Fe²⁺ binds iron responsive element-RNA, selectively changing protein-binding affinities and regulating mRNA repression and activation

Abstract

Iron increases synthesis rates of proteins encoded in iron-responsive element (IRE)-mRNAs; metabolic iron (“free,” “labile”) is Fe²⁺. The noncoding IRE-RNA structure, approximately 30 nt, folds into a stem loop to control synthesis of proteins in iron trafficking, cell cycling, and nervous system function. IRE-RNA riboregulators bind specifically to iron-regulatory proteins (IRP) proteins, inhibiting ribosome binding. Deletion of the IRE-RNA from an mRNA decreases both IRP binding and IRP-independent protein synthesis, indicating effects of other “factors.” Current models of IRE-mRNA regulation, emphasizing iron-dependent degradation/modification of IRP, lack answers about how iron increases IRE-RNA/IRP protein dissociation or how IRE-RNA, after IRP dissociation, influences protein synthesis rates. However, we observed Fe²⁺ (anaerobic) or Mn²⁺ selectively increase the IRE-RNA/IRP K_D. Here we show: (i) Fe²⁺ binds to the IRE-RNA, altering its conformation (by 2-aminopurine fluorescence and ethidium bromide displacement); (ii) metal ions increase translation of IRE-mRNA in vitro; (iii) eukaryotic initiation factor (eIF)4F binds specifically with high affinity to IRE-RNA; (iv) Fe²⁺ increased eIF4F/IRE-RNA binding, which outcompetes IRP binding; (v) exogenous eIF4F rescued metal-dependent IRE-RNA translation in eIF4F-depeleted extracts. The regulation by metabolic iron binding to IRE-RNA to decrease inhibitor protein (IRP) binding and increase activator protein (eIF4F) binding identifies IRE-RNA as a riboregulator.

Iron increases rates of ferritin protein synthesis in animals by facilitating messenger RNA/ribosome binding; metabolic iron (i.e., “labile” or “free” iron in cells) is considered to be ferrous (1). The iron response requires a noncoding riboregulator called the “iron-responsive element” (IRE), which is approximately 30 nt, folded into a distorted, bulged helix loop (2–5). This riboregulatory structure is also found in mRNAs for proteins of iron traffic (6–9), cell cycling (10), and the nervous system (11). IRP proteins bind with different stabilities to IRE-RNAs of the IRE-RNA family (12, 13), creating a graded or hierarchal set of mRNA responses to iron in vivo. Deletion of the 30 nt IRE-RNA not only removes IRP regulation but also decreases the rate of IRP-independent protein synthesis (14). A number of current models of IRE-RNA/IRP regulation feature iron-dependent degradation/modification of the IRP proteins as the main control point (8, 9, 15, 16). Such models do not answer two important questions: (i) How does iron increase release of IRP protein for [4Fe-4S]-modification and/or degradation? Overlap of the IRE-RNA and the Fe-S binding sites on IRP1 prevents Fe-S insertion in the IRP1/IRE-RNA complex (5). (ii) How does the IRE-RNA control rates of IRP-independent protein synthesis (14)? In an earlier study, we showed that Fe²⁺ ions (anaerobic) selectively increased the dissociation constant for the IRE-RNA/IRP1 complex in solution (12). Here we report the following observations: (i) Fe²⁺ binding to IRE-RNA in solution causing conformational changes at the two RNA/IRP contact sites; (ii) Fe²⁺ or Mn²⁺ increase in IRP-dependent protein synthesis in wheat germ extracts (WGE); (iii) IRE-RNA tightly binds eIF4F (a large protein complex required for ribosome binding) in addition to IRP; (iv) eIF4F outcompetes IRP for IRE-RNA binding as Fe²⁺ concentrations increase; (v) eIF4F binding to IRE-RNA is increased by metal ions as is IRP-independent protein synthesis in vitro. Regulation by metabolic iron binding to IRE-RNA is a genetic regulatory mechanism in eukaryotes that balances inhibitor protein (IRP)-RNA interactions and activator protein (eIF4F)-RNA interactions. The IRE riboregulator illustrates the potential of specific metal interactions with folded, noncoding mRNA structures for gene regulation.

