Specificity of the induction of ferritin synthesis by hemin

The pathology of superficial siderosis of the central nervous system

Chronic or intermittent extravasations of blood into the subarachnoid space, and dissemination of heme by circulating cerebrospinal fluid, are the only established causes of superficial siderosis of the central nervous system (CNS). We studied the autopsy tissues of nine patients by iron histochemistry, immunocytochemistry, single- and double-label immunofluorescence, electron microscopy of ferritin, and high-definition X-ray fluorescence. In one case, frozen brain tissue was available for quantitative assay of total iron and ferritin. Siderotic tissues showed extensive deposits of iron and ferritin, and infiltration of the cerebellar cortex was especially severe. In addition to perivascular collections of hemosiderin-laden macrophages, affected tissues displayed iron-positive anuclear foamy structures in the neuropil that resembled axonal spheroids. They were especially abundant in eighth cranial nerves and spinal cord. Double-label immunofluorescence of the foamy structures showed co-localization of neurofilament protein and ferritin but comparable merged images of myelin-basic protein and ferritin, and ultrastructural visualization of ferritin, did not allow the conclusion that axonopathy was simply due to dilatation and rupture of fibers. Heme-oxygenase-1 (HO-1) immunoreactivity persisted in macrophages of siderotic cerebellar folia. Siderosis caused a large increase in total CNS iron but high-definition X-ray fluorescence of embedded tissue blocks excluded the accumulation of other metals. Holoferritin levels greatly exceeded the degree of iron accumulation. The susceptibility of the cerebellar cortex is likely due to Bergmann glia that serve as conduits for heme; and the abundance of microglia. Both cell types biosynthesize HO-1 and ferritin in response to heme. The eighth cranial nerves are susceptible because they consist of CNS axons, myelin, and neuroglial tissue along their subarachnoid course. The persistence of HO-1 protein implies continuous exposure of CNS to free heme or an excessively sensitive transcriptional response of the HO-1 gene. The conversion of heme iron to hemosiderin probably involves both translational and transcriptional activation of ferritin biosynthesis.

Haem precursor delta-aminolaevulinic acid induces activation of the cytosolic iron regulatory protein 1

Control of cellular iron homoeostasis is performed by iron regulatory protein 1 (IRP1) through post-transcriptional modifications. This protein is sensitive to intracellular iron availability, being activated at low iron levels and inactivated at high iron levels, conditions that signal the increased expression of the transferrin receptor or of ferritin respectively. IRP1 is known to be activated by some oxidants such as H2O2 and NO. delta-Aminolaevulinic acid (ALA), previously found to produce reactive oxygen species and a carbon-centred radical, to release iron from ferritin, and to increase rat liver and brain non-haem iron and ferritin, was investigated for its effects on IRP1 activity in cultured hamster pulmonary fibroblasts. We have found that 1-2 mM ALA produced a 2-3-fold activation of IRP. On incubation with 1-4 mM succinylacetone methyl ester, a potent ALA dehydratase inhibitor, a 3-4-fold activation of the protein was observed, accompanied by a 40% increase in the intracellular ALA concentration. When cells were incubated in the presence of ALA or succinylacetone methyl ester, N-acetylcysteine inhibited IRP1 activation, suggesting that the observed effect is mediated by an oxidative process. We surmise that ALA-induced IRP1 activation might act as a co-sensor of iron homoeostasis.

Regulation of expression of ferritin H-chain and transferrin receptor by protoporphyrin IX

