Thyroid hormone modulates the interaction between iron regulatory proteins and the ferritin mRNA iron-responsive element.

The cytoplasmic iron regulatory protein (IRP) modulates iron homeostasis by binding to iron-responsive elements (IREs) in the transferrin receptor and ferritin mRNAs to coordinately regulate transferrin receptor mRNA stability and ferritin mRNA translational efficiency, respectively. These studies demonstrate that thyroid hormone (T) can modulate the binding activity of the IRP to an IRE in vitro and in vivo. T augmented an iron-induced reduction in IRP binding activity to a ferritin IRE in RNA electrophoretic mobility shift assays using cytoplasmic extracts from human liver hepatoma (HepG2) cells. Hepatic IRP binding to the ferritin IRE also diminished after in vivo administration of T with iron to rats. In transient transfection studies using HepG2 cells and a human ferritin IRE-chloramphenicol acetyltransferase (H-IRE-CAT) construct, T augmented an iron-induced increase in CAT activity by 45%. RNase protection analysis showed that this increase in CAT activity was not due to a change in the steady state level of CAT mRNA. Nuclear T-receptors may be necessary for this T-induced response, because the effect could not be reproduced by the addition of T directly to cytoplasmic extracts and was absent in CV-1 cells which lack T-receptors. We conclude that T can functionally regulate the IRE binding activity of the IRP. These observations provide evidence of a novel mechanism for T to up-regulate hepatic ferritin expression, which may in part contribute to the elevated serum ferritin levels seen in hyperthyroidism.

The cytoplasmic iron regulatory protein (IRP) modulates iron homeostasis by binding to iron-responsive elements (IREs) in the transferrin receptor and ferritin mRNAs to coordinately regulate transferrin receptor mRNA stability and ferritin mRNA translational efficiency, respectively. These studies demonstrate that thyroid hormone (T 3

) can modulate the binding activity of the IRP to an IRE in vitro and in vivo. T 3 augmented an iron-induced reduction in IRP binding activity to a ferritin IRE in RNA electrophoretic mobility shift assays using cytoplasmic extracts from human liver hepatoma (HepG2) cells. Hepatic IRP binding to the ferritin IRE also diminished after in vivo administration of T 3 with iron to rats. In transient transfection studies using HepG2 cells and a human ferritin IRE-chloramphenicol acetyltransferase (H-IRE-CAT) construct, T 3 augmented an iron-induced increase in CAT activity by ϳ45%.
RNase protection analysis showed that this increase in CAT activity was not due to a change in the steady state level of CAT mRNA. Nuclear T 3 -receptors may be necessary for this T 3 -induced response, because the effect could not be reproduced by the addition of T 3 directly to cytoplasmic extracts and was absent in CV-1 cells which lack T 3 -receptors. We conclude that T 3 can functionally regulate the IRE binding activity of the IRP. These observations provide evidence of a novel mechanism for T 3 to up-regulate hepatic ferritin expression, which may in part contribute to the elevated serum ferritin levels seen in hyperthyroidism.
The iron regulatory protein (IRP, 1 previously known as the iron-responsive element-binding protein, IRE-BP, and iron responsive factor, IRF) is a trans-acting RNA-binding protein which binds with high affinity to conserved stem-loop structures, iron-responsive elements (IREs), present in the ferritin, transferrin receptor (TfR), and erythroid 5-aminolevulinate synthase mRNAs (1)(2)(3). The IRP serves a central role in the regulation of iron (Fe) homeostasis (1). In the absence of iron, the IRP binds to the IRE in the 5Ј-untranslated region (5Ј-UTR) of ferritin and erythroid 5-aminolevulinate synthase mRNAs and represses translation (4 -6). Binding of the IRP to IREs in the 3Ј-untranslated region (3Ј-UTR) of TfR mRNA stabilizes the mRNA and prevents its degradation (7)(8)(9). In iron-replete states, the reverse holds, which results in increased ferritin translation and decreased TfR mRNA stability. This reciprocal regulation is achieved at the post-translational level and is independent of new protein synthesis (10).