Selective metal interactions with RNA structures are becoming increasingly well-known, such as those in ribozymes and riboswitches, which are predominantly found in bacteria (17, 18). Less is known about metal interactions with riboregulators in eukaryotic mRNAs, which function in cytoplasmic protein synthesis uncoupled from mRNA synthesis/DNA transcription, a contrast with bacterial mRNA riboswitches. Recently, metal-sensitive quadruplexes have been shown to function as a protein synthesis rate regulator in a human mRNA (ADAM10; 19, 20). Our earlier studies showed direct binding of a number of nonferrous metal ions to IRE-RNA, such as Mg²⁺ (21, 22) and Mn²⁺ (12) as well as complexes with Cu¹⁺ (21) and Ru²⁺ (23).

Regulation of stable mRNAs permits rapid cellular responses. Eukaryotic protein synthesis can be divided into three main phases: initiation, elongation, and termination. Different sets of cytoplasmic proteins are required for each phase; energy (ATP or GTP) is consumed in each phase (24, 25). Most regulation of protein synthesis rates occurs during the initiation phase when mRNAs are bound to and oriented on ribosomes by initiation factors (eIFs), and the mRNA AUG start codon is correctly aligned for recognition by the initiator tRNA^met and ribosome. eIF4F binding to eukaryotic mRNA is one of the first steps in protein synthesis and is thought to be rate limiting. eIF4F is a trimer of three protein subunits that are often studied separately. Each of the three eIF4F subunits binds directly to mRNAs. The eIF4F subunit eIF4E is the cap-binding protein that recognizes the 7-methyl guanosine mRNA cap structure on most eukaryotic mRNAs (24, 25). eIF4G, the large eIF4F subunit, is a scaffolding protein that recruits other eIFs and has a nonspecific RNA binding site. eIF4A is an ATP-dependent RNA helicase, which binds mRNA structures and is presumed to unwind secondary structure between the mRNA cap and the AUG during “scanning” (24). This noncoding region between the cap and coding region is called the 5′ untranslated region (5′ UTR). In spite of the potential regulatory role of eIF4F and its subunits, there have been few determinations of K_D values. In two studies of eIF4G/RNA binding, a mixture of eIF4A and eIF4G were used. The observed K_D for β-globin/eIF binding was 100 nM (26), and the K_D for poly(U)₄₀/eIF binding was in the μM range (27). In a third study, without eIF4A, the K_D was 4 nM for eIF4G binding to the structured picronovirus encephalomyocarditis virus (EMCV) stem loop (26).

Ferritin mRNA contains the oldest IRE-RNA in the IRE-mRNA family, based on evolutionary comparisons (6, 28) and has the most extensively characterized structure (2–5). The IRE-RNA hairpin in ferritin mRNA contains the conserved CAGUGN terminal loop and the conserved mid-helix bulge C (Fig. 1B) as well as a ferritin IRE-RNA-specific bulge U that is associated with tighter IRP binding and larger iron responses in vitro and in vivo (6, 28). In the IRE-RNA terminal loop, the base pair between conserved C and G bases create an AGU triloop (5, 28). The three main protein/RNA contact sites in IRE-RNA bound to IRP1 are in the terminal loop A15, terminal loop G16, and helix bulge C8 (Fig. 1) (5). A large RNA surface remains exposed, inviting RNA interactions with other molecules and ions even while bound to IRP1. Probing IRE-mRNA with chemical cleavage agents such as Cu¹⁺-1,10-phenanthroline or Mg²⁺ (21, 22) to locate metal-sensitive sites, combined with the X-ray crystal structure of the IRE-RNA/IRP1 complex, shows that some metal sites are located on exposed RNA surfaces and are available for additional RNA interactions even in the IRP1 complex (Fig. 1; 28). The selective role of Fe²⁺ in dissociating IRE-RNA from the protein synthesis inhibitor, IRP1, identified the chemical nature of the biological iron signal (12) and complemented regulatory models of IRP protein degradation by iron sensitive E3 ligase (29, 30) or IRP inhibition of IRE-RNA binding by insertion of a [4Fe-4S] cluster at the RNA binding site. Insertion of the [4Fe-4S] cluster converts IRP1 to c-aconitase (7, 31). How the IRE-RNA enhances protein synthesis rates in the absence of IRP (14) has remained obscure.

