms (see sec-tion on ‘‘Mechanisms and stoichiometry of clusteriron delocalization’’ and ref 6). Likewise, SOD andcatalase cannot prevent the bifunctional inactivationof IRP-1 and aconitase, perhaps because the oxida-tion of DOX-Fe(II) with oxygen generates Fe(II)/Fe(III) complexes that modify biologic targets byO2·0 and H2O2-independent mechanisms (cf refs 28,39 and Fig. 4B, Fig. 7B). Critical appraisal of the lit-erature also suggests that O2·0 and H2O2 have mul-tiple and often diverging effects on Fe-S clusters andaconitase/IRP-1; however, neither species would im-pose the same bifunctional and irreversible inactiva-tion as imposed by DOXol (43–48).

In conclusion, our studies demonstrate that the in-teraction of DOXol with cytoplasmic aconitase/IRP-1 encircles various aspects of DOX toxicity andprovides a framework to optimize cardioprotectionwith dexrazoxane or other newly designed chelators.The unusual reactivity of DOXol with aconitase/IRP-1 vis a vis its marginal contribution to the tumoricidaleffects of DOX (49) also implies there may be addi-tional means to decrease cardiotoxicity without af-



fecting anticancer activity. Potential strategiesinclude the development of drugs that inhibit aldo-keto reductases or the identification of DOX analogsthat form fewer alcohol metabolites.

This work was supported by grants from MURST (to G.M.,Special Project ex 40% ‘‘New assessment approaches in toxi-cology,’’ 1996) and from Consiglio Nazionale delle Ricerche.

REFERENCES

1. Weiss, R. B. (1992) The anthracyclines: will we ever find a betterdoxorubicin? Sem. Oncol. 19, 670–686

2. Dorr, R. T. (1996) Cytoprotective agents for anthracyclines. Sem.Oncol. 23 (Suppl. 8), 23–34

3. Olson, R. D., and Mushlin, P. S. (1990) Doxorubicin cardiotox-icity: analysis of prevailing hypotheses. FASEB J. 4, 3076–3086

4. Powis, G. (1989) Free radical formation by antitumor quinones.Free Rad. Biol. Med. 6, 63–101

5. Nies, A.S., and Spielberg, S.P. (1996) Principles of therapeutics.In Goodman and Gilman’s The Pharmacological Basis of Therapeutics,9th Ed (Hardman, J. G., Limbird, L. E., Molinoff, P. B., and Rud-don, R. W., eds) pp. 43–62, McGraw-Hill Co., New York

6. Minotti, G., Cavaliere, A. F., Mordente, A., Rossi, M., Schiavello,R., Zamparelli, R., and Possati, G. F. (1995) Secondary alcoholmetabolites mediate iron delocalization in cytosolic fractions ofmyocardial biopsies exposed to anticancer anthracyclines. J. Clin.Invest. 95, 1595–1605

7. Minotti, G., Mancuso, C. Frustaci, A., Mordente, A., Santini,S. A., Calafiore, A. M., Liberi, G., and Gentiloni, N. (1996) Par-adoxical inhibition of cardiac lipid peroxidation in cancer pa-tients treated with doxorubicin. J. Clin. Invest. 98, 650–661

8. Beinert, H., and Kennedy, M. C. (1993) Aconitase, a two-facedprotein: enzyme and iron regulatory factor. FASEB J. 7, 1442–1449

9. Klausner, R. D., Rouault, T. A., and Harford, J. B. (1993) Regu-lating the fate of mRNA: the control of cellular iron. Cell 72, 19–28

10. Harrison, P. M., and Arosio, P. (1996) The ferritins: molecularproperties, iron storage function and cellular regulation.Biochim. Biophys. Acta 1275, 161–203

11. Hentze, M. W., and Kuhn, L. C. (1996) Molecular control ofvertebrate iron metabolism: mRNA-based regulatory circuits op-erated by iron, nitric oxide, and oxidative stress. Proc. Natl. Acad.Sci. USA 93, 8175–8182

12. Miller, D. M., Buettner, G. R., and Aust, S. D. (1990) Transitionmetals as catalysts of ’autoxidation’ reactions. Free Rad. Biol. Med.8, 95–108

13. Kennedy, M. C., Emptage, M. H., Dreyer, J.-L., and Beinert, H.(1979) The role of iron in the activation-inactivation of aconi-tase. J. Biol. Chem. 258, 11098–11105

