Disorders of Iron and Copper Metabolism, the Sideroblastic Anemias, and Lead Toxicity

Iron and Oxidation

The key to the biologic utility of iron is its ability to exist in either of two stable oxidation states: Fe ²⁺ (ferrous) or Fe ³⁺ (ferric). This property permits iron to act as a redox catalyst by reversibly donating or accepting electrons. An excellent example is the electron transport chain of oxidative phosphorylation, in which adenosine triphosphate (ATP) is generated from glucose by the orderly transfer of electrons through a network of iron-containing mitochondrial cytochromes.

When dissolved in aqueous solution, ferrous iron rapidly oxidizes to ferric iron, which is insoluble at physiologic pH. The resulting ferric hydroxide salts (rust) are of no metabolic utility. To achieve stable solubility under physiologic conditions, iron complexes with iron-binding agents termed chelators. Chelators are synthesized by all organisms, ranging from microbes (e.g., deferoxamine produced by Streptomyces pilosus ) to humans (e.g., transferrin in human plasma). These molecules are important to environmental iron acquisition and transport as well as intracellular storage.

Iron-Protein Complexes

Iron-protein complexes capitalize on the properties of the metal to perform metabolic functions, the iron acting as the chemical workhorse and the protein dictating biologic specificity. Individual iron atoms can interact directly with amino acid side groups in proteins, as in ribonucleotide reductase. Alternatively, iron may form coordination complexes with other small molecules. For example, protoporphyrin IX donates four of the six electrons needed to form a stable coordination complex with iron. Heme, the iron–protoporphyrin IX complex, is so stable that removal of the iron moiety requires an enzyme, heme oxygenase. The functional properties of heme are determined by the nature of the associated protein or small molecule ligands supplying the remaining two electrons. The best-characterized heme protein is hemoglobin, in which a globin histidine residue donates the fifth electron and the sixth comes from molecular oxygen. This configuration enables hemoglobin to transport oxygen safely throughout the body.

Iron and sulfur atoms can also form complexes (clusters) that catalyze enzymatic reactions. The Krebs cycle enzyme aconitase, for example, contains an iron-sulfur (Fe/S) cluster. As discussed later in this chapter, the iron content of the Fe/S cluster of a related aconitase-like molecule allows it to “sense” iron concentrations within the cell and to act as an iron regulatory protein (IRP) to modulate the translation of genes of iron metabolism.

Iron Toxicity

The ability of iron to catalyze redox reactions also accounts for its toxicity. As an enzymatic cofactor, the metal is involved in the restructuring of cellular components. Unchelated iron has unbridled redox activity, however, and may contribute to cellular degeneration. We live in an oxygen-rich atmosphere, and our bodies require oxygen for many metabolic processes. However, oxygen is highly reactive and its interaction with iron potentiates its toxicity. Normal cellular reactions, including respiration, generate the reactive oxygen intermediates superoxide (O ₂ ⁻ ) and hydrogen peroxide (H ₂ O ₂ ). Oxidative stress develops when production of reactive oxygen species exceeds the body's processing capacity. Under these circumstances, reactive oxygen intermediates may be converted to injurious free radicals by the iron-catalyzed Fenton reaction : Hydroxyl radicals promote the peroxidation of proteins, DNA, and membrane lipids, problems that are exacerbated by iron overload. Certain organelles are particularly susceptible to iron-dependent peroxidation. In iron-overloaded cells, injured mitochondria and lysosomes become leaky. Mitochondrial damage and release of lysosomal proteases cause further cell injury and may ultimately lead to cell death. This process contributes to the severe tissue damage seen in the liver, heart, and endocrine organs of patients with iron overload disorders (see later).

Iron is not normally present in cell membranes. However, in both sickle cell disease and thalassemia, iron, heme, ferritin, and denatured hemoglobin adhere to the inner surface of the red cell plasma membrane and thereby contribute to the pathogenesis of these congenital anemias. The membrane complex containing denatured hemoglobin has been termed hemichrome . The red cell anion transport protein band 3 appears to nucleate the formation of these iron aggregates. Injured cells decorated with membrane iron deposits are removed by a functioning spleen in patients with thalassemia and sickle cell–hemoglobin C disease (hemoglobin SC disease), but they circulate in functionally asplenic patients with homozygous sickle cell–hemoglobin S disease (hemoglobin SS disease). Membrane-associated iron promotes free radical formation and further membrane damage, marked by generation of the lipid peroxidation product malonyldialdehyde and by cross-linking of membrane proteins. Membranes become rigid and thus contribute to the formation of irreversibly sickled cells that occlude the microcirculation.

Sometimes reactive oxygen intermediates can be beneficial. Neutrophils contain a membrane-associated reduced nicotinamide adenine dinucleotide phosphate (NADPH) oxidase that produces superoxide to kill ingested microorganisms (reviewed by Clark ). Superoxide and secondary reactive oxygen intermediates are potent antimicrobial agents. Congenital defects in this NADPH oxidase, collectively termed chronic granulomatous disease (see Chapter 22 ), are characterized by a serious defect in defense against bacterial pathogens.