Fig. 1.

Structure of the IRE-RNA/IRP1 complex. (A) The figure was prepared with data in PDB ID code 3SNP using Discovery Studio Visualizer 3.0. RNA = orange-gold; protein = green. C8 and A15,G16 are conserved RNA/IRP contacts. The exposed RNA surface includes some metal-complex binding sites. (B) Secondary structure ferritin (FRT) IRE; boxed residues are sites used for 2AP substitution.

We now show that Fe²⁺ binding to IRE-RNA alters its conformation and increases IRP-dependent translation of IRE-RNA in vitro. Secondly, an alternate IRE-RNA binding protein, eIF4F, was identified. Moreover, we demonstrate Fe²⁺ enhanced eIF4F binding to IRE-RNA, which contrasts with Fe²⁺ inhibition of IRP1 binding to IRE-RNA (12) and competitive binding of IRE-RNA by eIF4F in the presence of IRP. Finally, we observed rescue of IRE-mRNA translation by exogenous eIF4F in eIF4F-depleted cell extracts and recapitulation of metal ion enhancement. Because Fe²⁺ is the predominant chemical form of iron in the cytoplasmic “iron pool” of eukaryotic cells (1), our observations indicate how a metal ion metabolite binds to a riboregulator structure that is capable of interacting with either a protein synthesis inhibitor (IRP1) or enhancer (eIF4F), for feedback regulation of iron metabolism.

Results

We previously determined that Fe²⁺ (anaerobic) and Mn²⁺ decreased IRP binding to IRE-RNA. We now report Fe²⁺ binding to IRE-RNA alters the RNA conformation at the two IRP/RNA contact sites (Fig. 2A). The IRE-RNA terminal loop A15 and bulged C8, two critical regions for IRP1 binding, were substituted with 2-aminopurine (2AP; Fig. 1B), which can reflect differences in base stacking and solvent accessibility. Anaerobic addition of Fe²⁺ to the 2AP substituted A15 and C8 IRE-RNA resulted in opposite effects on fluorescence intensity; A15 increased and C8 decreased (Fig. 2A and Fig. S1 A and B). Fe²⁺ addition to 2AP alone quenched fluorescence approximately 10% (Fig. S1C). These results demonstrate that Fe²⁺ ions induce conformational changes in IRE-RNA at two critical protein binding sites. Fe²⁺ binding to anaerobic solutions of IRE-RNA, measured as the decreased fluorescence of solutions of ethidium bromide bound to structured IRE-RNA (1.0 μM; Fig. S2A), occurred over the same concentration range as Mn²⁺ measured in air (12). No Fe²⁺ binding was observed for a 30-nt stem loop derived from yeast 5 S RNA (Fig. S2B). Metal selectivity for IRE-RNA binding was Fe²⁺ > Mn²⁺ > Mg²⁺ (Fig. S3).

Fig. 2.

Fe²⁺ changes IRE-RNA conformation and protein repressor/initiation factor binding in solution as well as increasing proteins synthesis in IRP-supplemented WGE. (A) Fe²⁺ increases fluorescence of IRE-RNA substituted with 2AP at A15 and decreases fluorescence of IRE-RNA substituted at C8. (RNA = 50 nM, 20 mM HEPES Buffer, pH 7.2, 100 mM KCl). The data are representative of three experiments. (B) Ferritin mRNA [full length, poly(A) tailed and capped] translated in IRP-supplemented WGE (4 nM RNA; RNA: IRP = 1∶15). ³⁵S-Met incorporation was quantitated by autoradioradiography. Fe²⁺ experiments used anaerobic conditions. The data are averaged from 2–4 independent translation experiments using 2 preparations of mRNA. *Significantly different from IRP + RNA with added metal ions; P < 0.01.

To link solution studies of Fe²⁺ on ferritin IRE-RNA with studies of cultured cells and animals, we examined effects of Fe²⁺ on ferritin mRNA translation in WGE containing exogenous rabbit IRP. Fe²⁺ reversed IRP inhibition of protein synthesis (Fig. 2B), consistent with the effect on IRP binding stability (12) and induction of ferritin synthesis in living cells and animals (6–10). Mn²⁺ also reversed the IRP inhibition, reaching full derepression at 50 μM, a contrast with Fe²⁺ (Fig. 2B) which, at similar concentrations, was not able to fully derepress translation, likely reflecting the larger number of iron-binding proteins in cell extracts.

