Coordination of Cellular Iron Metabolism by Post-Transcriptional Gene Regulation Lukas C. Kiihn and Matthias W. Hentze LCK. Swiss Institute for Experimental Cancer MWH. European Molecular Biology Laboratory,

Research, Epalinges, Switzerland.Heidelberg, Germany

AB!XRACT Maintenance of cellular iron homeostasis demands the coordination of iron uptake, intracelhtlar storage, and utilixation. Recent investigations suggest that a single genetic regulatory system orchestrates the expression of proteins with central importance for all three aspects of cellular iron met&&m at the level of mRNA stability and translation. Two components of this regulatory system have been defined: a cis-acting mRNA sequence/sttucture motif called “iron-responsive element’* (IRE) and a specific trons-acting cytoplasmic binding protein, here referred to as “IRE-bindmg protein” @U&BP). As an early event in the regulatory cascade, cellular iron deprivation induces the IRE-bii activity of IRE-BP, whereas binding activity is reduced in iron-replete cells. IRE-BP is highly homologous to the iron-sulphur (Pe-S) protein aconitase which strongly suggests that IRE-BP is an Fe-S protein itself. Control over IRE-BP activity by the celhdar iron status is exerted post-translationally and likely involves changes between (4Fe4S) and (3Fe4S) states of the postulated IRE-BP Fe-S cluster. In addition, post-translational regulation of IRE-BP activity via heme has been proposed. Subsequent to its activation, IRE-BP binds with high affinity to single IREs contained in the 5’ untranslated regions (UTRs) of ferritin and erythroid 5-aminolevulhric acid synthase (eALAS) mRNAs. The binding represses translation of these proteins involved in iron storage and utilixation, respectively. In contrast, iron uptake is largely regulated via multiple IREs in the 3’ UTR of transkrrht receptor (TfR) mRNA. TtR-IREs are required for the iron-sensitive control of TW mRNA stability. IRE-BP binding stab&es TtR gene transcripts against as yet undefined ribonucleases. As a result of these regulatory hrteractions, iron starvation hxhrces the expression of TtR, thereby increasing iron uptake, and represses the synthesis of proteins involved in iron storage and utilixation. As cellular iron. levels rise, the homeosta tic balance is maintained by lowerhtg iron uptake and increasing iron storage in ferritin.

Address

reprintrequats to: hhs

des Boveresses, CH-1066 Epaliies,

C. Kiihn, Swiss Institute for Experhnental Cancer Research, Chemin Switxerland.

183 Journal of Inorganic BkhemWrp 47.183-195 (1992) 0 1992ElacvicrSciencePublishingCo.. Inc., 655 Avenue of the Amcrim, NY, NY 10010 01620134/92/$5.00

184 L. C. Kiihn and iU. W. Hen&e INTRODUCTION

Iron is a critical nutrient for living cells. In spite of its abundance in the environment, cellular iron’metabolism is complicated by two chemical characteristics of this metal. In an oxygenated atmosphere at physiological pH, iron occurs almost exclusively in its essentially water-insoluble ferric form. Iron can also exhibit lethal toxicity inside the cell [see Ref. 11. Consequently, cells have had to develop means to take up iron as a soluble chelate, to direct it to iron-dependent metabolic pathways and to “buffer” iron toxicity by intracelhrlar storage in a nontoxic form. In serum, iron is kept soluble by binding to the protein transferrin. A specific membrane protein, the transferrin receptor (TfR), mediates Tf endocytosis. Once inside the cell, iron dissociates from transferrin and leaves the endosome by an as yet unknown mechanism; it is then either incorpomted into haemoproteins, such as cytochromes, or enzymes such as ribonucleotide reductase, aconitase and other iron-sulphur proteins or is stored in a nontoxic form inside ferritin. The system of iron uptake and storage by the TfR and ferritin is highly conserved among different cell types and species. The importance of ferritin and the T!R in iron metabolism is reflected in their tight regulation in response to the cellular iron state. As predicted, high iron availability (which can be experimentally induced by addition of diferric transferrin, iron salts, or hemin to the medium of cultured cells) results in increased biosynthesis of ferritin and decreased production of T!R. In contrast, iron starvation (experimentally achieved by adding the iron chelator desferrioxamine to the medium) induces increased synthesis of TfR and decreases ferritin production [see Refs. 2, 31. This regulation is bidirectional and rapid; it has been observed in laboratory animals [see Ref. 41 and cultured cells [see Refs. 5, 61. Regulation of iron uptake and storage occurs mainly at the post-transcriptional level. We ,will describe the components of this regulatory system and consider how iron mediates their functional interaction. In addition to the “housekeeping” functions of iron uptake and storage, the major iron utilization pathway in the body, the biosynthesis of heme in erythroid cells, may also be controlled by this coordinated post-transcriptional regulatory network .[see Refs. 7, 81.