The effect of protoporphyrin IX (hemin without iron) on the expression of transferrin receptor and ferritin was investigated in Friend leukemia cells. Cells treated with protoporphyrin IX exhibit enhanced transferrin-receptor expression and markedly reduced ferritin synthesis. Stimulation of transferrin-receptor expression is observed at both the mRNA and protein level. The effect on ferritin synthesis is mediated by translational inhibition of the mRNA, which, in contrast, is transcriptionally stimulated by protoporphyrin IX treatment. The regulation of transferrin receptor and ferritin in response to iron perturbations has been studied extensively and is mediated by the binding of iron-regulatory proteins (IRP) to the iron-responsive elements (IRE) present in the 3' and 5' untranslated regions of the transferrin-receptor and ferritin mRNA, respectively. To elucidate the molecular mechanisms underlying the effects of protoporphyrin IX on ferritin and transferrin-receptor expression, the role of the IRE sequence was investigated both in vivo by transfection experiments, with a construct containing the coding region for the chloramphenicol acetyltransferase (CAT) reporter gene under the translational control of the ferritin IRE, and in vitro by RNA band-shift assays. Whereas, examination of IRP binding to the IRE by in vitro assays suggests an apparent inactivation of IRP by protoporphyrin IX treatment, CAT assays indicate that protoporphyrin IX is able to induce in vivo a translational inhibition similar to that obtained by treatment with the iron chelator Desferal. This observation raises the possibility of different effects on the IRP activity exerted by porphyrin treatment in intact tissue-culture cells and in vitro. We conclude that translation of ferritin mRNA and degradation of transferrin-receptor mRNA are inhibited in intact tissue-culture cells by protoporphyrin IX through a mechanism similar to that exerted by iron chelation, thus involving depletion of the intracellular iron pool. These results can improve the understanding of the regulation of ferritin gene expression in some pathological conditions associated with disturbed heme synthesis.

The ferritins: molecular properties, iron storage function and cellular regulation

The iron storage protein, ferritin, plays a key role in iron metabolism. Its ability to sequester the element gives ferritin the dual functions of iron detoxification and iron reserve. The importance of these functions is emphasised by ferritin's ubiquitous distribution among living species. Ferritin's three-dimensional structure is highly conserved. All ferritins have 24 protein subunits arranged in 432 symmetry to give a hollow shell with an 80 A diameter cavity capable of storing up to 4500 Fe(III) atoms as an inorganic complex. Subunits are folded as 4-helix bundles each having a fifth short helix at roughly 60 degrees to the bundle axis. Structural features of ferritins from humans, horse, bullfrog and bacteria are described: all have essentially the same architecture in spite of large variations in primary structure (amino acid sequence identities can be as low as 14%) and the presence in some bacterial ferritins of haem groups. Ferritin molecules isolated from vertebrates are composed of two types of subunit (H and L), whereas those from plants and bacteria contain only H-type chains, where 'H-type' is associated with the presence of centres catalysing the oxidation of two Fe(II) atoms. The similarity between the dinuclear iron centres of ferritin H-chains and those of ribonucleotide reductase and other proteins suggests a possible wider evolutionary linkage. A great deal of research effort is now concentrated on two aspects of ferritin: its functional mechanisms and its regulation. These form the major part of the review. Steps in iron storage within ferritin molecules consist of Fe(II) oxidation, Fe(III) migration and the nucleation and growth of the iron core mineral. H-chains are important for Fe(II) oxidation and L-chains assist in core formation. Iron mobilisation, relevant to ferritin's role as iron reserve, is also discussed. Translational regulation of mammalian ferritin synthesis in response to iron and the apparent links between iron and citrate metabolism through a single molecule with dual function are described. The molecule, when binding a [4Fe-4S] cluster, is a functioning (cytoplasmic) aconitase. When cellular iron is low, loss of the [4Fe-4S] cluster allows the molecule to bind to the 5'-untranslated region (5'-UTR) of the ferritin m-RNA and thus to repress translation. In this form it is known as the iron regulatory protein (IRP) and the stem-loop RNA structure to which it binds is the iron regulatory element (IRE). IREs are found in the 3'-UTR of the transferrin receptor and in the 5'-UTR of erythroid aminolaevulinic acid synthase, enabling tight co-ordination between cellular iron uptake and the synthesis of ferritin and haem. Degradation of ferritin could potentially lead to an increase in toxicity due to uncontrolled release of iron. Degradation within membrane-encapsulated "secondary lysosomes' may avoid this problem and this seems to be the origin of another form of storage iron known as haemosiderin. However, in certain pathological states, massive deposits of "haemosiderin' are found which do not arise directly from ferritin breakdown. Understanding the numerous inter-relationships between the various intracellular iron complexes presents a major challenge.