Neutrophils and iron also injure tissues in inflammatory diseases such as rheumatoid arthritis. Synovial macrophages ingest red cell hemoglobin introduced by intermittent joint hemorrhage. Iron is deposited in the synovial membrane, proximate to superoxide and hydrogen peroxide generated by neutrophils and macrophages participating in the inflammatory reaction. Iron catalyzes the conversion of these compounds to free radical species, which promote lipid peroxidation. Iron therapy exacerbates this process. In contrast, antioxidants and iron chelators retard free radical generation, thereby affording some theoretic protection against injury in rheumatoid arthritis.

These deleterious properties of iron are threatening only when the element is in a “free” state or in an abnormal form within the cell. Protection of cell structures from iron-dependent free radical damage is therefore of utmost importance. When iron is bound to protein either directly or indirectly, the generation of free radicals is largely abrogated. Thus, tight chelation of iron is a means of controlling its reactivity. As discussed later, cytoplasmic ferritin allows iron to be stored safely within cells by sequestering it in an innocuous form. Expression of ferritin is induced by oxidative stress. Both prokaryotic and eukaryotic cells contain ferritins, and mice lacking one of the two ferritin genes do not develop past the early embryonic stages. Thus, ferritin appears to be necessary for most, if not all, living cells.

Iron Absorption

Iron absorption occurs predominantly in the proximal duodenum. The physical state of iron entering the duodenum greatly influences its absorption. At neutral pH, ferrous iron is rapidly converted to the insoluble ferric form. Acid produced by the stomach serves to lower the pH in the duodenum and thereby enhance the solubility and uptake of iron (see later). Heme is absorbed separately from and more efficiently than inorganic iron, independent of duodenal pH. Consequently, meat is an excellent nutritive source of iron. Heme iron absorption is, however, very poorly understood, but a heme oxygenase inhibitor has been shown to block heme catabolism in the intestine and result in an iron-deficient state.

A number of dietary factors influence inorganic iron absorption. Ascorbate and citrate increase iron uptake, in part by acting as weak chelators to help solubilize the metal in the duodenum. Iron is readily transferred from these compounds to the absorptive epithelium. Conversely, plant phytates, bran, and tannins inhibit iron absorption. These compounds chelate iron and prevent its uptake by the absorption machinery.

Through a combination of genetic and biochemical approaches, much has been learned about the absorption of nonheme iron over the past two decades. Nonheme iron arrives at the apical surface of the absorptive duodenal enterocyte in its ferric (Fe ³⁺ ) form. It is reduced through the action of a brush border ferric reductase. This enzyme may be duodenal cytochrome b , a heme protein that is homologous to b 561 cytochromes. Expression of duodenal cytochrome b is significantly greater in the proximal duodenum than elsewhere and increases in iron deficiency.

Ferrous (Fe ²⁺ ) iron is then taken up by the enterocyte through the action of divalent metal transporter 1 (DMT1, also known as Nramp2, DCT1, and SLC11A2). Transport requires the coordinated movement of protons in the same direction (symport). Thus, DMT1 functions only at low pH, and it has little or no activity at neutral pH. DMT1 is widely expressed, but duodenal levels increase dramatically in iron deficiency.

DMT1 can also transport other divalent metal ions, including Cd ²⁺ , Co ²⁺ , Cu ²⁺ , Mn ²⁺ , Pb ²⁺ , and Zn ²⁺ . Competition studies have shown that lead, manganese, cobalt, and zinc can share the intestinal absorption pathway used by iron. Increased iron absorption induced by iron deficiency also enhances the uptake of these elements. Because iron deficiency often coexists with lead intoxication, this interaction has vast public health significance and can produce particularly serious medical complications in children. As discussed later, copper absorption and metabolism are handled by an entirely different mechanism.

After iron enters the absorptive enterocyte through the action of DMT1, it has at least two possible fates. It can be retained by the cell and subsequently be lost when the enterocyte dies and is sloughed into the intestinal lumen, or it can be transported across the basolateral membrane to enter the body. Iron retained by the enterocyte is used for cellular metabolism or incorporated into ferritin. Exported iron leaves the cell by way of a unique basolateral transmembrane iron transporter, ferroportin (FPN1, also known as SLC40A1, MTP1, and IREG1).

Basolateral iron transfer also requires a change in the oxidation state of the metal, likely facilitated by the homologous multicopper ferroxidases hephaestin, and ceruloplasmin, which are membrane associated and present in the plasma, respectively. A comprehensive model of intestinal iron absorption, primarily derived from studies in mice, is shown in Figure 11-2 . This model pertains only to nonheme iron transport; details of heme iron uptake have not yet been worked out. A candidate intestinal heme transporter was reported but subsequently proven to be a folate transporter, instead.

Normally, only about 10% of dietary nonheme iron entering the duodenum is absorbed. However, this value increases significantly with iron deficiency. In contrast, iron overload reduces, but does not completely eliminate, absorption. Finch designated this modulation the “stores regulator.” In addition, both iron deficiency anemia and forms of anemia associated with ineffective erythropoiesis induce a marked increase in iron absorption. This effect has been designated the “erythroid regulator.” The erythroid regulator appears to exert a dominant effect over the stores regulator, as individuals with chronic persistent ineffective erythropoiesis can become profoundly iron overloaded even in the absence of transfusion. Hypoxia also increases iron absorption, independent of anemia.

