Differential translational control of 5’ IRE-containing mRNA in response to dietary iron deficiency and acute iron overload

Abstract

Iron regulatory proteins (IRPs) are iron-responsive RNA binding proteins that dictate changes in cellular iron metabolism in animal cells by controlling the fate of mRNAs containing iron responsive elements (IREs). IRPs have broader physiological roles as some targeted mRNAs encode proteins with functions beyond iron metabolism suggesting hierarchical regulation of IRP-targeted mRNAs. We observe that the translational regulation of IRP-targeted mRNAs encoding iron storage (L- and H-ferritins) and export (ferroportin) proteins have different set-points of iron responsiveness compared to that for the TCA cycle enzyme mitochondrial aconitase. The ferritins and ferroportin mRNA were largely translationally repressed in the liver of rats fed a normal diet whereas mitochondrial aconitase mRNA is primarily polysome bound. Consequently, acute iron overload increases polysome association of H- and L-ferritin and ferroportin mRNAs while mitochondrial aconitase mRNA showed little stimulation. Conversely, mitochondrial aconitase mRNA is most responsive in iron deficiency. These differences in regulation were associated with a faster off-rate of IRP1 for the IRE of mitochondrial aconitase in comparison to that of L-ferritin. Thus, hierarchical control of mRNA translation by IRPs involves selective control of cellular functions acting at different states of cellular iron status and that are critical for adaptations to iron deficiency or prevention of iron toxicity.

INTRODUCTION

Sensory and regulatory mechanisms utilizing RNA-based control form essential homeostatic networks in prokaryotes and eukaryotes (1–4). RNA binding proteins, non-coding RNAs and mRNA targets form RNA regulons that coordinate fundamental cellular processes such as growth control and whose dysregulation is associated with common human diseases including oncogenesis (1). IRPs are iron-regulated RNA binding proteins that sense cellular iron levels and control the fate of mRNAs essential for the maintenance of metazoan iron homeostasis in diverse physiological scenarios (3,5). IRPs bind stem-loop structures termed iron responsive elements (IREs) in the untranslated regions (UTR) of mRNA encoding proteins required for controlling the uptake and metabolic fate of iron, the adaptive responses to iron deficiency and other physiological processes. IRPs directly control mRNA translation or stability depending on the location of the IRE. The first identified targets of IRP action are the mRNA encoding the H- and L-subunits of the iron storage protein ferritin (FTH1 and FTL) and the iron uptake protein transferrin receptor 1 (TFRC). IRPs bind to the single IRE near the 5′ end of ferritin mRNA and block an early step in the initiation phase of translation (6). In contrast, IRPs bind to multiple IREs in the 3’ UTR of Tfrc mRNA where they promote its stabilization. To date functional IREs have been identified in ten mRNA, and transcriptomic studies suggest that IRPs may control a much broader post-transcriptional regulon (3,7–9). Several of the more recently identified IRE-containing mRNAs encode proteins without direct roles in iron metabolism and for some mRNA IRPs have been proposed to interact with sequences lacking the canonical IRE structure (3,7,8,10). The breadth of the cellular processes controlled by proteins encoded by mRNA with established functional IRE(s) suggests that IRPs differentially regulate mRNA fate (5,11). However, it has not been determined if iron-dependent control of IRP action allows for selective control of proteins directly involved in iron metabolism versus proteins needed for adaptive changes in cellular function.

Identification of a functional IRE in the 5’ UTR of the tricarboxylic cycle enzyme mitochondrial aconitase (Aco2) mRNA provided the first indication that the physiological action of IRPs extended beyond the direct control of the uptake and metabolic fate of iron (12,13). IRE-containing mRNAs encoding proteins involved in cell cycle regulation (CDC14A) and oxygen-sensing (HIF-2 α; EPAS1) were then identified, further expanding the range and organismal roles of the IRP RNA regulon (14,15). Subsequently, a transcriptome-wide approach identified a broad array of potential IRP mRNA targets which led to the recent demonstration of the unanticipated role of the actin binding protein profilin 2 (PFN2) in iron metabolism (8). These findings support the concept that the iron-dependent control of mRNA fate by IRPs may be separable into primary targets required for the maintenance of cellular iron homeostasis and secondary targets involving proteins that mediate adaptive changes in cellular pathways during stress such as those needed for cell survival in iron-deficient environments.

The functional impact of naturally occurring IRE mutations supports the concept that RNA binding hierarchy is a key factor allowing for selective action of IRPs. Many mutations in human Ftl IRE individually give rise to hereditary hyperferritinemia cataract syndrome (HHCS) and high serum ferritin (16–18). The severity of the disease phenotype in HHCS is strongly related to the extent to which these IRE mutations reduce IRP binding affinity and is directly related to the accumulation of serum ferritin, a likely measure of translational derepression of Ftl mRNA (16). The strongest increase in serum ferritin accumulation occurs over the first 10-fold decrement of IRE binding affinity; mutations that cause a greater loss of RNA binding showed a smaller relative increase in serum ferritin (16). Interestingly, the affinity with which IRP1 recognizes the six established 5’ IREs varies over a 9-fold range (19). These differences in affinity of interaction of IRP1 with the known 5’ IREs also supports the existence of an iron-dependent translational regulatory hierarchy essential for the maintenance of metazoan iron homeostasis.

The presence of 5’ IRE in multiple mammalian mRNA encoding proteins of differing metabolic functions provides a means to determine the extent to which IRPs selectively determine mRNA fate. Several studies have demonstrated strong repression of ferritin subunit mRNA in animals or cell lines (20–23). In comparison, erythroid 5-aminolevulinate synthase (Alas2) and Hif-2α mRNAs are less strongly repressed in vivo (24–26), as is Aco2 mRNA in a cell-free system (27). While these findings indicate that the translation of 5’ IRE-containing mRNA is selectively regulated, the extent to which the set point may vary for iron regulation of the translation of these mRNA has not been determined. Furthermore, while results of previous studies support the existence of a translational regulon amongst IRE-containing mRNA (24), the full spectrum of iron-regulation of the translation by IRPs in a single physiological system has not been investigated. Whether this hierarchical mechanism can fully explain iron regulation of the steady-state level of proteins encoded by 5’ IRE-containing mRNA is not clear.