REGULATION OF IRON UPTAKE BY THE TRANWERRIN

RECEPTOR

Exchange of iron between tissues is predominantly mediated by iron-loaded serum transferrin and its,binding to specific cell surface receptors. Transferrin receptors are continuously internalized by endocytosis via coated vesicles and travel to the intracellular endosomal compartment .from where they recycle back to the surface [see Refs. 9-l 11. Presumably as a result of the slightly acidic pH of 5.5 in the endosome, ferric iron is released from transfertin and is then probably reduced. It reaches the cytoplasm by a poorly elucidated transmembrane transport. Apotransfertin, still bound to TfR, returns to the cell surface where it dissociates rapidly at the neutral PH. The net uptake of iron by TfR-mediated endocytosis is directly proportional to the level of TfR expression [see Refs. 12- 141. Several reasons contribute to this correlation: first, concentrations of diferric transferrin in serum am at least lOO-fold above the K, for the receptor-ligand interaction; second, binding of transferrin is rapid and likely saturates receptors continuously; and third, the

REGULATION OF IRON METABOLISM

,$85

multiple steps in the endocytic pathway are, as much as we know, barely regulated under physiological conditions. In order to acquire the correct amount of iron it is therefore crucial for cells to express an appropriate level of TfR at their surface. This is achieved through a regulatory feedback loop in which iron modulates TfR mRNA levels. Iron-dependent regulation of TfR was first observed in tissue culture cells upon incubation with hemin, iron-salts, or various iron-chelators [see Refs. Z-171. In each case, TfR was expressed at levels that were inversely proportional to iron availabiity. Transfection experiments showed that the regulation of TfR mRNA levels is largely not due to a promoter-dependent transcriptional control mechanism, but depends on specific sequences in the TfR mRNA 3’ untranslated region (3’ UTR) [see Refs. 18, 191. Joining these 3’ sequences behind the coding region of an unrelated indicator gene was sufficient to confer a similarly regulated phenotype to the transcribed hybrid mRNA [see Refs. 5, 201. Thus, the 3’ UTR of TfR mRNA contains the structural information which permits differential mRNA accumulation related to cellular iron supply. Based on run-on transcription assays and direct measurements of TfR mRNA in the nucleus and cytoplasm, it was concluded that the regulation of TW mRNA results in an altered cytoplasmic mRNA stability with up to 20-fold changes in its half-life [see Ref. 51. The &acting regulatory sites within the 3’ UTR have been mapped by detailed deletion and mutation analysis [see Refs. 5, 201 and are located in a region with high sequence conservation among different vertebrate species [see Refs. 21, 221. Relevant elements in human TfR mRNA encompass two areas of about 150 bases separated by 250 bases of dispensable sequences. At least two types of functional elements have been identified: five hairpin structures with properties of authentic IREs (Fig. 1) [see Ref. 20] and some additional adjacent stem loop structures that confer a high turn-over rate to TfR mRNA of cells grown in medium with high iron content [see Ref. 51. Deletion of the entire 3’ UTR or the “rapid turnover determinant” resulted in a constitutively high expression of TfR mRNA [5] whereas concurrent mutagenesis of all IREs prevented the binding of IRE-BP and yielded a nonregulated, rapidly decaying mRNA [see Ref. 231. Of the five IREs present, at least four bind simultaneously an IRE-BP in a 1: 1 stoichiometry [see Refs. 24, 251,

A A,C

4 4

N ,,--,

A,W

@.-.

FIGURE 1. Sequence and structure determinatits define the ‘konsensus” IRE. The depicted motif is derived from phylogenetic comparisons and experimentally tested variants. Deletion of encircled nucleotides causes loss of in vivo JRE function and/or in vitro IRE-BP binding. In contrast, variants identified by arrows were shown to cause only minor functional changes. For further discussion, see text.

186 L. C. Kiihn and M. W. Hen&e

but only three of them are stringently requimd for irondependent regulation [see Refs. 5, 231. These findings together with the observation that IRE-BP activity in vivo directly parallels TfR mRNA levels has lead to the concept that IRE-BP-bii protects TtR mRNA against degradation [see Refs. 23,241. The pathway of TtR mRNA degradation and the RNases involved remain to be characterixed. Apparently one of the components in this process is a short-lived polypeptide, since the protein synthesis inhibitor cycloheximide blocks TtR mRNA degradation. Recent experiments by Koeller et al. [see Ref. 261 argue against the possibility that cycloheximide prolongates the half-life of TtR mRNA by a direct c&-acting effect on translation. These authors found that modulating TW mRNA translation by insertion of an IRE into the 5’ UTR did not change the rate of its decay. Inhibition of transcription by actinomycin D has also been reported to prolongate the TfR mRNA half-life under conditions where this mRNA is expected to decay [see Ref. 241. An unexpected finding (Seiser et al;, IPIO Abs 041) concerns the decay of TtR mRNA during cell growth arrest of an interleukin-2dependent cytotoxic T cell clone. Upon removal of interleukin-2, binding of active IRE-BP to TfR mRNA was not su@icient to prevent the mRNA decay. This suggests that additional growth factor-dependent events may be involved in the protection of TtR mRNA. REGULATION OF IRON STORAGE