Fe²⁺-induced release of IRP binding alone is insufficient to account for IRE-stimulation of ferritin synthesis in vitro (14) and for the larger iron response of ferritin in vivo, compared to other IRE-mRNAs (7). If IRE-RNA bound a protein synthesis initiation factor, the large iron response in vivo and the approximately 10-fold decrease in IRP-independent synthesis associated with IRE deletion (14), would be explained. The effects of IRE deletion occur in both rabbit reticulocytes which contains endogenous IRP and in WGE without endogenous IRP, indicating a common component. To identify the IRE-RNA positive control element, we examined IRE-RNA binding to eIF4F. eIF4F binds the ferritin IRE-RNA with a K_D of 9 nM (Fig. 3A), similar to the K_D for EMCV RNA and severalfold lower (tighter binding) than β-globin mRNA (26). By contrast, the 30-nt oligoribonucleotide stem loop from yeast 5S RNA did not bind eIF4F under the same conditions, K_D > 1 mM, (Fig. 3A), suggesting that eIF4F specifically recognizes the IRE stem loop, as previously observed for the stem loop in EMCV RNA (26). High-affinity eIF4F binding to ferritin IRE-RNA, even without the m⁷G cap (Fig. 3A), explains the previous observation that omitting the cap of IRE-RNA has little effect on protein synthesis contrasting with the large effect of deleting the IRE-RNA structure (14).

Fig. 3.

eIF4F binding to IRE-RNA: Sensitivity to metal ions and competition with IRP. Protein binding to IRE-RNA was monitored by changes in 5′ fluorescein(^Fl) labeled IRE-RNA fluorescence anisotropy (excitation- 490 nm, emission- 520 nm). (A) eIF 4F/ FRT IRE-RNA binding ± Fe²⁺. RNA: 0.050 μM, eIF4F: 0–100 nM in 20 mM HEPES, pH 7.2, 100 mM KCl. eIF4F did not bind a stem-loop 30 nt oligoribonucleotide used as a negative control (K_D > 1 mM). Fe²⁺ effects were significant (P < 0.01). (B) Selectivity of metal ions on eIF4F-IRE binding. Fe²⁺/Mn²⁺ have 20∶1 larger effects than Mg²⁺. Metal ion effects are significant,P < 0.02. (C) Competitive binding of eIF4F and IRP1 to FRT-RNA. RNA = 0.05 μM, eIF4F and IRP as indicated. Convergence of lines [1/(IRP)vs1/Δr] on the Y axis (circle) indicates competitive binding. (D) Fe²⁺ (anaerobic) has opposite effects FRT IRE-RNA binding to IRP or eIF4F. eIF4F-IRE with Fe²⁺ (red) (from A); IRP1-IRE RNA with Fe²⁺ (blue) (12); the significance of Fe²⁺ effects is P < 0.01.

Fe²⁺ ion increased the binding affinity of eIF4F for IRE-RNA nearly fivefold (Fig. 3D and Table S1). Metal selectivity of the effect is shown by the differences in metal ion concentrations required for comparable effects: Mg²⁺ required nearly 1,000 times higher concentration for an effect equivalent to 5 μM Fe²⁺ or Mn²⁺ (Fig. 3B). When Fe²⁺ concentrations in the cytoplasmic “free” or “signaling” iron pool increase; IRE-mRNA/IRP protein complex concentrations decrease. Current models focus on iron-dependent, proteasomal IRP degradation (32) and depletion of IRP1 upon insertion of a [4 Fe-4S] cluster at the IRE-RNA terminal loop binding site, which, coincidentally, confers c-aconitase activity on the protein.