In this study, we determined the impact of iron overload and iron deficiency on the translational regulation of multiple 5’ IRE-containing mRNA in rat liver. Our results show that 5’ IRE-containing mRNA encoding proteins that directly control cellular iron metabolism are more translationally repressed in normal liver and respond most strongly to iron overload. In contrast, the 5’ IRE-containing mRNA encoding Aco2 mRNA is largely translated in normal liver and is most responsive to iron deficiency. We demonstrate that the IRE from a weak target, Aco2 mRNA, dissociates from IRP1 much faster than does L-ferritin IRE suggesting enhanced access of Aco2 mRNA to the translation machinery. Taken together, our results demonstrate that the iron-dependent regulation of 5’ IRE-containing mRNA translation selectively controls pathways involved in the adaptive responses to iron deficiency and those required for survival in iron overload.

EXPERIMENTAL

Polysome profile analysis:

A method similar to Anthony et. al was used (28). Livers were excised and washed in ice-cold polysome buffer (40 mM HEPES pH 7.4, 100 mM KCl, 5 mM MgCl₂, 2 mM citrate, and 1 mM DTT). The liver was minced into small (~0.25–0.5 cm³) pieces and a representative mixture of pieces were homogenized in 3 volumes of polysome using a Potter Elvehjem homogenizer fitted with a teflon pestle. The homogenate was centrifuged at 5000 × g at 4 °C for 20 min. The upper 2/3 of supernatant was collected and nine volumes of supernatant were diluted with one volume detergent (10% sodium deoxycholate, 10% Triton X-100). After gentle mixing 500 μL of the sample was layered over an ice-cold 11 ml linear 15% to 60% sucrose gradient, in PB. The samples were centrifuged at 180,000 × g in a Sorvall TH641 swinging bucket rotor for 2 h at 4 °C with slow braking. The gradients were fractionated on an ISCO model UA-6 gradient fractionator. The absorbance at 254 nm was continuously monitored and ten 1-min (~1.2 ml) fractions were collected and stored at –80 °C overnight. Total RNA was isolated from 500 μL of each gradient fraction using Trizol (ThermoFisher). The integrity and location in the gradient of the 18S and 28S rRNAs was determined by agarose gel electrophoresis. Using the image as a guide, it was determined that fractions 1–2 contained the free protein pool, some of the ribonucleoprotein particle (RNP) pool and a portion of the 40S ribosomal subunit; fractions 3–4 contained RNPs, both ribosomal subunits and the 80 S monosome; and fractions 5–10 contained the light and heavy polysomes. Since the sedimentation velocity of a translationally repressed RNA will vary depending on the size of the mRNA some repressed mRNA (RNP) may overlap with the ribosomal subunits or the 80S monosome. Hence finding an mRNA in the 40S or 80S region may indicate it is translationally repressed or it can also indicate it is on the pathway of translational initiation. RNA from each gradient was reverse-transcribed to synthesize total cDNA (Reverse Transcription System, Promega). The total amount of mRNA for m-acon, L-ferritin, citrate synthase, and transthyretin in each gradient fraction was quantified by real time PCR with SYBR Green using a Roche Light Cycler. PCR product size was confirmed by agarose gel electrophoresis.

The sequence of the PCR primers used are: Rat Fth1, sense: 5′ TACCACCAGGACTCGG; anti-sense: 5′ GGAAGATTCGTCCACCTC; Ftl, sense, 5′-CACTTCTTCCGCGAATTG-3’, anti-sense, 5′-TCAGAGTGAGGCGCTCAA-3’; Fpn1, Sense: 5′ TGGAAGCCCCTTGGA C; anti-sense: 5′ CCAAAGACCGATTCTAGC; Aco2, sense, 5′-GATCCGAGCCACTATCGA-3’, anti-sense, 5′-TGGATCAAAGTCCGATCG-3’; citrate synthase (Cs), sense, 5′-CACTCAACTCGGGACG −3’, anti-sense, 5′-CCTCGACACTCCGAAC-3’; and transthyretin (Ttr), sense, 5′-GGCAGCCCTGCTGTCGAT-3’, anti-sense, 5′-TGCTGTAGGAGTACGGGC-3’. The level of 18S ribosomal RNA was quantified using QuantumRNA Classic 18S primers (ThermoFisher).

Total RNA level by qPCR:

A portion of the homogenate was used for quantification of mRNA abundance by qPCR. The level of expression of these mRNA was normalized to the concentration of 18S ribosomal RNA in each sample. RNA was extracted from liver homogenates using Trizol according to the manufacturer’s directions. RNA integrity was confirmed as noted above.

Immunoblotting:

FTL, FTH1 and ACO2 protein level was determined by immunoblotting. Liver homogenate was solubilized by diluting it to a final concentration of 1% Triton X-100 followed by incubation in ice for 10 min. Insoluble material was removed by centrifugation at 14,000 × g for 10 min at 4 °C. Then 5 or 50 μg of protein from the solubilized homogenate was denatured for 10 min in Laemmli’s reducing sample buffer at 65 °C or 100 °C for m-acon and ferritin samples, respectively (29). Protein concentrations were determined using the bicinchronic acid (BCA) method (ThermoFisher). A 10% polyacrylamide-SDS gel was used for all immunoblotting except for ferritin where the Tricine-SDS buffer was used (29). After transfer to nitrocellulose (Schleicher and Schuell) proteins were detected with protein A-purified rabbit IgG against bovine heart m-acon and rat liver ferritin, respectively, followed by incubation with goat anti-rabbit IgG-horseradish peroxidase conjugate (Southern Biotech) (29) using SuperSignal (ThermoFisher). Immunoblots was quantified by densitometry. Examination of α-tubulin as a loading control revealed its abundance to be altered in iron deficient liver, so immunoblots were normalized to total protein load (results not shown).

Animal treatment:

The use of animals was reviewed and approved by the Institutional Animal Care and Use Committee of the University of Wisconsin Research Animal Resource Center.

Study 1 - acute iron overload:.

Adult male Sprague-Dawley rats weighing 167 ± 1.4 g were provided with the purified control (C) diet containing 50 mg Fe/kg diet for 5 to 6 days (29). Rats were injected with either 0.6 ml of 10 mg/ml ferric ammonium citrate (FAC) or 0.6 ml of 10 mg/ml ammonium citrate (AC) three hr prior to killing. FAC and AC were prepared in sterile 0.9% NaCl with a final pH of 7.4. At termination rats in the control group weighed 221 ± 3 g (n = 3) while rats in the iron injected group weighed 219 ± 3 g (n = 3). A previous study using similar aged rats injected with a similar dose of iron per 100 g body weight found a 168% increase in total liver iron at 4 hr after injection (22).

Study 2 - iron deficiency without anemia.