AND UTlLIZATION

Once iron has entered the cell, its storage and utilization must be regulated in a coordinate fashion. Z&ringer et al. [see Ref. 41 first suggested that the expression of the intracellular iron storage protein ferritin was controlled at the translational level, i.e., that the increase in ferritin biosynthesis after iron administration occurred without corresponding changes in ferritin mRNA levels. Subsequent studies confirmed these initial observations and identitied a single, highly conserved RNA element in the 5’ UTR of human and rat ferritin H- and L-chain mRNAs which is necessary and sufficient for the iron-dependent translational control [see Refs. 27-291. Changes in the intracellular iron status alter the affinity of a specific cytoplasmic IRE-binding protein (IRE-BP) (see below), such that under conditions of iron starvation increased binding to the ferritin IRE represses ferritin synthesis [see Ref. 30-331. Conversely, IRE-binding activity seems to be low in iron replete cells which permits efficient, unrepressed ferritin expression. The specific repression of ferritin mRNA translation has also been demonstrated in vitro, making use of endogenous IRE-BP activity contained in rabbit reticulocyte lysates or by adding purified IRE-BP to cell-free wheat germ translation systems [see Refs. 34, 351; for this property IRE-BP has also been termed “ferritin repressor protein” or “FRP.” Interestingly, it has been suggested that ferritin mRNA competes efficiently against other mRNAs for limited translation initiation factors in the absence of IRE-BP induced repression [see Ref. 361. The high translational efficiency of the ferritin transcript in its nonrepressed state makes it possible that IRE-BP-induced repression results in a large range of iron regulatory effects. Currently, we know relatively little about the mechanism by which the IRE/IRE-BP complex represses mRNA translation. It seems clear that translation initiation (as opposed to elongation or termination) is affected [see Ref. 371, but the molecular nature of this block in the initiation pathway has remained elusive. Recent in vitro

REGULATION OF IRON METABOLISM

187

translation experiments have demonstrated that IRE-BP media&d repression does not require polyadenylation of the mRNA [see Ref. 381. Furthermore, the formation of a high affinity RNA/protein complex of spliceosomal origin in the 5’ UTR of an indicator mRNA suffices to repress its translation in cis (Stripecke and Hentze, unpublished work). This result implies that no specific additional property of IRE-BP (other than iron-regulated high affinity binding to the IRE) would be required to control ferritin mRNA translation. Moreover, proximity of the IRE/IRE-BP complex to the 5’ cap structure of the mRNA is required for its ability to regulate translation in vivo [see Ref. 33, m and He&e, IPIO Abs P67]. This finding may explain why the position of the IRE in the 5’ UTRs of ferritin mRNAs is highly conserved even though tk total lengths of the 5’ UTR of ferritin mRNAs are not. Mechanistically, these observations suggest that an early step in the translation initiation pathway, possibly related to the initial interaction of the cap binding complex with the 5’ end of the mRNA, is the target for regulation. RNA motifs which closely resemble ferritin IREs have been identified in the 5’ UTRs of two vertebrate mRNAs, erythmid 5-aminolevulinate synthase (eALAS) and aconitase mRNA [see Refs. 8, 39, 401. Since eALAS is involved in the biochemical pathway which co~umes - 80% of the daily systemic iron turnover, this observation suggested involvement of the IRE/IRE-BP system in the regulation of iron utilization. Consistent with such a suggestion, the eALAS IRE motif is conserved between human and murine homologues and is located close to the cap structure of the eALAS mRNAs. It is spe&caIly recognized by IRE-BP in vitro [see Refs. 7, 81 and confers iron-dependent translational regulation to indicator genes in vivo [see Ref. 81. Recently, iron regulation of eALAS mRNA translation was demonstrated directly in mouse erythroleukemia cells (Hen&e et al., unpublished observations). Thus, the eALAS mRNA has been established as a second.translationally regulated target for the IRE/IRE-BP system. Since eALAS does not “consume” iron directly, but feeds 5-aminolevulinate into subsequent steps of the heme biosynthetic pathway (which utilizes iron in its final step), the significance of eALAS regulation for controlling heme synthesis and thus iron utilization requires more detailed investigation. Conceivably, the eALAS IRE could serve as a protective element under conditions of relative iron starvation to ensure iron availability for critical cellular enzymes at the expense of the heme biosynthetic pathway. The function and physiological relevance of the IRE motif bound in the 5’ UTR of porcine heart mitochondrial aconitase mRNA [see Ref. 81 is less clear. The IRE is located in a cap-proximal position [see Ref. 411 and can specifically bind. purified human IRE-BP (Constable and He&e, unpublished observations), but iron regulation of aconitase mRNA translation remains to be directly demonstrated. The possible physiological consequences of iron-dependent expression of the Krebs cycle enxyme aconitase are less obvious than in the case of eALAS. However, two features of aconitase provide pieces in the incomplete puzzle: first, aconitase is a classic example of an enzyme which utilizes an iron-sulphur cluster in its catalytic center [see Ref. 42; and review by A. J. Thomson in this issue]. Second, IRE-BP and aconitase share extensive amino acid homology [see Refs. 43, 441, such that aconitase activity of IRE-BP itself needs to be considered. Possible implications of this observation are discus& below. The somewhat obscure connection between the WordmaW regulation of intracelhrlar iron metabolism via IRE-BP and aconitase activity requires intensive investigation and may also reveal the physiological function (if any) of the IRE motif in aconitase mRNA.