Two IRPs have been defined in various human and rat tissues (3,11,12). The most widely expressed and abundant IRP in human tissues is IRP1 (1,3). A second human IRP (IRP2) has been described recently. IRP2 is 57% identical with IRP1 at the amino acid level and 2-10 times less abundant than IRP1 in most tissues, except in the brain (3). In contrast to IRP1, cellular concentrations of IRP2 are inversely regulated by iron levels due to iron-dependent regulation of the half-life of IRP2 protein (3). The relative contribution of each of these species to iron homeostasis remains to be elucidated. The two rat IRPs have been designated BP1 and BP2 (12) (also known as IRF and IRF B , respectively) (11) and may represent rodent counterparts for IRP1 and IRP2. Significant functional differences exist between BP1 and BP2. In particular, BP2 does not have functional aconitase activity and, in contrast to BP1, levels of BP2 protein are regulated by iron (12). In addition, IRF B (and presumably BP2) is expressed most abundantly in intestine, brain, and kidney (11).
Thyroid hormone (T 3 ) plays a central role in differentiation, development, and maintenance of body homeostasis (13). The actions of T 3 , like the steroid hormones, are mediated through intracellular T 3 -receptor proteins (TRs) (14,15) which act predominantly to modulate transcription by binding to specific T 3 -response elements in target genes (16). Recent studies have demonstrated, however, that T 3 also has important effects at the post-transcriptional level to regulate the expression of several genes, including the ␤-subunit of thyrotropin (TSH␤) (17), the thyrotropin releasing hormone receptor (18), and the retinoid-X receptor (19). To date, there is little understanding of the molecular mechanisms underlying these T 3 -induced changes in mRNA stability.
Several groups have documented an association between T 3 levels and ferritin expression. In earlier reports, hypothyroidism produced by thyroidectomy was associated with increased rat hepatic ferritin content, which was found to be due to post-transcriptional changes in the ferritin synthetic rate (20,21). More recently, however, and in contrast, hyperthyroid rats with elevated T 3 and T 4 levels were found to have an increased liver ferritin protein synthesis rate (38% above control) (22). Part of this increase may be due to elevated IRE-mediated ferritin translation, although T 3 has also been shown to increase the transcription rate of H-ferritin mRNA in rat C6 glioma cells (23), raising the possibility of a transcriptional effect in the liver. Of particular interest, are reports from several groups in which T 3 was shown to positively regulate serum ferritin measurements in humans (24 -28), similar to the changes reported in the rat (22). Elevated serum ferritin levels were observed in hyperthyroid individuals, and levels decreased significantly after antithyroid treatment with normalization of T 3 levels (24 -26, 28). Furthermore, administration of T 3 to hypothyroid individuals produced a significant increase in the serum ferritin level (26,27). Although the cause of the T 3 -induced increase in the serum ferritin level in humans is unknown, increased synthesis of ferritin in the liver may well be an important contributor. These links between T 3 and the regulation of ferritin expression suggest that a positive correlation exists between the levels of T 4 /T 3 and ferritin in the serum. However, the molecular mechanisms involved in the hepatic regulation of ferritin expression by T 3 remain to be determined.
We reasoned that a component of the effect of T 3 on ferritin expression in the liver was due to T 3 -induced modulation of IRP binding to the ferritin IRE. Therefore, we used 1) the RNA electrophoretic mobility shift assay (REMSA) and 2) transient transfection assays to investigate the regulation of IRE-dependent gene expression by T 3 , in vivo and in vitro. Our results demonstrate that T 3 can functionally regulate the binding activity of the human and rat IRP to a ferritin IRE. These data provide evidence for a role of T 3 in the post-transcriptional regulation of iron-responsive genes and new insights into the action of T 3 in the modulation of iron homeostasis. Furthermore, these data may, in part, explain the positive association between serum levels of T 3 and ferritin.