eIF4F effectively competes with IRP1 for IRE-mRNA binding (Fig. 3C). Since a competitive inhibitor produces an apparent increase in the K_D of the substrate by the factor (1 + i/K_i), where i is the inhibitor concentration and K_i is the dissociation equilibrium constant for the inhibitor, the apparent K_D for IRP1/IRE interaction is equal to K_D,IRP(1 + [eIF4F]/K_D,4F) for competitive binding of IRP1 and eIF4F. In the absence of eIF4F, the dissociation of IRP1/IRE increases ∼10∶1 for [Fe²⁺], from 0 to 50 μM (Fig. 3D). In the presence of eIF4F: , where the superscript indicates 50 μM Fe²⁺ and the K_D subscripts indicate the species. Assuming [eIF4F] is large compared to K_D,4F, (K_D approximately 40 nM), then the dissociation of IRE/IRP1 is increased by the ratio of . Thus, when concentrations of Fe²⁺ reach 50 μM in the presence of eIF4F, the dissociation of IRP1/IRE is increased to 80∶1 compared to IRP1/IRE at zero concentrations of “free” Fe²⁺ and eIF4F. The competitive advantage of IRE-RNA binding to eIF4F compared to IRP is illustrated in Fig. 3D.

To determine the effects of metal-induced increased eIF4F binding on protein synthesis in vitro, WGE (no endogenous IRP) was depleted of eIF4F/iso4F (33). Exogenous eIF4F (rabbit reticulocyte, RR) restored ferritin synthesis in depleted WGE to 80%, confirming the contribution of eIF4F binding to IRE-RNA for protein synthesis (Fig. 4A). The effect of metal facilitated IRE-RNA/eIF4F binding (Fig. 3) is illustrated by the increase in protein synthesis when Mn²⁺ (the oxygen-resistant surrogate for Fe²⁺) was added to the depleted WGE extracts supplemented with exogenous eIF4F (Fig. 4A).

Fig. 4.

Mn²⁺ (oxygen-resistant) increases eIF4F-dependent protein synthesis, directed by IRE-riboregulated mRNA. A model for IRE-RNA riboregulator action. (A) Ferritin mRNA (4 nM) translated in eIF4F/iso4F depleted WGE, supplemented with rabbit eIF4F and 25 μM Mn²⁺ (an oxygen-resistant surrogate for Fe²⁺); the general conditions in Fig. 2 were used. (B) IRE-RNA riboregulator/dual protein control: a model. 1. IRE-mRNA/ribosome binding inhibited by IRP/IRE-RNA binding. 2. Fe²⁺ binding to IRE-RNA causes a (conformational) change decreasing IRP binding to IRE-RNA and increasing eIF4F binding to IRE-RNA; free IRP is degraded or converted (IRP1) to c-aconitase. 3. eIF4F/IRE-RNA ± cap/eIF-4E recruit ribosomes, to begin protein synthesis (translation and expression). Bright green = IRE-RNA; Blue = IRP; Gold = eIF4F; olive green = cytoplasmic aconitase; brown = ribosome.

Discussion

The major contribution a noncoding mRNA structure can make to gene expression and protein synthesis rates in eukaryotic cells is illustrated by the ability of IRE-RNA to bind, competitively, two regulatory proteins. IRE-RNA binds the large protein complex known to gate protein synthesis, eIF4, in addition to the well-known binding of IRE-RNA to IRP, a protein synthesis inhibitor. Phosphorylation of eIF proteins themselves or of eIF binding proteins (25), alters binding to mRNA under a variety of cellular conditions. Here we show the stabilization of eIF/RNA interactions by binding of a metabolic metal ion (Fe²⁺) to a noncoding riboregulator, the IRE-RNA. Increases in eIF4F/ferritin IRE-mRNA binding facilitated by Fe²⁺ occur at concentrations 100 times lower than with Mg²⁺ and with stabilities 20–1,000 times higher than for eIF4F binding to β- globin mRNA or the m⁷G cap structure (26, 27). Further, metal ion stabilized binding leads to enhanced ferritin IRE-RNA in vitro translation, demonstrating the biological effect of this interaction.