188 L. C. Kiihn and M. W. Hentze

FUNCTIONAL CHARACTERIZATION OF IREs IREs function as specific binding sites for IRE-BP. As such, they control mRNA translation and turnover. Ferritin, eALAS, and TfR IREs display a higher degree of conservation with the respective elements from different species than the IREs of the different mRNAs of one particular species. In general, ferritin IREs are characterized by a relatively high content of A-U (or G-U) interactions in the top helix and G-C interactions in the bottom helix (Fig. 1). TfR IREs show a reversed pattern and eALAS IREs display an intermediate phenotype. Whether or not these differences have functional significance can not be answered with certainty, but two TtR IREs have been shown to function as translational regulators when placed as single IRE elements in the 5’ UTR of an indicator transcript [see Ref. 201. A structural definition of IREs can be derived from phylogenetic comparisons of functional, naturally occurring IREs (TfR, ferritin, eALAS), from transfection experiments using IRE variants, and from in vitro binding studies with IRE-BP as a ligand. Since it is quite possible that minor differences in binding affinities could have major implications for function as regulatory elements, care must be exercised with extrapolations from in vitro binding studies to in vivo function. With this caveat in mind and in the absence of extensive mutagenesis experiments, available data have been condensed to provide a preliminary definition of an IRE (Fig. 1). IREs are characteristic RNA stem-loop structures which share the following features: (i) a six-membered loop with the sequence 5’ CAGUGN 3’ (N = C, U, or A); (ii) a top helix of five (or occasionally four) paired bases; (iii) an unpaired 5’ C residue separated by five bases from the loop; and (iv) a bottom helix of somewhat variable length and position. Analysis of IRE-BP binding to full length’ ferritin mRNA by “toeprinting” showed that a region corresponding to the motif shown in Figure 1 served as the site of interaction for IRE-BP. This interaction was shown to affect the higher order structure of the region immediately flanking the IRE by short and long range interactions [see Ref. 451. The possible role of IRE-flanking sequences in translational regulation remains to be defined, whereas regions flanking the TtR IREs have been demonstrated to participate in the control of TfR mRNA stability [see Ref.

51. The IRE appears to be a motif with conserved sequence and structural features (Fig. 1). With regard to structural requirements, omission of the bottom helix was shown to result in a profound loss of affinity for IRE-BP [see Refs. 46, 471. Similarly, mutations of either nucleotides N,-N, or N,,-N,, of the top helix alone (which disrupt formation of the top helix) result in a loss of specific IRE-BP binding which is restored by combining the two compensatory mutations [see Ref. 471. In human ferritin L-chain and chicken ferritin H-chain mRNA, nucleotides N, and N,, do not have the potential for hydrogen bonding, but’ both mRNAs are translationally regulated by iron. Three deletion mutants reported thus far do not distinguish between sequence or structural perturbations. Deletion of the fully conserved C, nucleotide strongly reduces IRE-BP binding and abrogates function as a regulator of translation [see Refs. 33,481 and mRNA stability [see Ref. 231. The same effect on translation was observed for a deletion of N, [see Ref. 481. While the deletion of C , causes a - 50-fold reduction in IRE-BP binding [see Ref. 321, its replacement with an A residue has only minor effects on binding [see Refs. 46, 471. The loop sequence 5’ CAGUGN 3’ is highly conserved, but natural variants have been found in TfR and ferritin IREs. U,, is replaced by an A in human and rat

REGULATION OF IRON METABOLISM

189

TfR-IRE element A [see Refs. 22, 491 and by a C in the chicken homologue [see Ref. 211. In one human TtR cDNA clone [see Refs. 50, 511, G,, is substituted by an A in human TfR element C, but this substitution significantly reduces the apparent affinity for IRE-BP [see Ref. 251. N,, is most commonly a pyrimidine, but an A is found in human TfR element D (as above) and in the somal ferritin chain of the mollusc Lymnaea (W. Bottke, personal communication); a G residue in this position has not been found, perhaps because base pairing with C, has to be avoided. Mutagenesis studies support a functional role of the U,, variants [see Ref. 461, and further suggest that G, can be replaced by an A without a significant reduction in IRE-BP binding. It must be stressed however, that the effect of most of these variants or mutants on the in vivo function of IREs is still unknown. As a further reason for caution, an RNA motif which fulfills all criteria suggested by Figure 1 but fails to interact with human IRE-BP was recently identified in the Drosophila mefanogaster transcript toll [see Ref. 81. This apparent exception may be explained by the high content of predicted A-U and G-U base pairs in both helices of the toll IRE motif. CHARACTERIZATION OF THE IRE-BP The presence and activity of IRE-BP is generally measured in vitro by gelretardation assays in which a radiolabeled IRE-containing RNA is incubated with cytoplasmic protein extracts [see Ref. 301. The amount of RNA-protein complexes formed is quantitated by the retardation or “shift” of labeled RNA on nondenaturing polyacrylamide gels. An extension of this technique includes crosslinking of RNAprotein complexes with ultraviolet light prior to denaturing gel electrophoresis. This has permitted molecular weight estimates for IRE/IRE-BP complexes of about 100 kDa with rat liver extracts [see Ref. 301 and a doublet of 97 and 103 kDa with extracts from human placenta [see Ref. 241. Proteins with virtually identical properties were subsequently found in species from other vertebrate branches as well as insects and annelids [see Ref. 521, indicating a high conservation of IRE-BP in evolution. IRE-BP has since been purified to homogeneity either by affinity chromatography using an IRE-containing RNA-column [see Refs. 53, 541, or by a classical biochemical fractionation procedure in which rabbit liver IRE-BP was traced for its inhibitory activity on ferritin translation in vitro [see Ref. 351. Purified IRE-BP from both human and rabbit liver were reported as a single moiety of 90 kDa [see Refs. 35, 531, whereas the human placental protein migrated as a doublet with 95 and 100 kDa [see Ref. 541. Recently, the isolation of a cDNA clone for human IRE-BP with an open reading frame of 790 amino acids has been reported [see Ref. 551. The corresponding mRNA of about 3.6 Kb is encoded by a single copy gene on chromosome 9, a map location that matches the previous assignment [see Ref. 561. other groups have found independent cDNA isolates for rabbit (Walden and colleagues), and human IRE-BP (Hirling et al., unpublished work). In both cases, these clones match to a large extent the cDNA of Roiutult et al., such that it can be assumed that no new gene products have been cloned. However, the more recently reported cDNAs predict larger open reading frames with extensions in the deduced N-terminal region, specifying in the case of human IRE-BP an 889 amino acid protein with 97.5 kDa (Hi&g et al., unpublished work). Thus far no strong evidence for functional heterogeneity of celhrlar IRE-BPs has been found, but it seems possible that cells contain related proteins. Rouault et al. [see Ref. 551 reported a highly related cDNA