Hemin inhibits the interferon-beta-induced antiviral state in established cell lines

Hemin and other metalloporphyrins are known as very versatile compounds in nature, because they are able to carry out numerous functions in a free state or in association with specific proteins. When Friend murine erythroleukemia cells are treated with IFN-beta plus 100 microM hemin, the antiviral state is not observed, whereas the antiviral effect of IFN-gamma is unaffected by hemin treatment. This inhibitory effect of hemin is not restricted to erythroid cells. In fact, it is also observed in murine L929 and in human cell lines treated with IFN-beta. Neither trivalent iron in other forms nor hemin analogs (such as protoporphyrin IX or Sn(2+)-protoporphyrine IX) mimic this effect. Conversely, Co(3+)-protoporphyrin IX was as effective as hemin. At the transcriptional level, results obtained by run-on assays on nuclei from IFN-treated cells indicate that hemin does not completely inhibit IFN-beta induction of 2-5A synthetase gene(s) at 6 h of treatment but abolishes it at 24 h. In addition, hemin is able to inhibit the accumulation of IFN-induced 2-5A synthetase mRNAs. Experiments carried out to investigate the hemin effect on the early steps of the IFN signaling pathway indicate that hemin interferes with the ability of type I IFN to bind to its receptor, probably by a direct action on the IFN molecule.

The pathogenesis of superficial siderosis of the central nervous system

In advanced cases of superficial siderosis of the human central nervous system, the clinical triad of hearing loss, cerebellar ataxia, and myelopathy permits the diagnosis at the bedside, and magnetic resonance imaging readily confirms the hemosiderin deposits in brainstem, cerebellum, and spinal cord. To study the pathogenesis of this condition and explain the selective vulnerability of the cerebellum, experimental siderosis was induced in rabbits by the repeated intracisternal injection of autologous red blood cells. The earliest cellular response in the cerebellar molecular layer was hyperplasia and hypertrophy of microglia as displayed by immunocytochemistry for ferritin. Microglia also contained iron, but ferritin biosynthesis appeared to proceed without commensurate iron accumulation. This early apoferritin response probably occurred due to the presence of heme, rather than iron, in the cerebrospinal fluid and subpial tissue. Ferritin biosynthesis is accelerated when the ferritin repressor protein is dissociated from ferritin messenger ribonucleic acid. A specific antiserum localized ferritin repressor protein predominantly to astrocytes including Bergmann glia. It is proposed that abundance and proximity of ferritin repressor protein--immunoreactive Bergmann glia and ferritin-containing microglia in the cerebellar molecular layer permit prompt cellular interaction in the conversion of heme to ferritin and ultimately hemosiderin.

Iron sequestration by macrophages decreases the potential for extracellular hydroxyl radical formation

Alveolar macrophages (AM) from smokers contain a much higher quantity of intracellular iron than AM from nonsmokers. Since some forms of iron will catalyze the formation of hydroxyl radical (.OH) from superoxide and hydrogen peroxide, the ability of AM derived from smokers and nonsmokers to generate .OH was assessed. No detectable .OH was produced by AM from either source, suggesting that iron sequestration by AM may limit the potential for .OH-mediated lung injury. Consistent with this hypothesis, the ability of bronchoalveolar lavage fluid (BAL) from smokers and nonsmokers to act as an .OH catalyst decreased after exposure to AM. We found that, like AM, human monocyte-derived macrophages (MDM) have the ability to acquire large quantities of iron from small low molecular weight iron chelates as well as decrease the ability of BAL to act as a .OH catalyst. When MDM or AM were exposed to the iron chelates or BAL they were then able to generate .OH after phorbol myristate acetate stimulation. However, when acutely iron-loaded or BAL-exposed MDM were placed in culture, their ability to produce .OH decreased with time to the level of non-iron-exposed controls. This process correlated with iron translocation from the plasma membrane to the cytosol as well as a 3-9-fold increase in cellular ferritin. No increase in antioxidant enzyme levels or induction of the heat shock response was observed. Iron sequestration by macrophages may protect nearby cells from exposure to potentially cytotoxic iron-catalyzed oxidants such as .OH.