In this experiment weanling mice (21 d; 47 ± 1.1 g) were fed an iron adequate diet for 14 days to increase their iron stores before they were fed the C of iron-deficient (ID, < 2 mg Fe/kg diet) diet from the age of 35 to 49 days when they were considered adult age. At d 49 there was no difference in body weight in the control (223 ± 16) or ID (221 ± 7) groups. This approach allowed us to induce iron deficiency without anemia, in contrast to the approach in study 3 that produced rats with iron deficiency anemia.

Study 3 – iron deficiency with anemia.

Weanling (21 d) male Sprague-Dawley rats weighing 45 ± 0.6 g were housed and fed either the C or ID diet for 7 to 9 days (d 28 – d 30 of age). Three rats from each diet group were killed and their livers were isolated for polysome profile analysis. At the end of the experimental period there was no significant difference in body weight of rats in the control (100 ± 2 g) or ID (90 ± 4 g) groups (mean ± SEM).

Blood analyses:

Blood was collected from the inferior vena cava of the anesthesized rats after which livers were excised for polysome analysis. Hemoglobin (Hb) concentration was determined in heparinized blood (30). A portion of non-heparinized blood was used for preparing serum. Serum was used for the determination of percent transferrin saturation with iron, serum iron, and total iron binding capacity (Catalog # 565-A; Sigma).

RNA binding analyses:

IRP RNA binding activity was determined using [³²P]rat Ftl IRE as described (19). Since both IRP1 and IRP2 regulate IRE-containing mRNA fate we report RNA binding activity as the summation of spontaneous RNA binding activity of IRP1 plus IRP2 in the as-isolated cytosol (i.e., without activation by 2-mercaptoethanol).

RNA dissociation from IRP1:

Off-rate assays were performed using two approaches, electrophoretic mobility shift assay (EMSA) or nitrocellulose filter binding (31). Recombinant IRP1 was expressed in yeast, purified and the concentration of active IRP1 determined as described (31). [³²P]RNA was folded before the RNA binding assay as described (31). An IRP1:IRE complex was allowed to form using either 50 pM [³²P]Ftl IRE with 500 pM active IRP1 or 6 pM of [³²P]Aco2 IRE with 130 pM active IRP1, incubated at 30°C for 10 min in binding buffer (5 mM DTT, 20 μg/ml nuclease-free BSA, 5% glycerol, 1mM magnesium acetate, 20 mM Hepes pH 7.5 and 75 mM potassium chloride). Higher amounts of IRP1 and IRE were used for the Ftl assay because the slower dissociation rate of this complex did allow for measurable amounts of free RNA if the lower concentration was used. Unlabeled competitor RNA was then added (25 nM Ftl IRE, UGUAUCUUGCUUCAACAGUGUUUGGACGGAACAGA; 1.3 nM Aco2, GCGACCUCAUCUUUGUCAGUGCACAAAAUGGCGCC, Dharmacon) and dissociation was allowed to occur at 30 °C. The amount of RNA-protein complex remaining was determined by EMSA or filter binding assay. Control experiments were carried out the same as described, except using anti-sense L-ferritin IRE, UCUGUUCCGUCCAAACACUGUUGAAGCAAGAUACA, Dharmacon) as competitor RNA. Samples (30 μl) were withdrawn at specific time points and mixed with 3 μl of 0.5 mg/ml heparin and 25 μl of the mixture was loaded onto the gel with the current on at 90V. After all samples were loaded, the gel was run for 25 min at 250 V. Gels were loaded and run at 4°C to minimize dissociation of RNA from IRP1 during the run.

For the nitrocellulose filter binding assays samples were passed through nitrocellulose filters (Whatman, Protran BA85) that had been pre-wet with binding buffer (4°C) and incubated in buffer on ice until needed. Sample filtration was followed by addition of binding buffer (750 μl) at 4°C. The entire process from loading the sample to the drying of the filter by vacuum took 75 sec. The filter was then placed in scintillation vials with addition of scintillation mixture for counting. The data were fit to an exponential decay curve using GraphPad software.

Statistical analysis:

Differences between group means were determined by Student’s T-test. Differences were considered significant at a P < 0.05.

RESULTS

Differential translational repression of 5′ IRE-containing mRNA in liver:

To determine the extent to which IRPs hierarchically control the translation state of IRE-containing mRNAs we first performed studies in rats fed an iron sufficient diet that were injected with ferric ammonium citrate (acute iron overload) or ammonium citrate (control). A typical liver polysome profile from control animals revealed that the majority (~80%) of liver ribosomes were present as polysomes (Fig. 1A). We determined the translation state of four 5’ IRE-containing mRNA, those encoding Fth1, Ftl, Fpn1 and Aco2, in liver of acutely iron overloaded or control rats. The non-IRE-containing mRNAs encoding Tth and Cs mRNAs served as controls.

figure 1:

translation state of 5′ ire-containing mrna in control and acutely iron overloaded rat liver. adult male rats were intraperitoneally injected with 6 mg of either ammonium citrate (control) or ferric ammonium citrate (iron). polysome profile analysis was performed using liver post-mitochondrial supernatants separated on a sucrose gradient. the gradient was divided into 10 fractions of equal volume from which total rna was isolated and analyzed by qpcr. a. optical density at 254 nm, continuously monitored from the top (left) to the bottom (right) of the gradient. b. agarose gel showing rna from gradient fractions. fractions 1–2 contain the free protein pool, some of the messenger ribonucleoprotein pool and some of the 40s ribosomal subunit. fractions 3–4 contain rnps, the ribosomal subunits and 80 s monosome, and fractions 5–10 contain the light and heavy polysomes. c.-h. results of qpcr showing the distribution of the following mrnas across the gradients: c. transthyretin (tth); d. citrate synthase (cs); e. l-ferritin (l-ftn); f. h-ferritin (h-ftn); g. ferroportin (fpn); and h. mitochondrial aconitase (m-acon). results are expressed as mean ± standard error of the mean for n = 3 rats per group. an asterisk indicates significantly different from control, p < 0.05. see table 1 for quantification of results.

