Regulation of Ferritin and Transferrin Receptor mRNAs *

Iron regulates the synthesis of two proteins critical for iron metabolism, ferritin and the transferrin receptor, through novel mRNA/protein interactions. The mRNA regulatory sequence (iron-responsive element (IRE)) occurs in the 5’-untranslated region of all ferritin mRNAs and is repeated as five variations in the 3’-untranslated region of transferrin receptor mRNA. When iron is in excess, ferritin synthesis and iron storage increase. At the same time, transferrin receptor synthesis and iron uptake decrease. Location of the common IRE regulatory sequence in different noncoding regions of the two mRNAs may explain how iron can have opposite metabolic effects; when the IRE is in the 5’-untranslated region of ferritin mRNA, translation is enhanced by excess iron whereas the presence of the IREs in the 3’-untranslated region of the transferrin receptor mRNA leads to iron-dependent degradation. How and where iron actually acts is not yet known. A soluble 90-kDa regulatory protein which has been recently purified to homogeneity from liver and red cells specifically blocks translation of ferritin mRNA and binds IRE sequences but does not appear to be an iron-binding protein. The protein is the first specific eukaryotic mRNA regulator identified and confirms predictions 20 years old. Concerted regulation by iron of ferritin and transferrin receptor mRNAs may also define a more general strategy for using common mRNA sequences to coordinate the synthesis of metabolically related proteins.


Iron regulates
the synthesis of two proteins critical for iron metabolism, ferritin and the transferrin receptor, through novel mRNA/protein interactions. The mRNA regulatory sequence (iron-responsive element (IRE)) occurs in the 5'-untranslated region of all ferritin mRNAs and is repeated as five variations in the 3'-untranslated region of transferrin receptor mRNA. When iron is in excess, ferritin synthesis and iron storage increase.
At the same time, transferrin receptor synthesis and iron uptake decrease. Location of the common IRE regulatory sequence in different noncoding regions of the two mRNAs may explain how iron can have opposite metabolic effects; when the IRE is in the 5'-untranslated region of ferritin mRNA, translation is enhanced by excess iron whereas the presence of the IREs in the 3'-untranslated region of the transferrin receptor mRNA leads to iron-dependent degradation.
How and where iron actually acts is not yet known.
A soluble 90-kDa regulatory protein which has been recently purified to homogeneity from liver and red cells specifically blocks translation of ferritin mRNA and binds IRE sequences but does not appear to be an iron-binding protein.
The protein is the first specific eukaryotic mRNA regulator identified and confirms predictions 20 years old. Concerted regulation by iron of ferritin and transferrin receptor mRNAs may also define a more general strategy for using common mRNA sequences to coordinate the synthesis of metabolically related proteins.
Precise modulation of cellular iron is required both to provide for the synthesis of iron proteins such as ribonucleotide reductase and cytochromes and to prevent damage from free radicals produced by iron/dioxygen interactions. Cellular iron metabolism is self-regulated through iron-dependent changes in the abundance of ferritin, which sequesters excess iron, and transferrin receptors, which control iron uptake. The effect of changing intracellular iron concentrations is mainly on mature mRNA (Table I) rather than on transcription of DNA or protein degradation.
For example, when iron is in excess, cells use stored ferritin mRNA to synthesize more ferritin for iron storage (l-3, 5-18). At the same time the stability of the transferrin receptor mRNA decreases, which diminishes receptor synthesis and iron uptake (4, 19-24). Conversely, when iron levels are low, transferrin receptor mRNA is stabilized, more receptor is synthesized, and iron uptake increases, while ferritin mRNA is masked, ferritin * I am grateful for support from the Hematology Extramural Program of the National Institutes of Health (Grant DK20251) and the North Carolina Agricultural Research Service both for research and for the writing of the review. synthesis declines, and iron storage decreases. Iron regulation of iron storage and transport depends on common sequences in the noncoding regions of ferritin and transferrin receptor mRNAs and specific mRNA regulatory protein(s); the use of common regulatory sequences in ferritin and transferrin receptor mRNAs is the first known example of concerted regulation of eukaryotic mRNAs (Fig. 1). Ferritin and transferrin receptor expression is also regulated in the more conventional transcriptional mode during growth and differentiation (20,26,27) and by iron during extreme stress (9, 10). Why mRNA is the site of regulation by iron is not known, but the history of translational control of ferritin by iron is long (reviewed in Ref. 25).
Other proteins besides ferritin and transferrin receptors are regulated by controlling mRNA storage or stability. Examples include storage of mRNA for ribonucleotide reductase (28), heat shock housekeeping proteins (29), chlorophyll-binding proteins (e.g. Refs. 30 and 31), tubulin (32), and proto-oncogene proteins (33); variations in mRNA stability are reviewed in Refs. 34 and 58. Signals which control mRNA storage or stability include fertilization, heat shock, and light, as well as iron, suggesting that the effect of iron on the mRNAs for ferritin and transferrin receptors could exemplify a broader strategy for the regulation of developmentally or metabolically related proteins. In addition, ferritin and transferrin receptor mRNAs provide models for the specific interactions of eukaryotic mRNAs and regulatory proteins.