Hepcidin, an iron-regulated peptide hormone, links the actions of all these regulators through a unifying molecular mechanism. Hepcidin is a 25–amino acid peptide produced in the liver from a larger precursor (reviewed by Nemeth and Ganz ). It circulates in the plasma and binds to the iron exporter ferroportin on the basolateral surface of absorptive enterocytes, causing ferroportin to be internalized and degraded in a ubiquitin-dependent manner. In this way, hepcidin controls cellular iron egress. Hepcidin thus regulates iron absorption at the level of the intestinal epithelium in that any iron unable to leave the enterocytes is lost when these cells turn over.

Expression of hepcidin is induced in response to iron overload and inflammation and is repressed in response to increased erythropoietic activity and hypoxia. Regulation of hepcidin levels appears to be at the level of gene transcription. However, the signals that govern hepcidin transcription in response to these stimuli are incompletely understood. Only the inflammatory cytokine interleukin-6 has been definitely shown to be involved in regulation of hepcidin expression by physiologic changes outside the liver.

Most of the total body iron is ultimately incorporated into hemoglobin in erythroid precursors. An average adult produces 200 billion red cells daily to achieve a red cell renewal rate of 0.8% per day. Each red cell contains more than 1 billion atoms of iron, and each milliliter of packed red blood cells contains approximately 1 mg of iron. To meet this daily need for 2 × 10 ²⁰ atoms (or 20 mg) of elemental iron, iron is recycled from senescent red cells and returned to the circulation. Plasma iron turnover (PIT) represents the mass turnover of transferrin-bound iron in the circulation (expressed as milligrams per kilogram per day). Accelerated erythropoiesis increases PIT and enhances iron uptake from the gastrointestinal tract. Although hepcidin appears to be the effector molecule that alters intestinal absorption, an upstream, circulating factor—the erythroid regulator—that communicates the iron needs of the erythron must exist. There are some data to indicate that circulating bone morphogenetic protein (BMP) signaling pathway inhibitors (e.g., GDF15 and TWSG) may play a role in mediating this effect.

This erythroid control of hepcidin expression is particularly apparent in patients with thalassemia intermedia (non–transfusion-dependent thalassemia), in whom marked iron overload develops even without transfusions. The accelerated, but ineffective, erythropoiesis substantially boosts iron absorption by inhibiting hepcidin expression. By contrast, hepcidin is increased in patients with thalassemia and other anemias receiving chronic transfusion therapy as a consequence of iron overload and the suppression of erythropoiesis.

Intercellular Iron Transport

Only a small proportion of total body iron enters and leaves the body each day. Consequently, intercellular iron transport is quantitatively more important than intestinal absorption. The greatest mass of iron is found in erythroid cells, which make up about 60% to 80% of the total body endowment in normal individuals. To accommodate ongoing erythropoietic needs, the reticuloendothelial system recycles approximately 20 to 25 mg of iron from effete red cells each day.

A small fraction, 0.1% or 4 mg, of total body iron circulates in plasma in an exchangeable pool. In normal individuals, nearly all of the circulating plasma iron is bound to transferrin. Transferrin serves three purposes: (1) it renders iron soluble under physiologic conditions; (2) it prevents iron-mediated free radical toxicity; and (3) it facilitates transport into cells. Transferrin is by far the most important physiologic supplier of iron to red cells. In fact, plasma transferrin serves to deliver iron to most tissues of the body. It is an 80-kd glycoprotein that has homologous N-terminal and C-terminal iron-binding domains. The molecule is related to several other proteins, including ovotransferrin in bird and reptile eggs, lactoferrin in extracellular secretions and neutrophil granules, and melanotransferrin (p97), a protein produced by melanoma cells. The functions of these related proteins are incompletely understood. The human transferrin messenger RNA (mRNA) is 2.3 kilobases (kb) in length and encodes 679 amino acids, including a 19–amino acid leader sequence. It is located on chromosome 3q, near genes for the transferrin receptor and melanotransferrin. The liver is the major site of synthesis and secretion of transferrin. Other tissues, including Sertoli cells of the testis, oligodendrocytes of the central nervous system, lymphocytes, muscle cells, and mammary cells, can, however, also produce the protein. Local synthesis within the brain and testis apparently provides transferrin for these tissues because serum transferrin does not penetrate their unique capillary barriers.

Transferrin production is regulated at multiple levels. Several cis -acting control regions exist upstream of the gene. The transferrin promoter contains binding sites for tissue-specific nuclear factors that activate transcription differentially in the liver and other tissues (e.g., Sertoli cells). Iron, hormones, and inflammatory stimuli also modulate transferrin gene expression. In the setting of iron deficiency, serum transferrin levels rise substantially as a result of enhanced synthesis of transferrin mRNA by the liver. In contrast, inflammation depresses levels of both serum transferrin and serum iron, the latter through the action of the iron regulatory hormone hepcidin. The plasma concentration of diferric transferrin has been proposed to modulate expression of the iron regulatory hormone hepcidin. Liver-derived serum transferrin levels fall modestly in patients with genetic or acquired iron overload, although mRNA levels have been reported to be unchanged. This suggests that hepatic transferrin is also controlled at the level of translation or secretion. The quantity of transferrin mRNA in nonhepatic tissues (including the testis, kidney, and spleen) is not affected by iron deficiency. The liver-derived transferrin, therefore, appears to have the unique responsibility of responding to iron status.