Experimental superficial siderosis of the central nervous system: biochemical correlates

The pathogenesis of superficial siderosis of the central nervous system (CNS) may be examined by the repeated intracisternal injection of washed autologous red blood cells (RBC). In rabbits, the injections cause the accumulation of iron in the cytoplasm of microglial cells and astrocytes of cerebellar and cerebral cortices. Immunocytochemistry for ferritin reveals enhanced reaction product mainly in microglia but hemosiderin occurs only after extending the injections to 6 months. In an effort to determine the biochemical correlates of these morphological changes, iron, ferritin, ferritin subunits and the ferritin repressor protein (FRP) were quantitated. There was no increase of total iron or ferritin in the exposed cortical areas. However, the injections of RBC caused dramatic shifts of the relative contributions by heavy (H-) and light (L-) ferritin subunits. The initial response was a prompt increase of the H/L ratio to over 4.0 from the normal ratio near 1.0. Extended injections caused the ratio to drop to below unity, and the predominance of L-ferritin at 6 months coincided with the appearance of granular hemosiderin. This investigation also confirmed the presence of FRP in rabbit brain cytosols but the induction of experimental superficial siderosis did not change its levels or in vitro affinity for the iron-responsive element in ferritin messenger ribonucleic acid. It is proposed that the incrustation by hemosiderin which characterizes superficial siderosis of the CNS in humans occurs when prolonged exposure to hemoglobin produces persistent shifts of the H/L-ratios by accumulation of L-ferritin.

In vivo and in vitro modulation of the mRNA-binding activity of iron-regulatory factor. Tissue distribution and effects of cell proliferation, iron levels and redox state

The mRNA-binding protein, iron-regulatory factor (IRF) has a central role in iron metabolism. It coordinately increases transferrin-receptor mRNA stability and inhibits translation of ferritin and erythroid delta-aminolevulinate synthase mRNA by binding to specific mRNA structures, the iron-responsive elements (IRE). In gel-retardation assays, IRF had a broad tissue distribution, showing activity in cytosolic extracts from 12 mouse organs tested. In all these extracts, IRF could be further activated in vitro by 2-mercaptoethanol. In cultured mouse 3T6 fibroblasts, growth stimulation after low serum arrest increased IRF activity 10-fold, mainly through activation of existing inactive IRF. No change was observed during progression of 3T6 cells through the cell cycle. IRF activation by iron chelators has been postulated to result in the reduction of an intramolecular sulfhydryl group. In a search for redox conditions that regulate IRE binding of IRF, we studied several compounds in vitro or in vivo. Hemin, known to inactivate IRF in vivo, showed a similar, reversible effect in vitro, presumably by oxidizing IRF. However, this did not appear to be relevant for the mode of IRF regulation in vivo. Addition of protoporphyrin IX to intact cells induced IRF activity almost to the same extent as desferrioxamine. This effect was inhibited by iron salts, indicating that IRF is activated in vivo through depletion of a chelatable iron pool. In vitro activation by reductants other than 2-mercaptoethanol suggested some selectivity in their access to relevant sulfhydryl groups, but did not reveal which natural redox-sensitive compound might regulate IRF in vivo. However, in cultured cells, inactivation of free IRF by the sulfhydryl-specific oxidizing agent diamide was much more rapidly reversed than inactivation by iron salts. This indicates the direct involvement of a cellular reductant in setting IRF activity and suggests a rate-limiting IRF conformation that is reached only in the presence of iron, but not after diamide oxidation.

Modulation of the RNA-binding activity of a regulatory protein by iron in vitro: switching between enzymatic and genetic function?