Both Tth and Cs mRNAs were well translated in the control rats injected with ammonium citrate given that at least 80% of these mRNA was polysome associated (Fig. 1C and 1D, Table 1). In contrast, the translation state of each of the 5′ IRE mRNA examined was more highly repressed in control rats, although the degree of repression varied depending on the mRNA. In this context, repression refers to the extent to which a given mRNA is not associated with polysomes. Fth1 and Ftl mRNAs were most strongly repressed with only 12% of each mRNA being polysome-associated (Fig. 1E and 1F; Table 1). On a fractional basis and using the percent of Fth1 and Ftl mRNAs on polysomes as the reference about 48% more Fpn1 (17.8% Fpn1 vs. 12% Fth1 and Ftl) and 400% more Aco2 mRNA (62% m-acon vs. 12% ferritin) was present in the polysome fractions in liver if rats fed the iron sufficient diet (Fig. 1G and 1H, Table 1). These differences in translation state of the Fth1 and Ftl, Fpn1 and Aco2 mRNA suggest that IRPs variably repress the translation of 5′ IRE containing mRNAs in liver. While the degree of repression of Fth1 and Ftl mRNAs in rat liver under control conditions was similar to our previous study in mice (24) both Fpn1 and Aco2 mRNAs were more highly repressed in rat liver, suggesting a relatively iron-deficient state compared to mice.

Table 1.

Selective Translational Regulation of 5′ IRE-Containing mRNA in Adult Rat Liver

		Percent Total mRNA in Fraction1
mRNA	Treatment	RNP	40S/60S/80S	Polysomes
Tth	AC	9.9 ± 2.6	6.3 ± 1.2	83.8 ± 3.2
	FAC	5.8 ± 2.2	6.3 ± 0.9	87.9 ± 1.8
Cs	AC	11.2 ± 1.8	13.7 ± 0.6	75.1 ± 2.0
	FAC	6.3 ± 2.8	17.9 ± 1.9	75.8 ± 1.7
Ftl	AC	75.7 ± 4.9a	12.3 ± 4.3 a	12.0 ± 1.6a
	FAC	12.5 ± 2.8b	5.1 ± 1.1 b	82.4 ± 2.8b
Fth1	AC	79.7 ± 4.0 a	8.1 ± 4.0 a	12.2 ± 1.2 a
	FAC	15.0 ± 2.6 b	4.2 ± 1.3 b	80.8 ± 2.8 b
Fpn1	AC	72.9 ± 1.7 a	9.4 ± 0.3 a	17.8 ± 1.4a a
	FAC	12.3 ± 1.8 b	4.5 ± 0.6 b	86.2 ± 1.8 b
Aco2	AC	23.2 ± 5.5a	14.5 ± 1.1a	62.3 ± 4.4a
	FAC	2.9 ± 1.6b	13.3 ± 1.0b	83.9 ± 2.6b

¹The RNA content of gradient fractions was determined by real time RT PCR as described in Methods. Different superscripts between rows denote statistically significant differences between control rats treated with ammonium citrate (AC) or rats acutely iron-overloaded with ferric ammonium citrate (FAC) for individual mRNA, p < 0.05. Values are expressed as mean ± SEM (n = 3).

²Rats were injected with ammonium citrate as a control (AC) or with ferric ammonium citrate to induce iron overload (see Methods). Samples were analyzed 3 hr after injection.

To determine if the difference in translation state of 5′ IRE-containing mRNA is due to selective repression by IRPs, the extent to which acute iron overload caused translational derepression was examined. Treatment with ferric ammonium citrate (Iron) for 3 hr reduced IRP RNA binding activity (IRP1 plus IRP2) from 38 ± 2 fmol/mg protein to 24 ± 4 fmol/mg protein (P < 0.05) as measured by quantitative EMSA (see Supplementary material) (19). Acute iron overload strongly activated the translation of ferritin mRNAs as 80% of Fth1 and Ftl became polysome-associated, nearly a 7-fold increase relative to control (Fig. 1E and 1F). A similar response was observed for Fpn1 mRNA where more than 85% of this RNA was found in the polysome region after iron treatment (Fig. 1G). Interestingly, even in the case of Aco2 mRNA, iron significantly enhanced polysome association such that 84% of the messenger was polysome-bound, indicating that the relatively small fraction (~20%) m-acon mRNA found in the RNP region in control liver is being actively repressed by IRPs (Fig. 1H). In contrast to the IRE-containing mRNA, acute iron overload was without effect on the translation state of Tth and Cs mRNA which supports the conclusion that the impact of iron on mRNA translation is specific to 5’ IRE-containing mRNAs and that global changes in ribosome distribution are not occurring (Fig. 1C and 1D, Table 1). It is notable that even though iron substantially stimulated the translation of the IRE-containing mRNA examined, a small fraction of the population of each mRNA remained in the RNP pool. The fractional amount of RNP-associated mRNA in iron overloaded rats was highest for the Fth1 and Ftl mRNAs and lowest for Aco2 mRNA (Table 1). In summary, in adult rats fed a normal diet, 5’ IRE-containing mRNAs encoding the iron metabolism proteins FTH1 and FTL and FPN1 were highly repressed such that acute iron overload substantially increased their polysome association, while the tricarboxylic acid cycle enzyme Aco2 was largely polysome bound in control liver and consequently responded minimally to iron stimulation.

Differential impact of iron deficiency on repression of 5′-IRE mRNA:

To investigate further the selective control of 5′ IRE mRNA translation, the physiological response to dietary iron deficiency was determined. Since hypoxia can influence IRP RNA binding activity (32–35) we used a feeding regimen that produced iron-deficiency without anemia. Weanling rats (d 21) were fed an iron sufficient diet for two weeks before dividing them into a control or iron deficient group fed the C or ID diet, respectively, for two additional weeks. On the day of the experiment blood hemoglobin level was not significantly different between the two dietary groups being 13.2 ± 0.9 g hemoglobin/dl in rats fed the iron sufficient control (C) diet and 11.1 ± 0.6 g hemoglobin/dl in rats fed the iron-deficient (ID) diet (P > 0.05). In contrast, the percent saturation of serum transferrin with iron was strongly reduced (P < 0.0006) in the ID rats (9.0 ± 1.7 %) versus the control rats (35.3 ± 1.8 %). Liver IRP RNA binding activity (IRP1 plus IRP2) determined by EMSA (not shown) was 1.5-fold higher (P < 0.05) in the ID group (124 ± 6 fmol/mg protein) compared to the control group (83 ± 7 fmol/mg protein) (see Supplementary material). Thus, this experimental model allowed examination of the impact of iron deficiency without anemia on the translation state of 5′-IRE mRNA in liver.