X-ray crystal structures have been determined for transferrin and related proteins (reviewed elsewhere ). All members of the transferrin protein superfamily exhibit similar polypeptide folding. The N-terminal and C-terminal domains are globular moieties of about 330 amino acids; each of these domains is divided into two subdomains, with the iron- and anion-binding sites located in the intersubdomain cleft on either side of a central plane of symmetry, suggesting an origin by gene duplication from a primordial protein containing a single iron-binding site. The binding cleft opens with iron release and closes with iron binding. The N-terminal and C-terminal binding sites are very similar. No cooperativity exists in binding of iron by the two sites, and the protein can be proteolytically cleaved into two halves, each of which retains iron-binding capability. Transferrin binds ferric iron much more avidly than ferrous iron.

The precise mechanism by which iron is loaded onto transferrin as it leaves intestinal epithelial cells or reticuloendothelial cells is unknown. The copper-dependent ferroxidase ceruloplasmin and its homologue hephaestin likely play central roles in reducing ferrous iron exported by ferroportin 1 (FPN1) from the reticuloendothelial cells and intestine, respectively, although neither is absolutely required due to the tendency of iron to oxidize in aqueous environments. Transferrin binds ferric iron exceptionally avidly, with a dissociation constant ( K _d ) of approximately 10 to 22 mol/L in the company of an anion (usually carbonate), which serves as a bridging ligand between the metal and protein and excludes water from two coordination sites. Without the anion cofactor, iron binding is negligible; with it, ferric transferrin is resistant to all but the most potent chelators. Release of iron from transferrin involves protonation of the carbonate anion to loosen the metal-protein association.

The sum of all iron-binding sites on transferrin constitutes the total iron-binding capacity (TIBC) of plasma. Thus, on a molar basis, TIBC is twice the concentration of transferrin protein because each transferrin molecule can bind two iron atoms. The laboratory method used to measure the TIBC can either be truly the total iron-binding concentration of the plasma measured by iron binding or a calculation based upon measurement of the concentration of transferrin. Whatever the method, under normal circumstances, about a third of transferrin iron-binding sites are filled. Consequently, except for the situation of extreme iron overload, in which all transferrin-binding sites are occupied, non–transferrin-bound iron (NTBI) in the circulation is present at very low concentrations. The distribution of plasma and tissue iron can be traced with the use of ⁵⁹ Fe as a radioactive tag by reinfusing a subject with autologous transferrin loaded with radiolabeled iron. Blood samples can be analyzed at timed intervals to determine the rate of loss of the radioactive label. Such ferrokinetic studies indicate that the normal half-life of iron in the circulation is about 75 minutes. The absolute amount of iron released from transferrin per unit time is the PIT (see earlier).

Such radioactive tracer studies indicate that at least 80% of the iron bound to circulating transferrin is delivered to the bone marrow and incorporated into reticulocytes. Other major sites of iron delivery include the liver and the spleen, which are the primary depots for iron storage. Hepatic iron is found in both the reticuloendothelial cells and hepatocytes. Reticuloendothelial cells acquire iron primarily by phagocytosis and breakdown of effete red blood cells; they catabolize heme and return iron to the circulation, where it binds transferrin. Hepatocytes take up iron by at least two different mechanisms, which are transferrin-dependent and transferrin-independent. The relative amounts of iron in each of these cell types depend on clinical circumstances, as discussed in detail later.

Given the preeminent role of bone marrow in the clearance of labeled iron from the circulation, ferrokinetic studies provide a window on erythropoietic activity. Conditions that augment erythrocyte production increase PIT. For example, hemolytic anemias such as hereditary spherocytosis and sickle cell disease induce rapid delivery of transferrin-bound iron to the marrow. In contrast, disorders that reduce red cell production, such as Diamond-Blackfan anemia and aplastic anemia, prolong PIT.

When erythrocytes are produced and released into the circulation in a normal fashion, the process of erythropoiesis is termed effective . In patients with certain anemias, however, the abnormal, nascent red cells apoptose or are destroyed before they leave the marrow cavity. In this situation, erythropoiesis is ineffective , which means simply that the erythropoietic precursors have failed to accomplish their primary task: delivery of mature, enucleated erythrocytes to the circulation. The ferrokinetic profile in this case shows rapid removal of iron from transferrin with delayed entry of label into the pool of circulating red cell hemoglobin. β ⁺ -thalassemia is an important example of this pattern. In β ⁺ -thalassemia, ineffective erythropoiesis is associated with a markedly enhanced PIT.

Intracellular Iron Metabolism

Transferrin and the Transferrin Cycle

Although transferrin was first characterized more than 60 years ago, its receptor eluded investigators until the early 1980s, when monoclonal antibodies prepared against tumor cells were found to recognize the transferrin receptor glycoprotein. Subsequently, receptor-mediated endocytosis of iron bound to transferrin has been characterized in detail. A diagram showing key features of the transferrin receptor is presented in Figure 11-3 . Each subunit of the disulfide-linked homodimer contains 760 amino acids. Oligosaccharides account for about 5% of the 90-kd subunit's molecular mass. Four glycosylation sites (three N linked and one O linked) are found in the protein. Glycosylation-defective mutants have fewer disulfide bridges, less transferrin binding, and less cell surface expression. The transmembrane domain, located between amino acids 62 and 89, functions as an internal signal peptide because there is none at the N-terminal end. A molecule of fatty acid (usually palmitate) is also covalently linked to each subunit at the internal edge of the transmembrane domain and may play a role in membrane localization. The crystal structure of the ectodomain of the transferrin receptor has been reported. Although no one has yet succeeded in cocrystallizing the receptor with transferrin, molecular modeling and binding studies of mutant transferrin receptor proteins have yielded important insights, including the amino acid residues important for receptor binding to transferrin.