Preparation of Tissue and Cell Extracts-Frozen
Sprague-Dawley rat liver was homogenized in 5 volumes of ice-cold 10 mM HEPES (pH 7.5), 40 mM KCl, 3 mM MgCl 2 , 1 mM dithiothreitol, 0.32 M sucrose, 0.5 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, 2 g/ml aprotinin (Buffer A) with a Teflon homogenizer. After removal of nuclei by low speed centrifugation, the supernatant was centrifuged at 100,000 ϫ g for 1 h, and the supernatant was frozen rapidly in dry ice and stored at Ϫ70°C. Some animals received a single intraperitoneal injection of either T 3 (20 g/100 g body weight), ferric ammonium citrate (FAC) (2.3 mg/100 g body weight), or a combination of both, 8 h prior to preparing the extract. Cultured cells (HepG2, human liver carcinoma; CV-1, monkey kidney) were grown in Dulbecco's modified Eagle's medium and 10% fetal bovine serum supplemented with penicillin and streptomycin at 37°C in 5% CO 2 . Forty eight h before harvesting, the cells were washed with phosphate-buffered saline, and incubated with T 3 -deficient hypothyroid medium. In the 16 -24 h prior to harvesting, cells were grown for various times in the absence or presence of combinations of FAC (110 M), desferrioxamine (Df) (100 mg/ml, Desferal, Ciba Geigy), and T 3 (either 1 or 100 nM, Sigma). Cells were harvested by lysis in the same buffer as above plus 5% glycerol and 0.2% Nonidet P-40, but without the sucrose. After removal of the nuclei, the cells were centrifuged at 10,000 ϫ g, and the supernatant was stored in aliquots at Ϫ70°C. T 3 was removed from fetal bovine serum for use in hypothyroid medium as described (29). Protein concentrations were determined in duplicate by the Bradford method (Pierce).
RNA Electrophoretic Mobility Shift Assay (REMSA)-Binding reactions were performed as described (30, 33) with 5-10 g of cytoplasmic extract protein and 10 4 -10 5 cpm of RNA (ϳ0.5 ng) in Buffer A in a total volume of 15-25 l. Following incubation at 25°C for 30 min, 1 unit of RNase T1 (Boehringer Mannheim) was added for 10 min, followed by the addition of heparin (final concentration, 5 mg/ml) (Sigma) for an additional 10 min. Samples were subjected to electrophoresis on 4% native polyacrylamide gels (acrylamide/methylene bisacrylamide ratio, 70:1), which were dried and analyzed by autoradiography at Ϫ70°C. In some assays, extracts were preincubated for 10 min at 25°C with nonspecific and specific competitor sense unlabeled RNA (100-fold excess) before addition of the radiolabeled RNA. A Betagen 603 Blot Analyzer (Betagen Corp., Waltham, MA) was used to quantitate the radioactivity present in each lane.
UV Cross-linking of RNA⅐Protein Complexes-RNA-protein binding reactions were carried out as described above using 30 g of cytoplasmic extract and ϳ0.5 ϫ 10 5 cpm of labeled RNA per reaction. Following the addition of heparin, samples were placed on ice and irradiated for 5-30 min, 4 cm below the UV source in a Statalinker UV light box (Stratagene, 240 nm UV-bulb, La Jolla, CA) (33). After cross-linking, samples were incubated with RNase A (final concentration, 100 g/ml) (Boehringer Mannheim) for 15 min at 37°C. The samples were boiled for 3 min in SDS sample buffer, subjected to SDS-polyacrylamide gel electrophoresis (6 -10% SDS-PAGE), and analyzed by autoradiography. 14 C molecular weight markers (Amersham, Buckinghamshire, UK) were used as standards.
Transient Transfection and CAT Assays-HepG2 cells were seeded at 10 5 cells per well in a 4-well plate 48 h before transfection. Twenty four h before transfection, the medium was removed, the cells were washed twice with phosphate-buffered saline, and then incubated in hypothyroid medium. The cells were transfected for 16 h with 10 g of H-IRE-CAT/well and 0.8 g/well of Rous sarcoma virus-␤-galactosidase (␤-gal) control plasmid (34) by calcium phosphate co-precipitation (35). Cells were grown for an additional 24 h in the presence or absence of combinations of FAC (110 M), Df (100 mg/ml), and T 3 (either 1 or 100 nM). Lysates were prepared by harvesting the cells into a buffer containing 125 mM Tris, pH 7.6, 0.5% Triton X-100. Chloramphenicol acetyltransferase (CAT) activity was determined by counting directly the amount of CAT reaction product diffusing into liquid scintillation fluid (36). Each lysate was incubated at 37°C with [ 3 H]acetylcoenzyme A (0.1 Ci) and 2 mM chloramphenicol in 200 l of 100 mM Tris-HCl, pH 7.8, which had been overlaid with 5 ml of liquid scintillant (Econofluor, DuPont NEN). The CAT-catalyzed acetylation of chloramphenicol was monitored by counting the accumulation of tritiated acetylchloramphenicol product and provides an accurate measure of CAT protein levels. ␤-gal activity was determined for each extract as described (34).