Metal binding to selective sites on IRE-RNA by small, stereospecific molecules such as Cu¹⁺- phenanthroline and or Ru²⁺-tris(bipyridyl) has been known for some time (21, 23). In the presence of dioxygen and the RNA-bound complexes, reactive intermediates form. These cleave the folded RNA at the binding site of the metal-complex; sequencing of the cleaved RNA by primer extension analysis identifies the RNA binding sites. Such studies have shown specific binding of the metallo-complexes at sites near helix distortions or in the terminal loop of the IRE-RNA structure (21, 23). The possibility of Fe²⁺ binding to IRE-RNA in ferritin mRNA is suggested by the results of IRE-RNA cleavage with Fe²⁺ and hydrogen peroxide (34). The unusual pattern of hypersensitive cleavage sites was initially explained as solvent access prevented by RNA folding (34), a supposition that was not supported by NMR structures obtained later (2, 3). Alternatively, the hypersensitive sites could indicate Fe²⁺/RNA binding sites. Some of the IRE-RNA metal sites observed by metallo-chemical probing are at bases that are on the exposed RNA surfaces in the IRE-RNA conformation observed in the IRP/IRE-RNA crystal structure (Fig. 1), and thus could be accessible to metal ions while the IRE-RNA is bound to IRP. Experiments with 2AP substituted IRE-RNA showed that Fe²⁺ altered the conformation of both the terminal IRE loop (A15 was less stacked), and the bulge C8 (more ordered; Fig. 2A). A15 and C8 are part of the two IRE-RNA/IRP1 contact sites and have multiple protein contacts (2–5). Details of these conformational changes and direct identification of the Fe²⁺ binding site on IRE-RNA are outside the scope of this present study, however.

How do the observed solution effects of Fe²⁺ on IRE-RNA interactions with IRP and eIF4F proteins relate to physiological iron? Quantitative data on free cellular iron is limited. In normal mammalian cells, free [Fe²⁺] is approximately 0.5–1.0 μM measured by fluorescence quenching analysis using calcein (35) and more recently, by potentiometric methods (36). Ferritin protein accumulations are detectable, measured immunologically, at intracellular [Fe²⁺] approximately 5–10 μM. Our data show, at 5–10 μM [Fe²⁺] in solution, a 6∶1 decrease in IRP binding and a 1.5∶1 increase (P < 0.01) in eIF4F binding (Fig. 3C). Because eIF4F and IRP bind competitively (Fig. 3B), the IRE/eIF4F complex can increase 8∶1, for similar IRP:eIF4F concentrations at 5 μM [Fe²⁺]. The effect could be 80∶1, at 50 μM Fe²⁺.

A model of ferritin IRE-mRNA riboregulator function (Fig. 4B) shows Fe²⁺ metabolite-regulated IRE-mRNA-IRP inhibition of protein synthesis, where ribosome binding is prevented (7, 15, 28), combined with Fe²⁺-metabolite regulated IRE-mRNA-eIF4F enhancement of protein synthesis (Fig. 4A), where ribosome binding is facilitated. The model also includes the inactivation of IRP for IRE-RNA binding from older models (7, 8, 13, 29–31, 37, 38). Increasing cellular concentrations of Fe²⁺ sufficiently to bind IRE-RNA will lower RNA/IRP affinity and increase eIF4F binding, ribosome assembly, and mRNA translation. During revision, complementary, qualitative data appeared suggesting possible noncanonical translation of ferritin mRNA (39). The recent extensions of the IRE-RNA family to include mRNA encoding a cell cycle protein (40), α–hemoglobin chaperone (41) and amyloid precursor protein (11), in addition to the iron metabolic proteins (e.g., ferritin, ferroportin, and mt-aconitase), indicates that IRE-mRNA control of protein synthesis affects a number of metabolic processes in animal cells. As shown for the IRE-mRNA, other metal metabolite–riboregulator interactions, such as those in mRNA quadruplexes (19, 20) that control mRNA activity, may also involve RNA binding of multiple, regulatory proteins.

Experimental Procedure

Preparation of IRP1, eIF4F, and RNA.

Isolation of recombinant rabbit IRP1 from yeast used methods previously described in (12, 42). The eIF4F protein was purified from rabbit reticulocyte lysate (RRL) as described previously (43).