190

L. C. Kiihn and M. W. Hen&e clone that is encoded on chromosome 15, but for which a function has not been defined yet. Moreover, rat liver extracts were shown to form two IRE-protein complexes (Bl and B2) [see Ref. 301 due to cytoplasmic factors that can be separated biochemically [see Ref. 471. The most promising hints with respect to IRE-BP structure and function have come from computer-assisted compar&nsofthededucedaminoacidsequetMXof IRE-BP with available data banks [see Ref& 43,441. IRE&BP was found to belong most likely to a family of (4Fe4S) cluster proteins, the sequem~ homology being highest to mitochondrial aconitase and isopropylmalate isomerase. The sequence identity with mitochondrial aconitase from pig heart is about 3 1% over a 700 amino acid region, and the frequency of conserved residues close to 55%. Even more relevant is the complete conservation of all amino acid residues which contribute to the catalytic center of aconitase, including the three cyst&es known to coordinate the Fe-S cluster. The structural resemblances between mitochondrial aconitase and (cytoplasmic) IRE-BP have stimulated speculation whether IRE-BP may be identical to the poorly characterized cytoplasmic aconitase [see Refs. 43,441, in other words whether IRE-BP may be both an RNA-binding protein and an enxyme. Additional indirect support for such a notion comes from matching molecular weight assignments and the joint genomic locali&ion on human chromosome 9. Rouault and colleagues (IPIO Abs 046) reported that immunopurifled IRE-BP displays aconitase activity. Independently of a possible enxymatic activity, the likely presence of an Fe-S cluster will retain our attention because it may provide a key to understanding the iron-dependent changes in IRE-BP binding to IRIG. Several studies have documented a tight inverse correlation between the IRE-BP activity measured in vitro in celhilar extracts and the availability of intracelhilar “chelatable iron” before cell lysis [see Refs. 24, 31, 321. Thus, iron deprivation induces post-translationally a high affinity site (Rd = 10-‘” to lo- ” M) on the IRE-BP [see Refs. 32, 461, whereas the presence of iron salts or hemin in cell culture medium diminishes the affinity about 50-lOO-fold. Thus, in metabolicahy active cells, IRE-BP appears to exist in at least three pools, a high afllnity protein pool that is bound to endogenous cytoplasmic mRNAs, a free high-afllnity pool, and a pool of free inactive low-afIinity protein [see Ref. 461. At present, the simplest hypothesis to explain the effect of iron on IRE-BP activity is to envisage a direct effect on the putative Fe-S cluster. In aconitase one of the four Fe atoms in the cluster is known to be labile with the consequence of a functionally important conversion from an active (4Fe4S) containing enzyme to an inactive (3Fe4S) form [see Ref. 571. A similar situation might exist in IRE-BP. Treatment of purified human IRE-BP with ferrous iron salts dimhUes IRE-BP activity, whereas incubation with the iron chelator desferrioxamine induces a small increase in RNA-binding (Constable et al., unpublished work). While these findings support the notion of bidirectional (3Pe-4S)/(4Fe-4S) interconversions to activate or inactivate IRE-binding, we still lack direct proof for the existence of an Fe-S cluster in IRE-BP. Two findings of considerable interest are relevant to our picture of the mechanism of activation and inactivation of IRE-BP: the in vitro effects of redox compounds and hemin. Reducing compounds like &me rcaptoethanol are capable of converting free inactive IRE-BP into the high affmity state to the same extent as does iron chelation in vivo [see Ref. 3 11. In vitro activation can only be observed with IRE-BP that has not been activated in vivo prior to cell lysis, and requires a high nonphysiological