For the animals fed the control iron-adequate diet used in this study the relative translational activity of 5′-IRE mRNA exhibited the same differential pattern of repression as noted in a previous study in mice (24) and is consistent with the range of binding affinities previously determined for all 5’ IREs (19). Ftl mRNA was most extensively repressed followed by Fpn1 and then Aco2 mRNAs (Fig. 2C through 2E; Table 2). Iron deficiency resulted in a further repression of the translation of each of these mRNA. Polysomal-associated Ftl mRNA was most strongly affected exhibiting a 65% decrease in ID relative to C liver such that only 4% of Ftl mRNA appeared to be translationally active in iron deficient liver (Fig. 2C, Table 2). We also observed a 70% reduction in the total amount of Ftl mRNA in liver of the ID rats (Table 2). Iron is known to regulate Ftl gene transcription (22).

figure 2:

impact of iron deficiency without anemia on the translation state of 5′ ire-containing mrna in liver. weanling male sprague-dawley rats were fed an iron-adequate control (c) diet for two weeks (see methods). they were then switched to either an iron-deficient (id) diet (< 2 ppm iron) or they remained in the control group that received the iron adequate diet for an additional 2 weeks. polysome profiles of rat liver were generated as described in methods. a.- e. results of qpcr showing the distribution of the following mrnas across the gradients: a. transthyretin (tth); b. citrate synthase (cs); c. l-ferritin (l-ftn); d. ferroportin (fpn); and e. mitochondrial aconitase (m-acon). results are expressed as mean ± standard error of the mean for n = 3 rats per group. an asterisk indicates significantly different from control, p < 0.05. see table 2 for quantification of results.

Table 2.

Differential Impact of Iron Deficiency on Translational Repression of 5′ IRE-Containing mRNA in Adult Rat Liver

			Percent Total Gradient mRNA in Fraction1
mRNA	Diet2	Total mRNA3	RNP	40S/60S/80S	Polysomes
Tth	C	492 ± 62a	15.1 ± 4.7	6.9 ± 0.8	78.0 ± 5.3
	ID	153 ± 34b	10.1 ± 2.4	6.7 ± 1.2	83.1 ± 2.8
Cs	C	328 ± 20a	11.1 ± 2.4	17.7 ± 1.6	71.3 ± 2.1
	ID	140 ± 32b	12.1 ± 1.4	21.6 ± 1.2	66.3 ± 1.6
Ftl	C	419 ± 48a	75.2 ± 5.2	12.1 ± 1.3	12.7 ± 4.2c
	ID	131 ± 27b	86.7 ± 3.0	8.9 ± 2.7	4.4 ± 0.5d
Fpn1	C	349 ± 29a	57.3 ± 6.0	11.4 ± 1.7	31.6 ± 4.6 c
	ID	237 ± 37b	65.0 ± 7.5	18.9 ± 5.1	16.5 ± 2.7 d
Aco2	C	156 ± 18a	14.5 ± 5.1c	11.8 ± 2.3	73.8 ± 7.4c
	ID	122 ± 11a	42.6 ± 4.8d	20.2 ± 4.0	37.2 ± 3.8d

¹The RNA content of gradient fractions was determined by qPCR as described in Methods. Different superscripts between rows denote statistically significant differences between control (C) and iron deficient (ID) rats for individual mRNA, p < 0.05. Values are expressed as mean ± SEM for n = 3 rats for the control group and n=4 rats for the iron-deficient group. TfR1 mRNA concentration increased by 2.8 fold in iron-deficient (766 ± 169) vs. control (271 ± 38) liver (p < 0.05).

²Rats were fed the control (C) diet for two weeks to increase their iron stores. Then they were fed either the C or iron-deficient (ID) diets for an additional 13 or 14 days (see Methods).

³Total mRNA content in liver was determined by real time RT PCR as described in Methods. Results were normalized to the level of expression of 18S rRNA in each sample. RNA concentration is on a relative basis and cannot be compared across mRNA (see Methods).

We also observed enhanced translational repression of other 5′-IRE mRNA albeit not to the extent to which Ftl mRNA responded. The relative translation rates of Fpn1 and Aco2 mRNA were more highly repressed in ID liver but the relative decline in their polysome association (~50%) was not as extensive as was the case for Ftl mRNA (Fig. 2). In iron-deficient liver, nearly 40% of Aco2 mRNA remained polysome-associated, and this was more than 8-fold greater, on a fractional basis, than what was observed for Ftl (Fig. 2, compare panels C and E, Table 2). In the case of Fpn1, nearly 4-fold more mRNA (17%), on a fractional basis, remained polysome-associated as compared to Ftl (Fig. 2, compare panels C and D). As noted in the first study the translational activity of Tth and Cs mRNAs was high, as 70 to 80% of these mRNA were polysome-associated (Fig. 2A and 2B, Table 2). While a significant decline in the total amount of these mRNA was noted in ID liver their relative abundance on polysomes, and hence their translational efficiency, was not affected by iron deficiency (Table 2).

The abundance of proteins encoded by mRNAs that are strongly (FTL) or weakly (ACO2) controlled by IRPs was differentially affected by iron deficiency. At the termination of this experiment the level of ACO2 or total ferritin (FTL and FTH1) protein in rats fed the control diet was not different compared to control rats killed on day zero (results not shown). However, total ferritin protein declined by 81% (Fig. 3) for rats fed the ID diet. In contrast, ACO2 protein level declined by 21% (Fig. 3). The more extensive decline in ferritin (FTL and FTH1) protein in iron-deficient liver reflects the stronger repression of its mRNA in response to dietary iron deficiency compared to other targets of IRP action.

figure 3:

differential effect of iron deficiency on the abundance of liver proteins encoded by 5’ ire-containing mrna. liver homogenates from the control and iron deficient rats described in figure 2 were subjected to immunoblotting to determine the abundance of m-acon and ferritin protein as described in methods. a. results as quantified by densitometry. b. representative immunoblots. the sds-page conditions used did not permit separation of the h- and l-ferritin subunits. results are expressed as mean ± standard error of the mean for n = 4 rats per group. an asterisk indicates significantly different from control, p < 0.05.

Kinetics of IRP1:IRE dissociation contribute to differential translational regulation:

The differential translational regulation of the Ftl, Fpn1 and Aco2 mRNAs fits well with our previous finding demonstrating that IRP1 binds to 5′-IRE in an RNA binding affinity hierarchy (19). To better understand the mechanistic basis for the selective translational regulation of 5′-IRE mRNA we determined the dissociation rate (k_off) for IRP1 when bound to Ftl or Aco2 IRE. These IRE were chosen because they represent strongly versus weakly repressed mRNA and they display the largest (9-fold) difference in affinity for IRP1 (19,36). IRP1:[³²P]IRE complexes were allowed to form and the extent of RNA dissociation determined by gel shift or by nitrocellulose filter binding assay after addition of a 200-fold molar excess of competitor RNA. For each complex the specific competitor was an unlabeled version of the same IRE while the non-specific competitor was the antisense version of the Ftl IRE.