Fluorescein-labeled RNA 30 nt oliogonucleotides (^FIIRE-RNA), ferritin IRE-RNA, and 2-aminopurine substituted 30 nt IRE-RNA were purchased from Genelink and, after dissolving, were melted and reannealed as described in references 12 and 44 by heating, in 40 mM HEPES, pH 7.2, 100 mM KCl, 5% glycerol to 85 °C for 15 min with slow cooling to 25 °C. The concentrations of ferritin IRE-RNA and 2AP IRE-RNA were determined spectrophotometrically from the measured absorbance at 260 nm with 1.0 A₂₆₀/mL = 40 μg/mL RNA. The concentration of protein was determined by the Bradford method with bovine serum albumin as the standard using a Bio-Rad protein assay reagent.

Binding of Fe²⁺ to IRE-RNA.

The change in fluorescence intensity of ethidium bromide bound to structured IRE-RNA was used to detect Fe²⁺ binding to IRE-RNA, after heating and annealing the RNA as before (12, 44); the methods were used previously to show Mn²⁺ binding to IRE-RNA (12, 44). However, with Fe²⁺, anaerobiosis was maintained with argon purging in a glove box. Under such conditions, IRE-RNA is stable in the presence of Fe²⁺ (12, 44). Binding studies were carried out in buffer containing 20 mM HEPES, pH 7.2, and 100 mM KCl. RNA concentrations were 50–100 nM. Fe²⁺ binding to 2AP substituted oligonucleotides was carried out under the same conditions.

Protein Synthesis Assays.

For all protein synthesis assays, the transcript RNA [capped, with poly(A) tail] was heated at 65 °C for 5 min, binding buffer (24 mM HEPES pH 7.2, 60 mM KCl) at the same temperature was added and the sample annealed slowly to room temperature (30 min). After metal ion addition, solutions were incubated at 4 °C for 30 min for metal binding; anerobiosis for Fe²⁺ was obtained with sealed glass vials purged with argon. IRP, where indicated, was added to the RNA for 30 min prior to translation. WGE (Promega) mixtures were prepared according to the manufacturer’s instructions; eIF4F/iso4F depletion used methods described previously (33). RNA or RNA/IRP was added to WGE (final RNA concentrations were 4 nM) and incubated at 25 °C for 20 min. ³⁵S-Met incorporation into ferritin was determined with 12% SDS/PAGE with quantification by autoradiography with ImageQuant software.

Binding of eIF4F to IRE-RNA.

The equilibrium constant for eIF4F binding to ferritin IRE-RNA was determined by monitoring the fluorescence anisotropy change when increasing amounts of eIF4F were added to 50 nM ^Flferritin IRE-RNA in 20 mM HEPES Buffer, pH 7.4, 100 mM KCl at 25 °C. The K_D was determined by fitting the plot of changes in anisotropy vs. eIF4F concentration using the equation r_obs = r_min + {(r_max-r_min)/(2[IRE])}{b-(b² - 4[IRE][eIF4F])^0.5} where the r_obs is the observed anisotropy value; r_min the minimal anisotropy value; r_max is final saturated anisotropy value; [IRE] and [eIF4F] are the concentrations of IRE-RNA and eIF4F protein, respectively; b = K_D + [IRE] + [eIF4F] as described elsewhere (45, 46). Data were fit using KaleidaGraph (Abelbeck Software).

eIF4F and IRP Competitive Binding for IRE-RNA.

IRE-RNA (50 nM) was mixed with varying concentrations of eIF4F protein and titrated with IRP1 protein as described above. The data were analyzed using the Lineweaver–Burke plot, 1/r-r_obs vs 1/[IRP]. Intersection of the lines at the same point on the Y-axis indicated competitive binding contrasting with the parallel lines that would be observed for uncompetitive binding. Data were fit using KaleidaGraph (Abelbeck Software).

eIF4F-IRE Binding Responds to Changes in Iron Concentration.

Freshly prepared solutions of FeSO₄ in deoxygenated (argon-purged) water were added anaerobically to argon-purged solutions of reannealed IRE-RNA and the experiments were performed in an inert atmosphere as described previously (12, 44). The anisotropy of fluorescein-labeled ferritin IRE-RNA (50 nM) was measured with increasing eIF4F protein and various concentrations of Fe²⁺, in 20 mM HEPES buffer, pH 7.4, 100 mM KCl. K_D values were calculated as described for experiments in the absence of Fe²⁺.