Iron is taken into cells by receptor-mediated endocytosis of diferric transferrin ( Fig. 11-4 ). Receptors on the outer face of the plasma membrane bind iron-loaded transferrin with very high affinity. The C-terminal domain of transferrin appears to mediate receptor binding. Diferric transferrin binds with much higher affinity than monoferric transferrin or apotransferrin does. The dissociation constant ( K _d ) for bound diferric transferrin ranges from 10-7 to 10-9 mol/L at physiologic pH, depending on the species and tissue. The K _d of monoferric transferrin is approximately 10-6 mol/L. The concentration of circulating transferrin is about 25 µmol/L. Therefore, cellular transferrin receptors are ordinarily fully saturated.

After binding to its receptor on the cell surface, transferrin is internalized through a constitutive mechanism that begins with invagination of clathrin-coated pits and the formation of endocytic vesicles. This process requires the short, 61–amino acid intracellular tail of the transferrin receptor molecule. Receptors with truncated N-terminal cytoplasmic domains do not recycle properly. This portion of the molecule contains a conserved tyrosine-threonine-arginine-phenylalanine (YTRF) sequence that functions as a signal for endocytosis.

An unknown ATP-dependent proton pump lowers the pH of the internalized endosome to about 5.5. Acidification of the endosome weakens the association between iron and transferrin and promotes a conformational change in the transferrin receptor to enhance binding of apotransferrin and facilitate release of iron. An endosomal ferrireductase, STEAP3, reduces iron from the Fe ³⁺ state to Fe ²⁺ , either at the same time that it is released from transferrin or soon afterward. STEAP3 is a membrane-bound NADPH-dependent oxidoreductase localized to transferrin cycle endosomes and expressed at high levels in erythroid precursors. Related proteins may serve similar functions at other sites.

The iron released from transferrin must exit the endosome and enter the cytoplasm and mitochondria for use in heme biosynthesis, Fe/S cluster biosynthesis, and storage and for other purposes. This transmembrane transport step is also mediated by DMT1. DMT1 is unusual in that it is expressed on two very different types of membranes—the apical membrane of intestinal enterocytes and the endosomal membrane of transferrin uptake vesicles in nonpolarized cells. Once transported out of the endosome or across the plasma membrane, iron must be delivered to sites of use or stored in the form of ferritin. The cytosolic chaperone that traffics iron to ferritin appears to be PCBP1, a protein that also functions in the control of mRNA stability and translation, particularly in erythroid cells. In erythroid cells, iron is delivered across the mitochondrial inner membrane by mitoferrin 1 (MFRN1 or SLC25A37) ; once across the membrane it is incorporated into heme by the enzyme ferrochelatase.

The fate of transferrin and the transferrin receptor is distinct from that of iron. Rather than entering lysosomes for degradation, as do the ligands in other receptor-mediated endocytosis pathways, intact receptor-bound apotransferrin recycles to the cell surface, where the neutral pH promotes detachment into the plasma. Thus, preservation and reuse of transferrin are accomplished by pH-dependent changes in the affinity of transferrin for iron and the transferrin receptor. Exported apotransferrin binds additional iron atoms and undergoes further rounds of iron delivery to cells. Thus, a single transferrin molecule, with a half-life of 8 days, may be used hundreds of times for delivery of iron.

It was once a widely accepted belief that the transferrin cycle was important, if not necessary, for the uptake of iron by all mammalian cells. However, two lines of evidence indicate that such is not the case. First, mice and humans with severe deficiencies of plasma transferrin have iron deficiency anemia but vastly increased amounts of iron in nonhematopoietic tissues. This finding indicates that uptake of iron bound to transferrin is important for red blood cell hemoglobinization and maturation but that transferrin is not essential for delivery of iron to most other tissues. Second, mouse embryos lacking transferrin receptor die midway through gestation as a result of severe anemia, but most nonhematopoietic tissues appear generally intact. Aside from the erythron, only the developing central nervous system shows an apparent requirement for transferrin receptor. In its absence, primitive neuroepithelial cells undergo apoptosis at a greatly increased rate. Taken together, these observations indicate that the transferrin cycle is of primary importance in erythropoiesis and neurogenesis but of lesser importance in other mammalian tissues, at least early in embryonic development.

Ferritin

Once inside the cytoplasm, iron is almost certainly bound by carrier molecules that assist in delivering iron to intracellular locations, including mitochondria (for heme biosynthesis) and ferritin (for storage). The identities of the intracellular iron carrier molecules by and large remain unknown. However, there is recent evidence to suggest that poly-C–binding protein 1 (PCBP1), which is involved in mRNA stability and translation, also serves as an iron chaperone specific for iron acquisition by ferritin and the hypoxia-inducible factor (HIF) prolyl hydroxylase. The amount of iron in transit within the cell at any given time is small and difficult to measure. This minute pool of transit iron, which is believed to be in the Fe ²⁺ oxidation state, is the biologically active and potentially toxic form of the element. Metabolically inactive iron, stored in ferritin and hemosiderin, is nontoxic and in equilibrium with this exchangeable transit iron.