14. Constable, A., Quick, S., Gray, N. K., and Hentze, M. W. (1992)Modulation of the RNA-binding activity of a regulatory proteinby iron in vitro: switching between enzymatic and genetic func-tion? Proc. Natl. Acad. Sci. USA 89, 4554–4558

15. Haile, D. J., Rouault, T. A., Tang, C. K., Chin, J., Harford J., andKlausner, R. D. (1992) Reciprocal control of RNA-binding andaconitase activity in the regulation of the iron-responsive ele-ment binding protein: Role of the iron-sulfur cluster. Proc. Natl.Acad. Sci. USA 89, 7536–7540

16. Mullner, E. W., Neupert, B., and Kuhn, L. C. (1989) A specificmRNA binding factor regulates the iron-dependent stability ofcytoplasmic transferrin receptor mRNA. Cell 58, 373–382

17. Cairo, G., and Pietrangelo, A. (1994) Transferrin receptor geneexpression during rat liver regeneration. Evidence for post-tran-scriptional regulation by iron regulatory factor B, a second iron-responsive element binding protein. J. Biol. Chem. 269, 6405–6409

18. Haile, D. J., Rouault, T. A., Harford, J. B., Kennedy, M. C.,Blondin, G. A., Beinert, H., and Klausner, R. D. (1992) Cellularregulation of the iron-responsive element binding protein: dis-

assembly of the cubane iron-sulfur cluster results in high-affin-ity RNA binding. Proc. Natl. Acad. Sci. USA 89, 11735–11739

19. Philpott, C. C., Haile, D., Rouault, T. A., and Klausner, R. D.(1993) Modification of free Fe-S cluster cysteine residue in theactive iron-responsive element-binding protein prevents RNAbinding. J. Biol. Chem. 268, 17655–17658

20. Hirling, H., Henderson, B. R., and Kuhn, L. C. (1994) Muta-tional analysis of the [4Fe-4S] cluster converting iron regulatoryfactor from its RNA-binding form to cytoplasmic aconitase.EMBO J. 13, 453–461

21. Drapier, J. C., and Hibbs, J. B., Jr. (1986) Murine cytotoxic ac-tivated macrophages inhibit aconitase in tumor cells. J. Clin. In-vest. 78, 790–797

22. Takanashi, S., and Bachur, N. R. (1976) Adriamycin metabolismin man. Evidence for urinary metabolism. Drug. Metab. Dispos.Biol. Fate Chem. 4, 79–87

23. Aust, S. D., Miller, D. M., and Samokyszin, V. M. (1990) Ironredox reactions in lipid peroxidation. Methods Enzymol. 186, 457–463

24. Stoscheck, C. M. (1990) Quantitation of protein. Methods En-zymol. 182, 50–68

25. Phillips, J. D., Guo, B., Yu, Y., Brown, F. M., and Leibold, E. A.(1996) Expression and biochemical characterization of iron reg-ulatory proteins 1 and 2 in Saccharomices cerevisiae. Biochemistry 35,15704–15714

26. Bouton, C., Hirling, H., and Drapier, J.-C. (1997) Redox mod-ulation of iron regulatory proteins by peroxynitrite. J. Biol. Chem.272, 19969–19975

27. Lauble, H., Kennedy, M. C., Emptage, M. H., Beinert, H., andStout, D. C. (1996) The reaction of fluorocitrate with aconitaseand the crystal structure of the enzyme-inhibitor complex. Proc.Natl. Acad. Sci. USA 93, 13699–13705

28. Minotti, G. (1990) NADPH and adriamycin dependent micro-somal release of iron and lipid peroxidation. Arch. Biochem. Bio-phys. 268, 398–403

29. Beinert, H., and Kennedy, M. C. (1989) Engineering of protein-bound iron sulfur clusters. Eur. J. Biochem. 189, 5–15

30. Torchinsky, Y. M. (1981) Sulfur in Proteins, pp. 48–56, PergamonPress, New York

31. Vile, G., and Winterbourn, C. C. (1990) Thiol oxidation andinhibition of Ca-ATPase by adriamycin in rabbit heart micro-somes. Biochem. Pharmacol. 39, 769–774

32. Gianni, L., Corden, B. J., and Myers, C. E. (1983) The bio-chemical basis of anthracycline toxicity and antitumor activ-ity. In Reviews in Biochemical Toxicology (Hodgson, E., Bend, J.,and Philpot, P. M., eds) pp. 1–82, Elsevier-North Holland,New York