REGULATION OF IRON METABOLISM

191

concentration of reductant, 300 mM &mercaptoethanol being optimal. The effect is fully reversed by a specific sulfhydryl oxidizing agent, however, only as long as IRE-BP is not bound to an IRE [see Ref. 3 I]. The oxidized form can readily be reactivated a second time with a reductant even in the absence of any additional proteins [see Ref. 541. Furthermore, alkylation of SH-groups in IRE-BP irreversibly blocks the interaction with the IRE [see Ref. 311. These results have been taken as evidence that the iron-dependent modulation of IRE-BP involves the reduction of specific sulfhydryl-groups that affect the IRE binding site [see Ref. 311. It remains unknown which SH-groups are involved and how the iron status modifies the redox state of IRE-BP. A possibility would consist of an allosteric switch in which the number of Fe atoms or their charge in the hypothetical Fe-S cluster influences the conformational environment of a particular SH-group. Future experiments are needed to define to what extent the redox sensitivity of IRE-BP should be reinterpreted in light of the putative iron-sulphur chemistry. An alternative but not mutually exclusive possibility for the control of IRE-BP has been proposed by Thach et al. based on their observation that IRE-BP is also inactivated in vitro by incubation with hemin [see Refs. 58, 591. Lin et al. [see Ref. 601 suggest that hemin binds to a specific site on IRE-BP and controls in vivo the high affinity interaction between IRE-BP and IREs. The specificity of the hemin effect in vitro and its relevance to the in vivo situation has, however, been seriously challenged by Haile et al. [see Ref. 611, who found that several porphyrin compounds including hemin inactivated IRE-BP, but also other non-iron regulated nucleic acid-protein interactions. It is well established that hemin added to cells decreases TfR expression [see Refs. 15, 621. However, this effect is reversed by the iron chelator desferrioxamine, which by itself cannot chelate iron from hemin in vitro [see Ref. 631. The regulatory event with hemin in vivo was therefore interpreted to be the consequence of a release of chelatable iron [see Refs. 63, 641. An inhibitor of heme degradation, tin mesoporphyrin IX, reduces the ability of hemin to induce ferritin synthesis [see Ref. 641. Moreover, inhibition of heme synthesis by succinylacetone does not block iron-dependent regulation [see Refs. 64, 651, whereas stimulation of heme synthesis by &rninolevulinic acid or protoporphyrin IX has effects similar to iron chelation, presumably by lowering the chelatable iron pool [see Refs. 64, 661. Despite all the evidence in favor of chelatable iron as a dominant regulator of IRE-BP activity, heme or a heme-containing protein may play an additional role, for example, in electron transfer. PHYSIOLOGICAL

IMPLICATIONS

In regulating cellular iron uptake, intracellular storage and possibly erythroid heme synthesis, IRE-BP coordinates the most important pathways of iron metabolism (Fig. 2). This central regulatory role of IRE-BP is reflected in one of its several alternative names, “iron-regulatory factor” or “IRF” [see Ref. 241. Thus, iron homeostasis of each individual cell and, therefore, of the entire organism seems to depend on the regulatory function of IRE-BP. In every known case, iron deprivation activates IRE-BP which in turn induces mechanisms that compensate for the lack of iron by both raising its uptake via increased TfR expression and lowering its storage by inhibition of ferritin translation. Although TfR and ferritin expression are regulated in opposite ways, these effects are synergistic in terms of cellular iron physiology (Fig. 2). Iron is made available for synthetic pathways of heme or iron-containing

192 L. C. Kiihn and iU. W. Hen&e

Cellular Response:

IRE-BP

activity

low

IRE-BP

activity

high

Regulatory

-=w= TfR mRNA Degradation: blocked Ferritin

mRNATranslation:

eAl_AS mRNATranslation:

Homeostatic

Effect:

reduced iron uptake increased iron storage increased erythroid iron utilization

increased iron uptake reduced iron storage reduced erythroid iron utilizalion

FIGURE2. Regulatory feedback mechanisms controlling iron homeostasis. IRE-BP activity is modulated by a regulatory intracellular iron pool. Under conditions of low iron supply, this cytoplasmic mRNA-binding protein becomes active and binds to IREs in the S, respectively, 3’ untranslated regions of several mRNA species encoding proteins of central importance in iron metabolism. As a consequence, iron uptake, iron storage, and erytbroid heme synthesis are coordinately regulated in a way to compensate for iron deficiency. At increased iron levels, the regulatory cascade is inversed. Thus, under physiological steady-state conditions, iron is expected to maintain IRE-BP at an intermediate level of activation, and thereby to control its own homeostasis.

and will rise to a certain level, beyond which IRE-BP is again inactivated and the regulatory balance inversed (Fig. 2). The critical concentration and molecular nature of the regulatory iron pool remains unknown, but is likely to relate somehow to the equilibrium of its interaction with IRE-BP. It is possible to view the maintenance of iron homeostasis as an autoregulatory feedback loop. In this sense, IRE-BP is reminiscent of classic prokaryotic metabolite-controlled genetic switches. How IRE-BP senses the cellular iron balance, and what physical (allosteric?) changes influence its IRE-binding properties are questions at the heart of this coordinate control in iron metabolism. There is no doubt that adjusting iron to its appropriate level is of great importance for cells both in order to avoid the potential toxicity of excess iron and the deleterious effects of iron deprivation. With the recent discovery of a 5’ IRE in erythroid &rninolevulinic acid synthase, it becomes possible to consider the involvement of IRE-BP in the determination of the steady state iron equilibrium between tissues. Iron from senescent erythrocytes is continuously liberated by retoculoendothelial cells and contributes about 80% to the circulating transferrinbound iron pool in senmr. An equivalent amount is incorporated into freshly synthesized hemoglobin. The rate of heme biosynthesis in erythroid cells is, therefore, a major factor influencing the availability of iron for other tissues. The question arises whether and how IRE-BP responds in different tissues to changes in the proteins

REGULATION OF IRON METABOLISM

193

balance of serum iron, and which physiological events requite modulation of IRE-BP activity. The few results concerning these questions, suggest that IRE-BP is ubiquitously expressed and that its activity is modulated in liver and bone marrow following intraperitoneal administration of iron salts or an iron chelator [see Ref. 30 and Kiibn and colleagues, IPIO Abs 041, 0431. However, both studies have analyzed an acutely induced situation, which may not be representative for chronic iron overload or anemia. The only other physiological situation known to induce IRE-BP activity concerns the early phase of cell growth in lymphocytes that have been stimulated to proliferate [see Refs. 67, 681. In this ca.w, modulation of IRE-BP is linked to the suddenly increased requirements of iron supply in order to meet the de novo synthesis of iron-containing proteins. We conclude that the identification of IRE-BP has provided basic knowledge about post-transcriptional events regulating gene expression. Moreover, we now have a model to explain iron homeostasis at the cellular and organismic level. Due to this important progress, new questions can be approached. Further work is required to investigate the physico-chemical parameters of IRE-BP, its regulation by iron, and its interaction with IREs. The mechanisms by which IRE-BP controls translation and mRNA stability need to be understood in more detail. Finally, we need additional studies in vivo in order to document the predictions concerning the regulation of iron metabolism in a true physiological situation. We thank many of our colleagues in the field, particularly Robert Thach, Elizabeth Theil, and William Walden for sharing ideas and unpublished data with us. We are grateful to the members of our labs for comments on the manuscript.