Both prokaryotes and eukaryotes produce ferritin molecules for iron storage. Mammalian ferritin molecules are complex 24-subunit heteropolymers of H (for heavy or heart) and L (for light or liver) protein subunits. L subunits are 19.7 kd in mass, with isoelectric points of 4.5 to 5; H subunits are 21 kd in mass, with isoelectric points of 5 to 5.7. The subunits of the ferritin molecule probably arose by gene duplication, although the degree of nucleotide sequence homology between the two is only about 50%. They assemble to form a sphere with a central cavity in which up to several thousand atoms of crystalline iron can be stored in the form of poly-iron-phosphate oxide. Eight channels through the sphere are lined by hydrophilic amino acid residues, and six more are lined by hydrophobic residues. Strong interspecies amino acid conservation is seen in the residues that line the hydrophilic channels, whereas marked variation is seen in those along the hydrophobic passages. The hydrophilic channels terminate with aspartic acid and glutamic acid residues and are lined by serine, histidine, and cysteine residues, all of which potentially bind metal ligands.

Although the two ferritin chains are homologous, only H ferritin has ferroxidase activity, which is essential for iron incorporation into the ferritin polymer. Isoelectric focusing shows that ferritin polymers have a variable composition of H and L subunits. These isoferritins, as they are called, show tissue-specific variation. Ferritin from liver, for instance, is rich in L subunits, as is that from the spleen. In contrast, the heart has ferritin rich in H subunits. Increased H-subunit content correlates with increased iron use, whereas increased L-subunit content correlates with increased iron storage. The ratio of H to L subunits increases with increased cell proliferation. Thus, ferritin provides a flexible reserve of iron.

Despite the fact that it does not, itself, have enzymatic activity, a mutation in L ferritin has been shown to cause a distinctive iron deposition disease in humans termed neuroferritinopathy . In patients carrying a nucleotide insertion altering the carboxy-terminus of L ferritin, iron overload develops in the basal ganglia and leads to a neurodegenerative disease with variable extrapyramidal features. By contrast, pathologic accumulation of apo-L-ferritin leads to a distinctive form of hereditary cataracts (hereditary hyperferritinemia-cataract syndrome) that is unassociated with iron metabolism abnormalities (see below). A number of other disorders, including aceruloplasminemia (see below), as well as pantothenate kinase deficiency, can also lead to pathologic iron deposition in distinctive regions of the brain.

Ferritin molecules aggregate into clusters that are engulfed by lysosomes and degraded. The end product of this process, hemosiderin, is an amorphous agglomerate of partially denatured protein and lipid interspersed with iron oxide molecules. In cells overloaded with iron, lysosomes accumulate large amounts of hemosiderin, which can be visualized by Prussian blue (Perls) staining. Although the iron enmeshed in this insoluble compound constitutes an end-stage product of cellular iron storage, it remains in equilibrium with cytosolic ferritin. Ferritin iron, in turn, is in equilibrium with iron complexed to low–molecular-weight carrier molecules. Therefore, introduction of an effective chelator into the cell captures iron from the low–molecular-weight “toxic” iron pool, draws iron out of ferritin, and eventually depletes iron from hemosiderin as well, though only very slowly. As might be expected, the bioavailability of hemosiderin iron is much lower than that of iron stored in ferritin.

The large number of processed pseudogenes that exist for each subunit initially confounded pinpointing the chromosomal location of ferritin genes. Functional ferritin genes for the H and L subunits are located on human chromosomes 11 and 19, respectively. In addition, there is an intronless gene encoding a mitochondrial, H-like ferritin molecule located on human chromosome 5q. The function of mitochondrial ferritin remains uncertain, although the fact that its expression is increased in erythroid precursors from patients with sideroblastic anemia suggests that it is involved in mitochondrial iron management.

Ferritin formation is controlled at multiple levels—transcription, message stabilization, translation, and subunit assembly. In the liver and in HeLa cells, iron rapidly induces the synthesis of L-subunit mRNA, with no effect on H-subunit transcription. In contrast, induced differentiation of HL-60 promyelocytic leukemia cells and mouse erythroleukemia cells increases production of H-subunit mRNA. Tumor necrosis factor induces H-chain transcription in human myoblasts. Iron, heme, reactive oxygen species, and oxidative stress all enhance transcription and translation of the ferritin heavy chain (reviewed by Rouault ).

Cytoplasmic ferritin mRNA forms a stable complex with several proteins. Both iron and interleukin-1β enhance translation of ferritin mRNA. Influx of iron into cells shifts the message onto the ribosomes, thereby enhancing the synthesis of ferritin subunits. This translational control mechanism involves an RNA-protein interaction that links the expression of genes encoding ferritin, the transferrin receptor, enzymes of heme biosynthesis, DMT1, ferroportin, and other proteins. Zahringer, Baliga, and Munro initially showed that ferritin synthesis was regulated at the level of message translation. Comparison of the 5′untranslated regions of ferritin mRNA encoding both heavy and light chains showed striking conservation of a 28-bp sequence motif that was predicted to form a stable RNA stem-loop structure and was necessary for translational control of ferritin ( Fig. 11-5 , inset).