RNase Protection Analysis of CAT mRNA-CAT mRNA from HepG2 cell transfectants was characterized by RNase protection. T7 polymerase transcribed a 261-nucleotide 32 P-labeled cRNA from HindIII-digested DNA template isolated from the CAT gene subclone (pBSCAT-1), as described (32). Labeled CAT cRNA (1 ϫ 10 6 cpm) was hybridized to 20 g of HepG2 cell RNA in a buffer containing 80% formamide, 40 mM Pipes (pH 6.7), 0.4 M NaCl, 1 mM EDTA for 16 h at 45°C. Digestion with RNase A (40 g/ml) and RNase T1 (2 g/ml) removed unhybridized cRNAs. Protected cRNAs were separated by electrophoresis on a 6% polyacrylamide gel containing 7 M urea. HaeIII-digested x-174 DNA fragments were kinase-labeled and used as molecular weight markers to confirm the lengths of protected fragments.
Slot-Blot Analysis of CAT mRNA-CAT mRNA levels were quantitated by slot blotting (32). Transfected HepG2 cell RNA (4 -20 g) was denatured for 15 min at 65°C in a buffer containing 6.15 M formaldehyde, 0.15 M sodium citrate, 1.5 M NaCl in duplicate (1ϫ and 5ϫ). The RNA was applied onto slots aligned on nitrocellulose membranes. A 1.64 kilobase HindIII/BamHI fragment coding for the complete CAT gene from pSV2CAT was labeled by random priming and used as a hybridization probe in a standard buffer (35). The filters were stringently washed 3 times at 55°C, and the blot was subjected to autoradiography. Specifically bound CAT gene probe was quantitated using the Betagen Analyzer.

IRP⅐IRE Complex is Species-specific In T 3 -responsive Tis-
sues-To determine if there were qualitative differences between the IRP⅐IRE RNA⅐protein complex (RPC) in the T 3 -responsive tissue and cells selected for use in this study, cytoplasmic extracts from rat liver and human hepatoma (HepG2) cells were used in REMSA with labeled pgem-IRE. Fig. 2A shows that the nature of the IRP⅐IRE complex is species-specific. In HepG2 cells, a single complex was seen (IRP1, lane 1). A second human IRP has recently been identified, IRP2 (3), although it co-migrates with IRP1 in HepG2 cells. With the exception of the brain, there is ϳ2-10 times more IRP1 than IRP2 in human tissues (3). In rat liver (lanes 3-6), two characteristic RPCs, denoted BP1 and BP2 (12,30), were identified, both of which represent specific bona fide IRP⅐IRE complexes. Accumulated evidence from several reports suggests that BP1 and IRP1 are species homologues, and that BP2 and IRP2 are likely rat and human counterparts, respectively (3,11,12).
We then confirmed that the cytoplasmic regulation of IRP binding to the ferritin IRE was preserved in these cells. The IRE binding activity of the IRP is typically increased in vitro by reducing agents, such as 2-mercaptoethanol (2-ME) (37). As predicted, 2-ME (2%) increased the binding of the IRP to IRE RNA with both HepG2 and rat liver cytoplasmic extracts ( Fig.  2A, lanes 2 and 4). Interestingly, the BP2 RPC in rat liver was consistently abolished in the presence of 2-ME ( Fig. 2A, lane 4), and the intensity of the BP1 RPC was increased. A 100-fold excess of an unlabeled specific competitor RNA (pgem-IRE) competed efficiently for binding to the IRP and abolished BP1 and BP2 RPC formation (lane 5). However, addition of excess FIG. 2. The IRP⅐IRE complex is species-and tissue-specific. A, REMSA was performed using 32 P-labeled pgem-IRE and 5 g of different cytoplasmic extracts: HepG2 cells (lanes 1 and 2) and rat liver (lanes 3-6). The binding reactions were incubated for 30 min at room temperature prior to sequential addition of RNase T1 and heparin, as described under "Experimental Procedures." The binding mixtures were analyzed by electrophoresis on a 4% nondenaturing polyacrylamide gel (REMSA). In lanes 2 and 4, 2-ME (2%) was added to the reaction mixture at the beginning of the incubation. A 100-fold excess of specific (unlabeled pgem-IRE, lane 5) or nonspecific (pgem-vec, lane 6) competitor RNA was incubated with the extract for 10 min at 22°C prior to addition of labeled probe. B, UV cross-linking analysis of rat liver IRE⅐IRP complexes. Arrows at BP1 and BP2 denote rat RPCs containing IRPs and IRE (11,12). Following incubation with RNase T1 and heparin, 32 P-labeled RPCs were UV-cross-linked as described under "Experimental Procedures." The complexes were treated with RNase A at 37°C for 15 min. After addition of SDS-sample buffer and boiling, the cross-linked products were separated by 7% SDS-PAGE and analyzed by autoradiography. To confirm the size of the two rat RPCs, BP1 and BP2, a portion of the reaction mixture from Fig. 2A, lane 3, was analyzed after UV cross-linking and digestion with RNase A (100 g/ml) for 30 min at 37°C. Two RPCs were identified which migrated at the positions predicted for the BP1 and BP2 proteins (ϳ90 -95 and ϳ105 kDa, respectively; Fig. 2B) (11,12). We concluded that these cells contained IRPs, BP1 and BP2, which were of the predicted size and which displayed appropriate regulation with reducing agents.