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15.

J. B. Neilands, Biol. Metals. 4, 1 (1991). E. C. Theil, J. Biof. Chem. X5,4771 (1990). R. D. Klausner and J. B. Harford, Science 246, 870 (1989). J. Ziihringer, B. S. Baliga and H. N. Munro, Proc. Natf. Acad. Sci. USA 73, 857 (1976). E. W. Miillner and L. C. Kiihn, Cell 53, 815 (1988). E. Mattia, D. Josic, G. Ashwell, R. Klausner and J. van Renswoude, J. Biol. Chem. X1,4587(1986). T. C. Cox, M. J. Bawden, A. Martin and B. K. May, EMBO J. 10,1891 (1991). T. Dandekar, R. Stripeclte, N. Gray, B. Goossen, A. Constable, H. E. Johansson and M. W. Hentxe, EMBO J. 10, 1903 (1991). E. H. Morgan, Mol. Aspects Med 4, 1 (1981). A. Dautry-Varsat, A. Ciechanover and H. F. Lodish, Proc. Natl. Acud. Sci. USA 80 2258 (1983). R. D. Klausner, G. A&we& J. van Renswoude, J. B. Hatford and K. R. Bridges, Proc. Natl. Acad. Sci. USA 80, 2263 (1983). M. Karin and B. Mink, J. Biol. Chem. 256, 3245 (1981). B. J. JacoFettaand E. H. Morgan, J. Biol. Chem. 2%, 9108 (1983). A. Ciechanover, A. L. Schwartz, A. Dautry-Varsat and H. F. Lodish, J. Biol. Chem. 258, 9681 (1983). U. Testa, P. Thomopulos, 0. Vinci, M. Titeux, A. Bettaieb, W. Vainchenker and H. Rochant, l&p. Cell. Res. 140,251 (1982).

194 L. C. Kiihn and M. W. Hentze

16. E. Mattia, K. Rao, D. S. Shapiro, H. H. Sussman and R. D. Klausner, J. Biol. Chem. 259, 2689 (1984). 17. J. H. Ward, J. P. Kushnex andJ. Kaplan, J. Biol. Chem. 257, 10317 (1982). 18. D. Gwen, L. C. Kiihn, EMBO .I. 6, 1287 (1987). 19. J. L. Casey, B. Di Jeso, K. Rao, R. D. Klausner and R. B. Harford, J%c. Natl. Acad. Sci. USA 85, 1787 (1988). 20. J. L. Casey, M. W. Hen&e, D. M. Koeller, S. W. Caughman, T. A. Rouault, R. D. Klausner and J. B. Harford, Science 240, 924 (1988). 21. L. N. Chan, N. Grannnatikakis, J. M. Banks and E. M. Gerhardt, Nucf. Acids Res. 17, 3763 (1989). 22. K. P. Roberts and M. D. Griswold, Molec. Endocrinol. 4, 531 (1990). 23. J. L. Casey, D. M. Koeller, V. C. Ramin, R. D. Klausner and J. B. Harford, EMBO J. 8, 3693 (1989). 24. E. W. Mbllner, B. Neupert and L. C. Kuhn, Cell 58, 373 (1989). 25. D. M. Koeller, J. L. Casey, M. W. Hentxe, E. M. Gerhardt, L. N. L. Chan, R. D. Klausner and J. B. Hat-ford, Proc. Natl. Acad. Sci. USA 86, 3574 (1989). 26. D. M. Koeller, J. A. Horowitz, J. L. Casey, R. D. Klausner and J. B. Harford, Proc. Nat/. Acad. Sci. USA 88, 7778 (1991). 27. J. Rogers and H. Munro, Proc. Natl. Acad. Sci. USA 84, 2277 (1987). 28. N. Aziz and H. N. Munro, Proc. Natl. Acad. Sci. USA 84, 8478 (1987). 29. M. W. Hentxe, S. W. Catqghman, T. A. Rouault, J. G. Barriocanal, A. Dancis, J. B. Hat-ford and R. D. Klausner, Science 238, 1570 (1987). 30. E. A. Leibold and H. N. Munro, Proc. Natl. Acad. Sci. USA 85, 2171 (1988). 31. M. ,W. Hentxe, T. A. Rouault, J. B. Harford, R. D. Klausner, Science 244, 357 (1989). 32. D. J. Haile, M. W. Hentxe, T. A. Rouault, J. B. Harford and R. D. Klausner, Mol. Cell. Biol. 9, 5055 (1989). 33. B. Goossen, S. W. Caughman, J. B. Harford, R. D. Klausner and M. W. Hentxe, EMBO J. 9, 4127 (1990). 34. W. E. Walden, S. Daniels-McQueen, P. H. Brown, L. Gaffield, D. A. Russel, D. Bielser, L. C. Bailey and R. E. Thach, Proc. Nati. Acad. Sci. USA 85, 9503 (1988). 35. W. E. Walden, M. M. Patino, and L. Gaffield, J. Biol. Chem. 264, 13765 (1989). 36. G. E. Shull and E. C. Theil, J. Biol. Chem. 258, 7921 (1983). 37. W. E. Walden and R. E. Thach, Biochemistry 25, 2033 (1986). 38. P. H. Brown, S. Daniels-McQueen, W. E. Walden, M. M. Patino, L. Gaffield, D. Bielser and R. E. Thach, J. Biol. Chem. 264, 13383 (1989). 39. P. Die&s in Biosynthesis of heme and chlorophylls, H. A. Dailey, Ed. McGraw-Hill, New York, 1990. 40. B. K. May, R. C. Bhasker, M. J. Bawden and T. C. Cox, Mol. Biol. Med. 7, 405 (1990). 41. L. Zheng, P. C. Andrews, M. A. Hennodson, J. E. Dixon and H. Zalkin, J. Biol. Chem. 265, 2814 (1990). 42. A. H. Robbins and C. D. Stout, Proteins 5, 289 (1989). 43. T. A. Rouault, C. D. Stout, S. Kaptain, J. B. Harford and R. D. Klausner, Cell 64. 881 (1991). 44. M. W. Hentxe and P. Argos, Nucl. Acids. Res. 19, 1739 (1991). 45. C. M. Harrell, A. R. McKenzie, M. M. Patino, W. E. Walden and E. C. Theil, Proc. Natl. Acad. Sci. USA 88, 4166 (1991). 46. H. A. Barton, R. S. Eisenstein, A. Bomford and H. N. Munro, J. Bioi. Chem. 265, 7Ow (1990). 47. E. A. Leibold, A. Laudano and Y. Yu, Nucl. Acids Res. 18, 1819 (1990). 48. M. W. Hentxe, S. W. Caughman, J. L. Casey, D. M. Koeller, T. A. Rouault, J. B. Hat-ford and R. D. Klausner, Gene 72, 201 (1988).