Subsequently designated the iron response element (IRE), this RNA stem-loop is recognized by at least two specific RNA-binding proteins called IRP1 and IRP2 for iron regulatory proteins 1 and 2. IRP1 is a 98-kd soluble polypeptide with homology to the mitochondrial tricarboxylic acid–cycle enzyme aconitase (ACO1). Indeed, IRP1 is the cytoplasmic isozyme of aconitase. Like mitochondrial aconitase, the enzymatic active site of IRP1 contains a 4Fe/4S cluster. When the iron concentration in the cytosol is high, the cluster is complete and the protein is in its holoenzyme form and possesses aconitase activity and cannot bind mRNA. Under these conditions, the ferritin message is translated efficiently. Conversely, when the iron concentration is low, the protein is in its apo-form and aconitase activity is absent, RNA binding is avid, and ferritin message translation is blocked because translation initiation complexes cannot form properly. Thus, IRP1 is a dual-function protein, with ambient iron controlling the switch by assembly and disassembly of a 4Fe/4S cluster. Other molecules that modulate the switch include ascorbic acid, nitric oxide, and heme, all of which contribute to the oxidation status of the cell. This mechanism allows ferritin protein expression to be rapidly downregulated without degradation of mRNA. The importance of IRP-mediated translational control is underscored by the fact that plasma hyperferritinemia and ferritin-containing cataracts develop in patients with mutations in the L-ferritin IRE. Nevertheless, these patients with hereditary hyperferritinemia cataract syndrome do not have abnormalities in iron homeostasis. Furthermore, it appears that an iron overload disorder resembling hereditary hemochromatosis develops in rare patients with mutations in the H-ferritin IRE, presumably because increased H-ferritin expression leads to increased cellular iron assimilation or retention, or both.

A second IRE-binding protein, IRP2, also modulates posttranscriptional expression of mRNA by binding to IREs in much the same way. Initially isolated in the liver, IRP2 has been found in all tissues examined and is more abundant than IRP1 in some. IRP2 lacks aconitase activity and does not contain an Fe/S cluster to function as an iron sensor. Rather, IRP2 protein levels are modulated directly by iron or by heme and the molecule is actively degraded when iron or heme (or both) is abundant. Recently, it has been shown that an E3 ubiquitin ligase complex that contains FBXL5 controls IRP2 protein abundance by targeting it for proteasomal degradation. FBXL5 contains an iron- and oxygen-binding hemerythrin domain that acts as a ligand-dependent regulatory switch that mediates the stability of FBXL5 under iron- and oxygen-replete conditions.

The Transferrin Receptor

IRPs also modulate expression of the transferrin receptor gene. The gene encoding the transferrin receptor is located on human chromosome 3q26.2-qter, near the gene encoding transferrin at 3q21. It consists of 19 exons spread over 31 kb. The transferrin receptor mRNA is approximately twice the length needed to encode the receptor protein. Its lengthy 3′untranslated region contains five RNA stem-loop structures that are structurally similar to those of the ferritin IREs. IRP binding to these sites increases the stability of the transferrin receptor message by obscuring an endonucleolytic cleavage site (see Fig. 11-5 ). The result is a larger amount of transferrin receptor mRNA in the cell. A deficit of cellular iron raises transferrin receptor mRNA levels at least in part through enhanced message stability. However, IRE regulation is probably not responsible for the primary increases in transferrin receptor number in erythroid cells, because this occurs at the transcriptional level and is unaffected by iron status. Relatively little is understood about the tissue-specific transcriptional regulation of the transferrin receptor.

The versatile regulatory functions of IREs may also confer iron-dependent regulation on other proteins. IREs are present in the 5′ untranslated regions of the mRNA encoding the erythroid form of the heme biosynthetic enzyme δ-aminolevulinic acid synthase, mitochondrial aconitase, and ferroportin. A single 3′ IRE is found in one splice isoform of DMT1 mRNA. As the rate-limiting enzyme in heme biosynthesis, the IRE in the erythroid form of δ-aminolevulinic acid synthase produces a conceptually satisfying link between iron and heme production. However, the functions of the IREs in the other mRNAs are not yet understood. It is postulated that all IREs serve the common purpose of coupling changes in the iron status of the cell with its ability to use and store the element (see Fig. 11-5 ). When the intracellular iron concentration is low, the IRE of ferritin mRNA binds IRP and reduces ferritin synthesis because additional iron storage capacity is not needed in this circumstance. Simultaneously, the level of transferrin receptor mRNA increases as the IRP stabilizes the message, thereby increasing expression of the transferrin receptor. Together, these events increase the flow of iron into cells while protecting against iron toxicity. When iron levels in the cell are high, the opposite scenario is operative. Although the IRPs accomplish these regulatory feats under extremes of iron status, the importance of IREs and IRPs in modulating gene expression in normal humans and animals is not fully understood. Recently, novel IRE-containing genes have been identified, with roles less directly related to iron homeostasis.