33. Doroshow, J. H. (1991) Doxorubicin-induced cardiotoxicity.NewEngl. J. Med. 324, 843–845

34. Buss, J. L., and Hasinoff, B. B. (1993) The one-ring open hydro-lysis product intermediates of the cardioprotective agent ICRF-187 (dexrazoxane) displace iron from iron-anthracyclinescomplexes. Agents Actions 40, 86–95

35. Paraskeva, E., and Hentze, M. W. (1996) Iron-sulfur clusters asgenetic regulatory switches: The bifunctional iron regulatoryprotein-1. FEBS Lett. 389, 401–403

36. Gray, N. K., Pantopoulos, K., Dandekar, T., Ackrell, B. A. C., andHentze M. W. (1996) Translational regulation of mammalianand Drosophila citric acid cycle enzymes via iron responsive ele-ments. Proc. Nal. Acad. Sci. USA 93, 4925–4930

37. Kim, H.-Y., La Vaute, T., Iwai, K., Klausner, R. D., and Rouault,T. A. (1996) Identification of a conserved and functional iron-responsive element in the 5*-untranslated region of mammalianmitochondrial aconitase. J. Biol. Chem. 271, 24226–24230

38. Schoen, F. J. (1994) The heart. In Robbins Pathologic Basis of Dis-ease, 5th Ed (Cotran, R. S., Kumar, V., and Robbins S. L., eds)pp. 517–582, W. B. Saunders Co., Philadelphia

39. Minotti, G. (1993) Sources and role of iron in lipid peroxidation.Chem. Res. Toxicol. 6, 134–146

40. Ryan, T. P., and Aust, S. D. (1992) The role of iron in oxygen-mediated toxicities. Crit. Rev. Toxicol. 22, 119–141

41. Weitzman, S. A., Lorell, B., Carey R. W., Kaufman, S., andStossel, T. P. (1980) Prospective studies of tocopherol pro-phylaxis for anthracycline cardiac toxicity. Curr. Therap. Res.28, 682–686

42. Myers, C., Bonow, R., Palmieri, S., Jenkins, J., Corden, B.,Locker, G., Doroshow, J., and Epstein, S. A. (1983) A random-



ized controlled trial assessing the prevention of doxorubicincardiomyopathy by N-acetylcysteine. Sem. Oncol. 10 (Suppl.),53–55

43. Gardner, P. R., Ranieri, I., Epstein, L. B., and White C. R. (1995)Superoxide radical and iron modulate aconitase activity in mam-malian cells. J. Biol. Chem. 270, 13339–13405

44. Pantopoulos, K., Mueller, S., Atzberger, A., Ansorge, W., Strem-mel, W., and Hentze, M. W. (1997) Differences in the regulationof iron regulatory protein-1 (IRP-1) by extra- and intracellularoxidative stress. J. Biol. Chem. 272, 9802–9808

45. Rouault, T. A., and Klausner, R. D. (1996) Iron-sulfur clusters asbiosensors of oxidants and iron. Trends Biochem. Sci. 21, 174–177

46. Hentze, M. W. (1996) Iron-sulfur clusters and oxidant stress re-sponses. Trends Biochem. Sci. 21, 282–283

47. Pantopoulos, K., and Hentze, M. W. (1995) Rapid response tooxidative stress mediated by iron regulatory protein. EMBO J. 14,2917–2924

48. Cairo, G., Castrusini, E., Minotti, G., and Bernelli-Zazzera, A.(1996) Superoxide and hydrogen peroxide-dependent inhibi-tion of iron regulatory protein activity: a protective stratagemagainst oxidative injury. FASEB J. 10, 1326–1335

49. Kuffel, M. J., Reid, J. M., and Ames, M. M. (1992) Anthracyclinesand their C-13 alcohol metabolites: growth inhibition and DNAdamage following incubation with tumor cell lines. Cancer Che-mother. Pharmacol. 30, 51–57

Received for publication November 17, 1997.Accepted for publication December 22, 1997.

The secondary alcohol metabolite of doxorubicin irreversibly inactivates aconitase/iron regulatory...

Documents

Transcript of The secondary alcohol metabolite of doxorubicin irreversibly inactivates aconitase/iron regulatory...