T 3 Modifies IRP⅐IRE Complex in Rat Liver-To address our hypothesis, we first investigated whether T 3 affected the IRE⅐IRP complex in rat liver in vivo by REMSA using cytoplasmic extracts from Spraque-Dawley rats injected intraperitoneally with either vehicle alone (n ϭ 3), T 3 (n ϭ 3), ferric ammonium citrate (FAC) (n ϭ 4), or a combination of both (n ϭ 4), together with labeled pgem-IRE. Administration of T 3 alone had no significant effect on the BP1 and BP2 RPCs (compare lane 1 and 2). Animals injected with FAC alone showed a decrease in the BP1 complex (ϳ30%) and complete loss of the BP2 complex (Fig. 3, compare lane 3 and 1). When FAC and T 3 were injected together, BP1 expression decreased further, and BP2 expression returned (lane 4). There was a reduction in total binding activity when T 3 was combined with FAC (compare lane 3 and 4, ϳ30%, Betagen scan). Although the RPCs are generally more intense in Fig. 3 Fig. 2A due to longer autoradiography exposure time, the relative ratios of BP1 to BP2 are similar throughout. These T 3 -induced changes in IRE binding activity of BP1 and BP2 appear to be a posttranslational event, because in the presence of FAC and T 3 , maximal IRE binding activity was recovered in vitro by treatment with 2-ME (2%) (lane 5). These findings, which are representative of 3 different REMSA experiments, provided in vivo evidence that T 3 could regulate the binding of the IRP to the ferritin IRE and thus modify hepatic ferritin expression.

than in
T 3 Modifies IRE Binding Activity of IRP in Iron-replete HepG2 Cells-The data from the in vivo studies suggested that T 3 was able to modify, either directly or indirectly, the binding activity of the IRP to a ferritin IRE. Furthermore, the in vivo results supported the accumulated data for humans where elevated T 3 levels are associated with increased serum ferritin levels (24 -28). We then examined whether T 3 had an effect on IRP binding activity in vitro using human HepG2 cells. This hepatoma cell line has been used extensively to characterize IRP⅐IRE interactions, and HepG2 cells contain well characterized thyroid hormone receptors (TRs) (38). HepG2 cells were plated into 6-cm dishes, cultured in hypothyroid medium (T 3deficient) (29) for 36 h, and thereafter in the presence or absence of combinations of FAC, desferrioxamine (Df), and T 3 , for various times, prior to harvesting. Cytoplasmic extracts from the cells were mixed with labeled pgem-IRE and analyzed by REMSA. Fig. 4 is a representative REMSA (n ϭ 4) showing that the IRP⅐IRE complex was diminished to ϳ85% of control (Betagen quantitation) after incubation with 110 M FAC for 16 h (lane 2). Although co-culture with 100 nM T 3 for 16 h diminished the complex further to 75% (lane 3), the most significant changes were seen with shorter incubations. When 100 nM T 3 was added to the medium for 4 h, the intensity of the complex was reduced to approximately 60% of control (lane 4), while incubation for only 2 h reduced it to approximately 50% of control (lane 5). This effect was only evident when the HepG2 cells were iron-replete and was similar to that observed in T 3 -treated rat liver extracts described above.