Iron and Ferritin
Synthesis: mRNA Storage and Utilization How does iron regulate ferritin synthesis over a 40-fold range (Fig. 2) without a major change in ferritin mRNA concentration?
First, mature functional ferritin mRNA can be stored in a masked or inactive form (1, 2); in iron-rich cells, the mRNA is released from storage and enters polyribosomes (2,35). Second, unmasked mRNA can be translated very efficiently or preferentially in some cell types (5,8,25). The different types of ferritin subunit mRNAs in a cell (3,7) are usually translated at equal rates both in uiuo (12) and in vitro (36).
Unmasking Stored Ferritin mRNA-Does iron alter ferritin mRNA? Ferritin mRNA isolated from normal and iron-loaded cells could not be distinguished structurally (12) or functionally ( Fig. 2) in cytoplasmic extracts that lack regulatory factors such as wheat germ (1). Therefore, iron must act indirectly through cytoplasmic factors. Such factors occur in red cell lysates which mimic intact iron-poor red cells (36). In contrast to other mRNAs, ferritin mRNA cannot form polyribosomes in red cell lysates (36); the first source for a ferritinspecific regulatory protein was a lysate of red cells (41) (see "Regulatory Protein(s) for Ferritin and Transferrin Receptor mRNAs").
What is known about the iron signal? Hemin and diferric transferrin induce ferritin synthesis in whole cells or transfected cells, but effects of iron in cell extracts have been elusive. For example, the reticulocyte lysates, which regulate ferritin mRNA translation, were unresponsive to added transferrin or hemin (36) but wheat germ extracts, which normally do not regulate ferritin mRNA translation, responded to hemin when a purified regulator protein was added (37). The results suggest that normally the link between iron and ferritin mRNA in whole cells involves several intermediates.
Ferritin mRNA Efficiency-When is the efficiency of translation of ferritin mRNA exploited in iron-loaded cells? If the ratio of mRNA to ribosomes and initiation factors is so high that initiation factors and ribosomes become limiting, protein encoded in efficient mRNAs will be synthesized more rapidly. Such a situation occurs in reticulocytes where ferritin is actually synthesized in preference to globin in iron-rich cells (5). However, in cells with rapid rates of protein synthesis, e.g. fibroblasts or hepatocytes, ribosomes and initiation fac- tors are in relative abundance and ferritin mRNA has no particular translational advantage (8,38).