The iron-responsive element-binding protein (IRE-BP) is an RNA-binding protein that regulates the expression of several mRNAs in response to availability of cellular iron. The iron-dependent control of IRE-BP activity has been reconstituted in vitro. Incubation of purified IRE-BP with iron salts in the presence of the reducing agent cysteine decreases IRE-BP binding to the cognate RNA element. The specificity of this effect is established by several parameters: (i) the interaction of the spliceosomal protein U1A with its U1 small nuclear RNA target sequence as an internal control is unaffected by iron perturbations, (ii) non-iron metals fail to mimic the iron effect, and (iii) iron chelator activates the IRE-binding activity of IRE-BP and titrates the effect of iron salts. Modulation of IRE-BP activity by chelatable iron is reversible and thus does not involve permanent alterations of the integrity of the protein. These findings accurately mirror the physiological basis for iron regulation of transferrin receptor mRNA stability as well as ferritin and erythroid 5-aminolevulinate synthase mRNA translation in vivo. We discuss these data vis-a-vis the structural homology of IRE-BP with the iron-sulfur protein aconitase and propose a mechanism by which the same cytoplasmic protein serves a dual function as an RNA-binding factor and an enzyme.

Enhanced degradation of the ferritin repressor protein during induction of ferritin messenger RNA translation

Induction of ferritin synthesis in cultured cells by heme or iron is accompanied by degradation of the ferritin repressor protein (FRP). Intermediates in the degradative pathway apparently include FRP covalently linked in larger aggregates. The effect of iron on FRP degradation is enhanced by porphyrin precursors but is decreased by inhibitors of porphyrin synthesis, which implies that heme is an active agent. These results suggest that translational induction in this system may be caused by enhanced repressor degradation. While unique among translational regulatory systems, this process is common to a variety of other biosynthetic control mechanisms.

Expression of active iron regulatory factor from a full-length human cDNA by in vitro transcription/translation

Iron regulatory factor (IRF), also called iron responsive element-binding protein (IRE-BP), is a cytoplasmic RNA-binding protein which regulates post-transcriptionally transferrin receptor mRNA stability and ferritin mRNA translation. By using the polymerase chain reaction (PCR) and the sequence published by Rouault et al. (1990) a probe was derived which permitted the isolation of three human IRF cDNA clones. Hybridization to genomic DNA and mRNA, as well as sequencing data indicated a single copy gene of about 40 kb specifying a 4.0 kb mRNA that translates into a protein of 98,400 dalton. By in vitro transcription of a assembled IRF cDNA coupled to in vitro translation in a wheat germ extract, we obtained full sized IRF that bound specifically to a human ferritin IRE. In vitro translated IRF retained sensitivity to sulfhydryl oxidation by diamide and could be reactivated by beta-mercaptoethanol in the same way as native placental IRF. An IRF deletion mutant shortened by 132 amino acids at the COOH-terminus was no longer able to bind to an IRE, indicating that this region of the protein plays a role in RNA recognition. Placental IRF has previously been shown to migrate as a doublet on SDS-polyacrylamide gels. After V8 protease digestion the heterogeneity was located in a 65/70 kDa NH2-terminal doublet. The liberated 31 kDa COOH-terminal polypeptide was found to be homogeneous by amino acid sequencing supporting the conclusion of a single IRF gene.

Crosslinking of hemin to a specific site on the 90-kDa ferritin repressor protein

Incubation of a 90-kDa ferritin repressor protein (FRP) with small amounts of radiolabeled hemin resulted in the formation of a strong interaction between the two that was stable to SDS/PAGE. (We refer to this interaction as a "crosslink," without intending to imply knowledge as to its chemical nature.) Of seven other proteins tested individually, only apohemopexin and bovine serum albumin showed similar crosslinking ability, albeit to a much lower extent. [14C]Hemin specifically crosslinked to FRP in the presence of a 50-fold excess of total wheat germ proteins. Inclusion of catalase did not prevent the reaction of hemin with FRP, suggesting that H2O2 is not involved. The subsequent addition of a stoichiometric amount of apohemopexin did not reverse the reaction. Exhaustive digestion of the complex with Staphylococcus aureus V8 protease produced a major labeled peptide of 17 kDa. These results show the existence of a highly specific, uniquely reactive hemin binding site on FRP.