As determined by EMSA the amount of Ftl IRE- and Aco2 IRE-IRP1 complexes were stable in the absence of any RNA competitor (not shown) or in the presence of the non-specific competitor (Fig. 4C and 4F). However, when challenged with the unlabeled Ftl IRE, the [³²P]Ftl IRE-IRP1 complex dissociated with k_-1 = 0.019 ± 0.0029 at 30° and a t_1/2 = 42.7 ± 7.9 min (mean ± SEM; n = 7) (Fig. 4A and 4C, Table 3). In contrast, the Aco2 IRE:IRP1 complex dissociated significantly more rapidly. When challenged with unlabeled Aco2 IRE, the [³²P]Aco2 IRE dissociated with k_-1 = 0.13 ± 0.016 at 30° or a t_1/2 = 5.7 ± 0.8 min (mean ± SEM; n = 6) (Fig. 4D and F; Table 3). The off-rate for the Aco2 IRE from IRP1 was 7.5-fold more rapid than was observed for the Ftl IRE. We also determined the off-rate for the Aco2 and Ftl by nitrocellulose filter binding assay. A similar result was obtained as Aco2 was found to dissociate nearly 6-fold more rapidly from IRP1 than was the case for the Ftl IRE (Table 3).

figure 4:

kinetics of dissociation of ferritin and m-acon ire from irp1. recombinant irp1 was incubated with either [32p]l-ferritin ire (a.-c.) or the [32p]m-acon ire (d.-f.) to form an rna protein complex at 30 °c. after 10 min a 200-fold excess of unlabeled cognate (filled symbols) or non-cognate (open symbols) rna was added to the reaction mixture. at indicated time points, samples were mixed with heparin and loaded on an emsa gel at 4°c. images of emsa gels are shown in a.-b., d.-e., with lane numbers indicated below each gel image.: a. emsa with labeled l-ferritin ire with unlabeled l-ft competitor rna, b. emsa with labeled l-ft ire with unlabeled non-cognate rna; d. emsa with labeled m-acon ire with unlabeled m-acon competitor rna; e. emsa with labeled m-acon ire with unlabeled non-cognate rna. quantified emsa results are shown in c. and f. for l-ft and m-acon, respectively. see table 3 for calculated dissociation rate constants and half-lives, and for results of similar experiments done by filter binding assays.

Table 3.

Kinetics of Dissociation of IREs from IRP1^a

	EMSA		Filter Binding
RNA	t1/2 (min)	k−1 (min−1)	t1/2 (min)	k−1 (min−1)
Ftl	42.7 ± 7.9	0.019 ± 0.0029	92.1 ± 10.2	0.0083 ± 0.0013
Aco2	5.7 ± 0.8b	0.13 ± 0.016c	16.7 ± 3.5d	0.070 ± 0.020e

^aFor the EMSA and filter binding analyses of L-ferritin seven separate experiments were performed. For m-acon six separate EMSA and eleven separate filter binding experiments were used. Results are reported as mean ± SEM.

^bSignificantly different versus L-ferritin (P = 0.0013).

^cSignificantly different versus L-ferritin (P = 0.0001).

^dSignificantly different versus L-ferritin (P = 0.0077).

^eSignificantly different versus L-ferritin (P = 0.0272).

Iron-dependent control of ferritin protein level in the absence of translational control:

It is well established that FPN1 and the FTH1 and FTL subunits, all encoded by 5’ IRE-containing mRNA, are also strongly iron-regulated through targeted protein degradation mechanisms (37–39). In the case of ferritin, NCOA4 targets ferritin shells to the lysosome for degradation in the presence of iron (40). Our findings described below support the concept that iron-dependent control of ferritin protein stability provides an additional level of control that acts when translational repression is maximized. Weanling rats (d 21) were fed the ID or C diet for one week. Rats fed the ID diet were iron-deficient and anemic as indicated by their reduced blood hemoglobin concentration (7.7 ± 0.1 g/dL) relative to that observed in control rats (11.1 ± 0.3 g/dL) (P < 0.05). Similar to what was observed in adults fed the control diet, only 10% of Ftl mRNA was polysome-associated in liver, and more than 70% of Aco2 mRNA was in this fraction in weanling control rats (Fig. 5, Table 4). However, a different picture emerged in response to iron deficiency compared to what was observed in adult rats. Surprisingly the translation state of Ftl mRNA was not further repressed by iron deficiency (Fig. 5, Table 4) even though IRP RNA binding activity (IRP1 plus IRP2) increased from 80 ± 7 fmol/mg protein to 211 ± 12 fmol/mg protein (P < 0.05) (see Supplementary material). In contrast, translation of Aco2 mRNAs was significantly repressed in iron-deficiency compared to liver from rats fed the control diet (Fig. 5E, Table 4). Iron-deficiency resulted in a 2.5-fold increase of Aco2 mRNA in the RNP fraction, and a 37% decrease in the polysome fraction was observed relative to what was observed in liver of rats fed the iron sufficient diet (Fig. 5E, Table 4). Fpn1 mRNA appeared to respond similarly to Aco2 mRNA as the fraction of Fpn1 mRNA that was polysome bound decreased from 29% to 17% (Fig. 5D, Table 4; n = 2). As was observed for adult rats, the translation state of Tth and Cs mRNA was unaffected by iron deficiency (Fig. 5A and 5B). Taken together, these results suggest that Ftl mRNA translation is maximally repressed in the liver of weanling rats fed an iron-adequate diet. In contrast, Fpn1 and Aco2 mRNA which are more weakly repressed versus Ftl and Fth1 mRNAs in rats fed a control diet, remained strongly responsive to iron deficiency even in weanling animals

figure 5:

iron regulation of ferritin protein abundance in the absence of translational regulation. weanling male sprague dawley rats were fed an iron-deficient (id) diet (< 2 mg fe/kg diet) or iron adequate (50 mg fe/kg diet) for 7 to 9 days. polysome profile analysis was performed as described in figure 1. a.- e. results of qpcr showing the distribution of the following mrnas across the gradients: a. transthyretin (tth); b. citrate synthase (cs); c. l-ferritin (l-ftn); d. ferroportin (fpn); and e. mitochondrial aconitase (m-acon). results are expressed as mean ± standard error of the mean for n = 3 rats per group. an asterisk indicates significantly different from control, p < 0.05.