Iron and Tranaferrin
Receptor Synthesis: mRNA Loss Changes in protein synthesis which depend on changes in transcription and/or RNA processing are the most common forms of genetic regulation currently known. How does the iron-dependent decrease in transferrin receptor mRNA concentration differ from the usual mode of regulation? First, iron has only a small effect on the transcription of the transferrin receptor gene (4, 24). Second, the transferrin receptor gene promoter has only a small effect on irondependent regulation of the transferrin receptor (20). Finally, sequences in the transferrin receptor mRNA have a major effect on the iron-induced loss of the mRNA (19-24). HOW iron leads to the loss of transferrin receptor mRNA and whether specific nucleases participate is not yet known. cell-free extracts or deletion/protein mRNA binding measured by changes in electrophoretic mobility in gels; and 2) sequence comparison. The unexpected result was detection of a common regulatory sequence in the 5'-untranslated region of ferritin mRNA and the 3'-untranslated region of the transferrin receptor mRNA (Fig. 3) (21, 22, 39). Location of the common sequence in different regions of the two mRNAs may be related to the opposite effects iron has on the two mRNAs: increased translation of ferritin mRNA and destabilization of transferrin receptor mRNA (Table I, Fig. 1). Both mRNAs have additional regulatory sequences which are not shared.
at position 9 or 12 (shown with an asterisk in Fig. 3) results in loss of regulation (17, 18) by iron. A 70-nucleotide sequence in the 3'-untranslated region of ferritin mRNA also contributes to translational regulation of ferritin synthesis (16, 36), but the effect is independent of iron. Intact cells or cell extracts which normally regulate ferritin mRNA translation appear to be required to detect the effect of the 3'-untranslated region of ferritin mRNA. The regulatory element in the 3'-untranslated region of ferritin mRNA has conserved features of secondary structure (36). How the 3'-untranslated sequence of ferritin mRNA functions in translation is still unknown, but the long distance between the sequence and the initiation site suggests that complex folding is involved.
Transferrin Receptor mRNA-The 3'-untranslated region of the transferrin receptor mRNA is unusually long, with 2650 nucleotides compared with 150 nucleotides in ferritin mRNAs. A sequence of -600 nucleotides contains two types of sequences (Figs. 1, 3) required for iron-dependent regulation of transferrin receptor mRNA (19-24); three critical nucleotides in the center of the sequence are flanked by five variable copies of the ferritin 5'-untranslated region IRE! Transferrin receptor IREs can be exchanged with the ferritin IRE for iron regulation of protein synthesis in transfected cells (21) or binding of proteins (22)(23)(24). IREs A and E are less important for regulation than IREs B-E (24), but the C residue required for regulation of ferritin mRNA is also required in each IRE for regulation of transferrin receptor mRNA (indicated by an * in Fig. 3). The presence of the IRE in both ferritin and transferrin receptor mRNAs allows iron levels in the cell to influence both mRNAs in concert.
Secondary/Tertiary Structure of the FerritinlTransferrin Receptor IRE-Secondary structure predictions (53,54) show that the IREs can be folded into a stem-loop structure with flanking sequences of variable length and sequence that lengthen the stem (Fig. 3). Experiments now in progress to analyze the actual structure of the IRE indicate more complexity than the two-dimensional prediction (55). The regulatory protein(s) may, therefore, recognize a three-dimensional surface rather than a two-dimensional structure.

Regulatory Protein(s) for Ferritin and Transferrin Receptor mRNAs
The idea of translational control of specific mRNAs by specific proteins was first stated 20 years ago (56). However, a functional assay became available for detecting mRNAspecific regulatory factors only recently (36); ferritin is the first eukaryotic mRNA for which a specific regulatory protein has been identified (41). A soluble 90-kDa ferritin mRNA regulatory protein from rabbit red cells and liver has now been purified to homogeneity (44); the regulatory protein also binds IRE sequences. Using binding rather than translational regulation as an assay, a protein of similar size has also been purified from human liver (45). Multiple IRE binding proteins may also occur (42). Thus, the story of specific mRNA regulatory proteins is far from complete. Free sulfhydryl groups are important for binding of specific protein to the IRE of ferritin mRNA (40,(43)(44)(45).
For example, binding is enhanced by 2-mercaptoethanol or dithiothreitol, particularly in crude cell extracts. In addition, the regulatory protein can be inactivated by N-ethylmaleimide (43). Binding constants for the protein/IRE interaction are 0.02-4 nm (45, 57) for a 35-mer containing the IRE, but binding to full-length mRNA has yet to be measured.

Perspectives
Concerted regulation of two mRNAs which encode two metabolically related proteins has not been observed before in eukaryotes. The use of a common mRNA structural motif is even more remarkable. Location of the common mRNA sequences in different noncoding regions of the two mRNAs, the 5'-untranslated region of ferritin mRNA and the 3'untranslated region of the transferrin receptor mRNA, may be related to the opposite effects of iron, i.e. translation of ferritin mRNA uersuS degradation of transferrin receptor mRNA. Although the rate at which information is accumulating is rapid, many questions remain. For example, ferritin is much more widespread than the transferrin receptor. Will the IRE be found in nonvertebrate or plant ferritin mRNA as well or only in those organisms which have transferrin and the receptor?
Questions about iron and the mechanisms of mRNA storage and stability are still largely unanswered. What is the molecular form of the iron signal? Does the regulator proteinferritin mRNA complex prevent binding of initiation factors or unwinding of mRNA secondary structure? Does the regulator protein/IRE interaction in transferrin receptor mRNA prevent nuclease attack? What are the actual structures of the ferritin/transferrin mRNA IREs? Where does the regulator protein bind? What is the structure of the regulator protein? Do concentrations of the regulator protein vary in different cell types? With iron concentrations? To be able to pose such specific questions about mRNA structure and function is exhilarating itself. More importantly, the experimental paths to the answers for ferritin and transferrin receptor mRNAs, and possibly for other examples of concerted translational regulation, are at hand. ii 9: ::: 12.