Similar experiments were conducted using HepG2 cells after the addition of Df (100 mg/ml) in the presence and absence of T 3 . However, we were unable to detect any effect of T 3 on the IRP⅐IRE complex in iron-depleted cells using REMSA (data not shown).

T 3 Augments Iron-induced CAT Gene Expression in HepG2 Cell H-IRE-CAT Transfectants-These in vivo and in vitro
REMSA results showed that T 3 was able to reduce IRP binding to a ferritin mRNA IRE. To investigate whether this T 3 -induced effect on IRP binding activity observed in REMSA studies was associated with functional changes in IRE-dependent gene expression, we utilized a transient transfection assay in HepG2 cells with a plasmid containing a human H-ferritin IRE inserted upstream of a CAT reporter gene driven by the Hferritin promoter (H-IRE-CAT). Previous studies with this and similar constructs containing ferritin IREs have demonstrated that changes in CAT activity represent changes in translational efficiency of ferritin mRNA and not changes in the rate of transcription (32,39). Furthermore, no typical T 3 -response elements are present within the sequence of the H-IRE-CAT plasmid.
HepG2 cells, cultured in hypothyroid medium, were transiently transfected with H-IRE-CAT and then incubated in the presence or absence of various combinations of FAC, T 3 , and Df.  4, n ϭ 4). The rats were killed 8 h later, and liver cytoplasmic extracts were prepared and used in REMSA with labeled pgem-IRE as described in Fig. 2. Each lane, with the exception of lane 5, represents an analysis of liver extract from a different rat. A portion of the reaction mixture from lane 4 was incubated separately with 2-ME (2%) prior to PAGE analysis (lane 5). Arrows at BP1 and BP2 denote rat liver cell RPCs containing IRPs and IRE (compare with Fig. 2) .   FIG. 4. T 3 modifies IRP binding to a ferritin IRE in iron-replete human liver HepG2 cells. REMSA (4% PAGE) was performed using HepG2 cell cytoplasmic extracts and 32 P-labeled pgem-IRE. The cells were cultured in hypothyroid medium for 36 h, before FAC was added to the medium 16 h prior to harvesting (110 M, lanes [2][3][4][5]. T 3 (100 nM) was also added at various times (16,4, and 2 h) prior to harvesting. REMSA was performed using 5 g of cytoplasmic extract from each of these different cells and analyzed as described in Fig. 2 2 In the absence of iron loading, no increase in CAT activity was seen after the addition of 100 nM T 3 alone (lane 2). As expected, incubation with 110 M FAC alone increased CAT activity (lane 3). However, co-culture of 100 nM T 3 with iron further increased the CAT activity (lane 4) approximately 35-45%. Even co-culture of 1 nM T 3 with iron increased CAT activity (lane 5), consistent with a dose-response relationship. In contrast, incubation of the cells with Df (100 mg/ml) reduced the CAT activity by ϳ50% from basal (lane 6), reflecting high affinity binding of the IRP to the IRE and reduced translational efficiency. Addition of 100 nM T 3 did not further decrease the CAT activity significantly (lane 7). The 3-fold overall difference in translational efficiency of H-IRE-CAT after stimulation with FAC and Df is consistent with similar findings in other recent reports (5,6,11).
To ensure that the increased CAT activity induced by T 3 was not a consequence of increased transcription of H-IRE-CAT, CAT mRNA was quantitated by slot blotting and RNase protection. CAT mRNA levels were measured in RNA extracted from the same H-IRE-CAT HepG2 transfectants by slot blotting. The slot blot was performed in duplicate for two separate loadings of 4 and 20 g of RNA. Compared to control (hypothyroid medium alone), no significant increase in CAT mRNA was detected after the addition of T 3 to the cultured cells (Fig. 6A). The radioactivity in each of the bands was quantitated, and no increase in counts was detectable in the T 3 -treated samples (data not shown). These results are consistent with the data in RNase protection analysis confirmed that the levels of protected CAT mRNA in H-IRE-CAT-transfected HepG2 cells do not change in response to iron, T 3 , or a combination of iron and T 3 (Fig. 6B, lanes 2-4, respectively). This is consistent with previous data confirming the post-translational nature of the regulation of IRP binding activity by iron. Furthermore, these results indicated that the effect of T 3 was not due to transcriptional up-regulation of H-IRE-CAT. Taken together, these results indicate that T 3 augments Fe-induced displacement of the IRP from the ferritin IRE independent of changes in H-IRE-CAT transcription. Moreover, these functional data complement our findings in the REMSA studies and support the notion that, in iron-replete cells, T 3 acts to augment the effect of iron, by modifying the binding activity of the IRP to the IRE in ferritin mRNA.