REGUJATION

49. 50. 51. 52. 53.

C. L. A. S. T.

OF IRON METABOLISM

1%

Schneider, M. J. Owen, D. Banville and J. G. Williams, Nature 311, 675 (1984). C. Kuhn, A. McClelland and F. H. Ruddle, Cell 37, 95 (1984). McClelland, L. C. Killm and F. H. Ruddle, Cell 39, 267 (1984). Rothenberger, E. W. Miillner and L. C. Kll, Nucl. Acids Res. 18, 1175 (1990). A. Rouault, M. W. Hentxe, D. J. Haile, J. B. Harford and R. D. Klausner, Proc.

Natl. Acad. Sci. USA 86. 5768 (1989). 54. B. Neupert, N. A. Thompson, C. Meyer and L. C. Kiihn, Nucl. Acids Res. 18, 51 (1990). 55. T. A. Rouault, C. K. Tang, S. Kaptin, W. H. Burgess, D. J. Haile, G. Samaniego, 0. W. McBride, J. P. Harford and R. D. Klausner, Proc. Nat/. Acad. Sci. USA 87, 7958

(1990). 56. M. W. He&e, H. N. Seuanez, S. J. O’Brien, J. B. Harford and R. D. Klausner, Nucl. Acids Res. 17, 6103 (1989). 57. H. Beinert and M. C. Kennedy, Eur. J. Biochem. 186, 5 (1989). 58. J. J. Lin, S. Daniels-&Queen, L. Gatlield, M. M. Patino, W. E. Walden and R. E. Thach, Biochem. Biophys. Actu 1050, 146 (1990). 59. J. J. Lin, S. Daniels-McQueen, M. M. Patino, L. Gaffield, W. E. Walden and R. E. Thach, Science 247, 74 (1990). 60. J. J. Lin, M. M. Patino, L. Gaffield, W. E. Walden, A. Smith and R. E. Thach, Proc. Nuti. Acad. Sci. USA 88, 6068 (1991). 61. D. J. Haile, T. A. Rouault, J. B. Harford and R. D. Klausner, J. Biol. Chem. 265, 12786 (1990). 62. J. H. Ward, I. Jordan, J. P. Kushner and J. Kaplan, J. Biol. Chem. 259, 13235 (1984). 63. T. Rouault, K. Rao, J. Harford, E. Mattia and R. D. Klausner, J. Biol. Chem. 268, 14862 (1985). 64. R. S. Eisenstein, D. Garcia-Mayol, W. Pettingel and H. N. Munro, Proc. Nat/. Acad. Sci. USA 88, 688 (1991). 65. S. S. Bottomley, L. C. Wolfe and K. R. Bridges, 1. Biol. Chem. 268, 6811 (1985). 66. F. Louache, U. Testa, P. Pelicci, P. Thomopoulos, M. Titeux and H. Rochant, J. Biol. Chem. 259, 11576 (1984). 67. U. Testa, L. C. Kiihn, M. Petrini, M. T. Quaranta, E. Pelosi and C. Peschle, J. Biol. Chem. 266, 13925 (1991). 68. S. Teixeira and L. C. Kiihn, Eur. J. Biochem., 202, 819 (1991).

Received January 27, 1992; accepted February 3, 1992