Table 4.

Ferritin mRNA Translation is Fully Repressed Irrespective of Iron Status in Weanling Rat Liver

			Percent Total Gradient mRNA in Fraction1
mRNA	Diet2	Total mRNA3	RNP	40S/60S/80S	Polysomes
Tth	C	320 ± 97	4.7 ± 0.5	5.6 ± 0.9	89.7 ± 1.3
	ID	278 ± 42	4.4 ± 1.1	4.3 ± 1.3	91.3 ± 2.2
Cs	C	350 ± 113	7.5 ± 0.9	13.4 ± 2.4	79.1 ± 3.2
	ID	291 ± 44	9.8 ± 2.6	12.2 ± 2.4	77.9 ± 4.9
Ftl	C	240 ± 79	77.9 ± 1.7	12.0 ± 1.6	10.1 ± 2.6
	ID	199 ± 10	76.7 ± 5.1	15.4 ± 3.6	8.0 ± 1.9
Fpn1	C	515 ± 165	60.7	10.8	28.5
	ID	460 ± 90	63.8	20.8	17.1
Aco2	C	214 ± 60	12.4 ± 3.2a	15.6 ± 1.9a	72.0 ± 4.8a
	ID	265 ± 62	30.9 ± 1.9b	23.5 ± 1.7b	45.6 ± 3.5b

¹The RNA content of gradient fractions was determined by real time RT PCR as described in Methods. Different superscripts between rows denote statistically significant differences between control (C) and iron deficient (ID) rats for individual mRNA, p < 0.05. Values are expressed as mean ± SEM for n = 3 rats. The concentration of Tfrc mRNA increased by 6.2-fold in iron-deficient (1161 ± 162) vs control (188 ± 21) liver (p < 0.05).

²Rats were fed the control (C) or iron-deficient (ID) diets for seven to nine days (see Methods).

³Total mRNA content in liver was determined by real time RT PCR as described in Methods. Results were normalized to the level of expression of 18S rRNA in each sample. RNA concentration is on a relative basis and cannot be compared across mRNA (see Methods).

⁴For this study one sample of the C and one from the ID failed during polysome analysis so the average of n = 2 samples/condition is reported. RNA analyses had n = 3.

We then asked if the different response of Ftl mRNA translation to iron deficiency in weanling rats relative to the adult rats used in our previous study was reflected at the protein level (Fig. 6). The 50% decline in ACO2 protein level in liver of weanling ID rats agreed well with previous observations (29) and is in line with the increased translational repression of this mRNA. Interestingly, ferritin protein (FTL and FTH1) level declined by 75% in the liver of ID weanling rats even though the translation state of Ftl mRNA was not altered relative to that observed in the liver of weanling rats fed an iron sufficient diet. Thus, substantial modulation of ferritin protein accumulation can occur in the absence of translational control of ferritin mRNA and this likely involves regulated changes in ferritin protein stability.

figure 6:

ferritin protein accumulation is iron regulated in the absence of translational regulation in weanling rat liver. liver homogenates from the control and iron deficient rats described in figure 5 were subjected to immunoblotting to determine the abundance of m-acon and ferritin protein as described in methods. a. results as quantified by densitometry. b. representative immunoblots. the sds-page conditions used did not permit separation of the h- and l-ferritin subunits. results are expressed as mean ± standard error of the mean for n = 4 rats per group. an asterisk indicates significantly different from control, p < 0.05.

DISCUSSION

Our analyses of the impact of dietary iron deficiency and acute iron overload on a direct action of IRPs, the translational control of 5′ IRE-containing mRNA, support novel conclusions concerning how mammalians cells coordinate the modulation of cellular iron metabolism with control of pathways involved in the adaptive response to iron deficiency (i.e. citrate metabolism). First, the different set-points for translational control of 5’ IRE-containing mRNA observed here supports the concept that IRP action occurs over a regulatory continuum that ranges from weakest (Aco2) to strongest (Fth1 and Ftl) targets. That IRPs are significant determinants of this hierarchy is supported by the range of affinities of IRP1 for natural 5’ IREs which reflects the translational hierarchy observed here and the relationship between IRE mutations in HHCS and the impact of IRP1 and IRP2 binding affinities (16,19). Second, the hierarchical control of IRE-containing mRNA generates a regulatory landscape where some targets of IRP action respond more strongly to iron excess while others respond preferentially to iron deficiency. Our study provides compelling evidence that the mRNA regulon controlled by IRPs is wired in a manner that allows separate but overlapping control of cellular iron metabolism and cellular response to iron deficiency in order to optimize the response to the continuum of iron availability from deficient to excessive.

Coordinate control of mRNA fate through the control of RNA regulons has central roles in cell proliferation, inflammatory responses, cancer and an array of other critical cellular and organismal processes (1,3,4). Our work comparing the iron-dependent control of the fate of 5’ IRE-containing mRNAs supports the concept of separate regulatory groups within this RNA regulon. At the translational level a first-line defense against iron-induced oxidative stress is exhibited by the FTH1 and FTL subunits and FPN1. These three mRNAs are most strongly translationally repressed in iron-sufficient cells and consequently respond most robustly to iron excess through a substantial increase in translation. Thus, coordinated high-level production of these proteins when iron levels rise lowers the risk of iron-dependent cellular damage by sequestering iron in the assembled ferritin shell and exporting iron via FPN1 which supports the concept that this regulatory group of mRNAs subject to IRP action is one of cellular defense. We suggest that a second regulatory group of mRNA subject to IRP-dependent control of translation involves mRNAs that are weakly targeted including m-acon as analyzed here but also ALAS2 and HIF-2α as examined previously (24,26). These mRNAs encode proteins essential for cell function when iron levels are optimal and whose function need not be strongly increased in iron overload. Instead, these mRNAs encode proteins whose function can be deleterious in iron deficiency. In the case of m-acon, it is well known that iron deficiency leads to impaired expression of iron-containing proteins essential for oxidative phosphorylation and this may be associated with increased formation of oxygen radicals (41). IRP-dependent suppression of m-acon in iron deficiency may limit TCA cycle flux thus reducing the level of oxygen radicals as is the case regarding the control of aconitase activity in E. coli (42). In the case of ALAS2, suppression of its activity in iron deficiency prevents accumulation of the heme precursor protoporphyrin IX which can be toxic as it is in X-linked dominant protoporphyria in humans or the mild erythropoietic protoporphyria seen in Irp2^−/− mice (43,44). Lastly, the transcription factor HIF-2α is the primary driver of erythropoietin (Epo) production and other genes critical for the adaptive response to hypoxia. IRP1 deficiency leads to inappropriate levels of Epo and polycythemia (24,45,46). Iron deficiency itself induces a block in erythropoiesis (47) and the failure to suppress Epo production via IRP1-mediated regulation may dysregulate erythropoiesis.