T 3 -induced Modulation of IRP Binding Activity and Thyroid Hormone Receptors-Rat liver and human hepatoma HepG2 cells contain well characterized T 3 -receptors (TRs) (38). To investigate if TRs were involved in mediating this T 3 -induced effect on the IRP⅐IRE RPC, cytoplasmic extracts were produced from monkey kidney (CV-1) cells, which do not contain TRs (40), and analyzed by REMSA. After incubation of CV-1 cells with FAC (110 M), addition of T 3 (100 or 1 nM) had no effect on the IRP⅐IRE RPC (Fig. 7, lanes 2 and 3). These experiments utilized iron-replete cells because this is the state in which T 3 appears to have the most effect on IRP binding both in vitro and in vivo. This finding suggests that TRs are required to mediate this effect in HepG2 cells and argues against the possibility of a direct cytoplasmic effect of T 3 .
To determine if a component of this T 3 effect was a direct cytoplasmic effect, and independent of nuclear TRs, we incu-2 P. Leedman, unpublished observation. bated various concentrations of T 3 (10 Ϫ5 -10 Ϫ9 M) with HepG2 cell cytoplasmic extracts containing labeled pgem-IRE probe, at the commencement of the reaction. As can be seen in Fig. 8, addition of T 3 directly to the cytoplasmic reaction mixture (lanes 2-4), did not modify IRP binding activity. These data support our results in CV-1 cells and are consistent with the notion that in liver cells T 3 acts via nuclear TRs to facilitate displacement of the IRP from the ferritin IRE. DISCUSSION The molecular mechanisms governing the regulation of ferritin gene expression by iron are based on regulated changes in the IRE binding activity of the IRPs. Here we provide the first evidence that T 3 , a hormone critical for maintaining body homeostasis, can modulate the binding activity of the IRP to a ferritin IRE both in vitro and in vivo. Our REMSA results indicate that T 3 acts post-translationally to augment the ironinduced displacement of the rat and human IRP from an IRE present in the 5Ј-UTR of ferritin mRNA. Furthermore, this T 3 -induced effect was associated with a similar sized functional increase in IRE-dependent gene expression (ϳ40 -50%), as demonstrated in transfection studies using a human ferritin IRE-CAT construct in human hepatoma cells. Our results suggest that T 3 can, possibly in a TR-dependent manner, functionally regulate the IRE binding activity of the IRP.
Our data provide further evidence that the binding activity of the IRP can be modified by agents other than iron and the redox state. Recent data indicate a direct association with the nitric oxide/nitric oxide synthase pathway, in which increases in NO activates IRP binding to IREs in ferritin and TfR mRNAs (39). The reactive oxygen intermediate hydrogen peroxide (H 2 O 2 ) has recently been shown to increase binding of the IRP, resulting in reduced ferritin synthesis and increased transferrin receptor expression (41). In contrast to activation of IRP by iron depletion which is okadaic acid-insensitive, induction of IRP by H 2 O 2 is okadaic-sensitive suggesting the involvement of stress-induced kinase/phosphatase pathways (41). Changes in the phosphorylation status of the IRP mediated by protein kinase C may provide another level of regulation (42). T 3 , however, is the first endocrine hormone that has been shown to modulate the IRP⅐IRE interaction and ferritin translation, and our studies suggest that nuclear TRs are possibly involved, as evidenced by the absence of effect in CV-1 cells which lack endogenous nuclear TRs. Moreover, T 3 was unable, even at high concentrations (10 Ϫ5 M), to reproduce the effect when added directly to the cytoplasmic extracts. This IRE-dependent action of T 3 on ferritin translation differs from other agents, such as interleukin 1, which modify ferritin translation through IRE-independent mechanisms (43). It is not known which TR isoform(s) (40) is involved in mediating this T 3 -induced effect, and this question requires further investigation.