To gain additional insight concerning the mechanism through which 5’ IRE-containing mRNA are differentially controlled we determined the dissociation rates of IRP1 with IREs at the extremes of the RNA binding hierarchy, Ftl1 and Aco2. The substantially faster off-rate of IRP1 with the Aco2 IRE relative to the Ftl1 IRE suggests increased accessibility of Aco2 mRNA to the translational initiation machinery as IRPs are predicted to spend less time associated with Aco2 mRNA. Presumably other 5’ IREs that bind weakly to IRPs relative to ferritin mRNA possess a similar translational advantage although differential interaction of these mRNA with translation initiation factors must also be considered (36,48,49). The faster off-rate for the Aco2 IRE is consistent with the enhanced translational activity of this mRNA observed in liver under steady state conditions and the reduced response of this messenger to iron deficiency. The faster off-rate predicts a more rapid kinetics of translational activation of Aco2 mRNA in response to agents that inhibit IRP RNA binding activity. While our previous studies in IRP-deficient mice suggested a preferential role of IRP2 in controlling 5’IRE-containing mRNA other than HIF-2α, we note that the relative proportion of mRNAs remaining in the repressed (RNP) pool in Irp2^−/− mice was greatest for the ferritin subunits followed by Fpn1 and Aco2 mRNA (24). A logical interpretation of these previous results is that the hierarchical differences in K_D and the associated difference in off-rate of IRP1 for 5’ IREs is a key determinant of the differentially repression of mRNA in both wildtype as well as Irp2^−/− mice.

When comparing our studies of the high affinity (picomolar K_D) complex of IRP1 bound to IRE, the fold-difference in K_D across the IRE RNA binding hierarchy of 9-fold (19) exceeds the 6- to 7-fold range in off-rate we report here when comparing L-ferritin and m-acon IRE. The non-equivalence in these values suggests that the association rate of IRP1 in forming the high affinity complex is also different for specific IRE. Previous studies using a fluorescence anisotropy assay to study a lower affinity (nanomolar K_D) IRP1-IRE complex observed a significant impact of IRE species on the rate of association with IRP1 (48–50). No difference in the off-rate for this low affinity complex was observed (48). Taken together, these studies may suggest that IRP1 binds to IREs to form an initial low affinity unstable complex where the rate of association is a key factor distinguishing different IRE. We propose that then there is transition to the higher affinity (picomolar K_D) complexes that differ in their inherent stability.

Since the discovery of translational control of ferritin expression in the 1960s much effort has focused on unraveling the mechanism and determining the extent to which it represented a commonly used paradigm of gene regulation (51,52). However, it is less well recognized that the original studies on iron regulation of ferritin protein accumulation by Drysdale and Munro demonstrated a substantial and comparable impact of acute iron overload on both the synthesis and degradation of ferritin in rat liver (37). These and subsequent studies revealed that ferritin protein is unstable under iron-deficient conditions which recently has been shown to involve the lysosomal targeting protein NCOA4 (39,53). Our studies reveal a developmental regulation of the relative roles of translational and post-translational mechanisms in dictating the steady state expression of ferritin. In adult rats we observed the classic pattern for control of ferritin expression with a 3-fold reduction in translational activity of Fth1 and Ftl mRNAs that accompanied the 5-fold impact on ferritin protein level when comparing liver of rats fed the iron deficient versus the iron sufficient diet. In contrast, in weanling rats a substantial 4-fold change in ferritin protein abundance in liver without any change in mRNA translation state was observed in response to dietary iron deficiency. We conclude that translational control is not an obligatory mechanism for iron-dependent control of ferritin expression. Taken together with the key role of hepcidin-dependent control of FPN1 protein degradation (38) it is clear that coordinated modulation of mRNA translation and protein stability act in concert to modulate the ultimate level of expression of proteins encoded by IRE-containing mRNA.

Studies of IRP deficiency established tissue-specific roles of each IRP in critical physiological processes including key aspects of brain function, erythropoiesis, mitochondrial function and intestinal iron absorption (43,45,46,54–60). A more refined understanding of the regulatory breadth of the IRE-IRP regulon is needed in order to fully define its roles not only in normal physiology but also in diseases where iron dysregulation contributes to pathology. Contributors to the phenotypic consequences of IRP dysregulation likely include differences in the tissue-specific expression level of IRE-containing mRNA, the signaling pathways controlling IRP1 versus IRP2 RNA binding activity, and their affinity for natural or mutant IREs (3,16,19,61). Our finding that 5’ IRE-containing mRNAs are hierarchically controlled at the translational level to changes in IRP RNA binding activity, as induced in our study by changes in iron status, predicts that the specific aspects of iron metabolism contributing to disease etiology will depend on whether IRP action is diminished or enhanced. Thus, loss of IRE RNA binding activity likely induces an iron-deficiency phenotype due to enhanced iron storage and export. In contrast, activation of IRPs would be associated with maladaptive increases in cellular iron accumulation with insufficient storage, impairment of the hypoxia response or impaired energy metabolism. Given the evidence that both canonical and putative non-canonical IREs are found in mRNAs encoding proteins without clear roles in iron metabolism additional regulatory groups of mRNA controlled by IRP action are yet to be elucidated (9,14,62,63).

CONCLUSIONS:

Hierarchical translational control of 5’ IRE-containing mRNA involves a graded response to changes in animal cell iron status. mRNAs encoding iron storage or export proteins are the most tightly controlled by IRPs. In liver of rats fed an iron sufficient diet these mRNAs encoding proteins required for iron sequestration or export are largely translationally repressed and consequently respond most robustly to excessive levels of iron. In contrast, mRNAs encoding proteins involved in cellular iron utilization or the adaptive responses to iron deficiency are actively translated in liver of rats fed an iron sufficient diet, show only a small response to iron overload, and are most strongly affected by iron deficiency. Differences in the iron-dependent setpoint modulating these two phases of translational regulation is central to the proper maintenance of cellular iron metabolism and viability.