A significant body of evidence exists showing a positive correlation between the serum levels of T 3 /T 4 and ferritin (24 -28) in individuals with thyroid abnormalities. All of these studies documented elevated serum ferritin levels in patients with hyperthyroidism which normalized when the T 3 /T 4 levels returned to normal. Interestingly, the positive relationship between serum ferritin and T 3 /T 4 levels holds in hypothyroidism as well. A similar positive relationship between serum ferritin and T 3 /T 4 levels has been observed in rats rendered hypo-and hyperthyroid (22). Furthermore, the hepatic ferritin synthesis rate increased significantly in hyperthyroid rats (22), consistent with the increased serum ferritin levels. These observations are consistent with our own data. To date, there are no human data to suggest that serum ferritin levels rise in hypothyroidism. Thus, differences exist between the recent data and the results presented herein compared to the earlier reports documenting increased rat hepatic ferritin synthesis in hypothyroidism (20,21). The reasons for this are unclear, but may relate, in part, to the nature of thyroid dysfunction utilized in each study (e.g. thyroidectomy versus T 3 supplementation). Interestingly, however, the earlier work did document that the changes induced by T 3 were at the post-transcriptional level which would be consistent with our results (21).
The IRP is one of two trans-acting RNA-binding proteins whose binding activity is modified by T 3 . The other is an 80 -85-kDa pituitary protein that recognizes a specific region within the 3Ј-UTR of TSH␤ mRNA, an anterior pituitary hormone (33). Remarkably, the RNA binding site for this T 3regulated pituitary trans-acting factor within TSH␤ 3Ј-UTR mRNA contains features of a ferritin IRE. The TSH␤ 3Ј-UTR sequence similarity with a ferritin IRE extends over 12 nucleotides (9 of 12 nucleotides, which includes the loop and a portion of the IRE stem, Fig. 9). To investigate whether this sequence could compete with the IRE for IRP binding, REMSA studies were performed. These showed that excess unlabeled TSH␤ 3Ј-UTR mRNA could compete efficiently with labeled pgem-IRE for IRP binding in pituitary cells (33). Further, pgem-IRE competed efficiently with TSH␤ 3Ј-UTR mRNA for binding of this T 3 -regulated trans-acting factor (33). In summary, TSH␤ mRNA has an IRE-like element which can interact with pituitary factors, including the IRP, in a T 3 -dependent manner. The cellular consequences of interactions between the IRP with other mRNAs sharing sequence similarity with the IRE is under further investigation.
Our results are consistent with a model of iron homeostasis in which IRE-dependent ferritin gene expression is positively FIG. 7. Lack of effect of T 3 on IRP binding to a ferritin IRE in iron-replete CV-1 cells. REMSA (4% PAGE) was performed using 5 g of CV-1 cell (monkey kidney cell that lacks TRs) extract and 32 Plabeled pgem-IRE. The cells were cultured in hypothyroid medium for 36 h, before iron (FAC (110 M), lanes 1-4) was added to the medium 16 h prior to harvesting. T 3 (1 nM, lane 2; 100 nM, lane 3) was added to the reaction mixture at the beginning of the incubation. REMSA was performed and analyzed as described in Fig. 2. This is a representative experiment (n ϭ 2). REMSA was performed and analyzed as described in Fig. 2. This is a representative experiment (n ϭ 2). regulated by T 3 . T 3 alone did not alter expression at the transcriptional or post-transcriptional level. However, in iron-replete cells, there was a significant T 3 effect to up-regulate ferritin gene expression, through modulation of the IRP binding activity and enhanced IRE-dependent translation. These results provide further insight into a rapidly emerging model for the regulation of iron homeostasis, by providing evidence that T 3 acts post-translationally to augment displacement of the IRP from a ferritin IRE. Other endocrine hormones may also modify IRP binding activity and have profound metabolic effects (e.g. retinoic acid, glucocorticoids etc.). Further experiments are in progress to investigate this possibility. Given the important role that both T 3 and iron play in the maintenance of body homeostasis, further elucidation of the control mechanisms governing the interactions between T 3 and IRE-dependent gene expression will be an important goal of future studies. FIG. 9. Nucleotide sequence alignment of the rat ferritin IRE with rat TSH␤ 3-UTR region. The stem-loop structure depicts the rat ferritin IRE contained within the pgem-IRE plasmid (30). The six-membered loop contains five nucleotides that are almost invariant (boxed nucleotides) (44). The 12-nucleotide region of sequence similarity comprises the sequence between the two lines. Nucleotide differences between the IRE and the consensus region sequence are indicated with arrows.