In an attempt to sort out the unique functions of IRP1 and IRP2 and to evaluate their roles in vivo, targeted mutagenesis was used to generate mutant mice lacking each of the proteins. Initial characterization of IRP1 knockout mice indicated that they had no identifiable defects, but more recent studies have shown that the animals have a transient polycythemia that is associated with an elevated erythropoietin level due to dysregulation of HIF-2α translation, providing further evidence for crosstalk between hypoxia-sensing and iron-regulatory pathways. IRP2 knockout mice have abnormalities in several tissues. Initially, the most striking aspect of the IRP2 knockout phenotype was a late-onset neurologic disorder associated with abnormal deposition of iron in white matter tracts and nuclei throughout the brain. The mice demonstrated ataxia, bradykinesia, and tremors. However, these findings could not readily be explained by known activities of IRP2, and a second group reported that an independently targeted line of mice did not have obvious neurodegeneration. It is possible that the discrepancy may be explained by differential severity of a neurologic phenotype on different genetic backgrounds. More recently, both groups have reported that IRP2-deficient mice have mild microcytosis and erythropoietic protoporphyria as well as subtle perturbations in systemic iron homeostasis.

Serum Ferritin and Soluble Transferrin Receptors

Although most ferritin is located within cells, a measurable amount of the protein exists in serum. Intracellular concentrations, particularly in the liver, are several orders of magnitude higher than serum concentrations. Therefore, a small amount of cellular lysis could liberate a relatively large amount of ferritin. However, nearly all serum ferritin appears to be secreted. There are probably several cell types that are sources of secreted extracellular ferritin, including iron-recycling macrophages and hepatocytes. Circulating ferritin consists almost exclusively of L-chain subunits. In contrast, intracellular ferritin contains a mixture of H and L subunits. In addition, unlike intracellular ferritin, circulating ferritin is glycosylated, thus suggesting that it passes through the endoplasmic reticulum and Golgi apparatus in a manner similar to that of other secreted proteins.

Serum ferritin levels decline with iron deficiency and rise with iron overload. In the absence of severe liver disease, infection, or chronic inflammation, serum ferritin is roughly proportional to total body iron stores. The correlation between serum ferritin levels and body iron stores is useful in the evaluation of patients with possible iron deficiency or iron overload. A low serum ferritin level (<12 µg/L) invariably represents iron deficiency, and high serum ferritin levels are found in patients with iron overload. Extremely high serum ferritin levels should be interpreted cautiously, however, because the correlation between ferritin levels and body iron stores is approximately linear only for storage reserves of iron ranging between 1 and 3 g. In addition, normal circulating ferritin values vary with sex and age. These considerations must be factored into any evaluation of iron stores based on ferritin values, particularly in children.

Various conditions modify serum ferritin levels. Inflammation increases the serum ferritin concentration severalfold. Infections, particularly chronic conditions such as tuberculosis or osteomyelitis, may also increase levels substantially, presumably because of the associated inflammatory response. Autoimmune disorders, chronic renal disease, and chronic liver disease are likewise associated with elevated levels. A number of tumors are variably associated with an increased level of circulating ferritin. For example, ferritin is an important prognostic factor in childhood neuroblastoma, in which serum ferritin levels correlate with disease severity.

The mechanisms by which inflammation and tumors increase the quantity of plasma ferritin are not fully understood. Treatment with interleukin-1β, a prime mediator of the inflammatory response, increases syn­thesis of ferritin in human hepatoma cells, whereas tumor necrosis factor has been shown to increase ferritin mRNA levels in murine cells in culture. These cytokines may also increase ferritin synthesis in cells secreting the protein. In addition, the inflammatory cytokine interleukin-6 induces hepcidin expression, which results in increased macrophage iron storage and probably contributes to increased serum ferritin levels. Furthermore, serum ferritin elevated out of proportion to iron stores and the transferrin saturation is a hallmark feature of ferroportin disease (see later).

Soluble transferrin receptors are also found in serum. Small vesicles containing transferrin receptors are shed from reticulocytes during their maturation to erythrocytes. In addition to these vesicle-associated receptors, the extracellular portion of the transferrin receptor lacking its transmembrane anchor can be found in the circulation. One mechanism by which soluble transferrin receptors are generated is a membrane-associated protease activity that clips the molecule between amino acids 100 and 101, at the base of its extracellular stem. This cleavage is potentiated by mutation of an O -linked glycosylation site at amino acid 104, thus suggesting that differential glycosylation may play a regulatory role. The transferrin-binding domain is intact in these soluble receptors, and they are likely to be complexed with transferrin in serum.

Because maturing red cells shed their transferrin receptors, the amount of soluble transferrin receptor in plasma reasonably reflects the degree of erythropoiesis. Measurement of plasma levels of soluble transferrin receptor provides a means of estimating erythropoietic activity that is simpler than the relatively cumbersome PIT determination. Soluble transferrin receptor (sTfR) is present in substantially lower amounts in patients with aplastic anemia relative to normal individuals. In contrast, values in patients with anemia caused by ineffective erythropoiesis are markedly increased. Patients with iron deficiency also have increased levels of circulating soluble transferrin receptors. This may result in part from the increase in cellular transferrin receptor expression produced by iron starvation and in part from the increased erythroid turnover associated with the ineffective erythropoiesis of iron deficiency. sTfR and serum ferritin values can be considered together as the ratio of sTfR to the log of ferritin (sTfR-F index). Values greater than 1.5 suggest iron deficiency alone or in combination with an inflammatory condition; values less than 1.5 are characteristic of the anemia of chronic inflammation. The sTfR-F index also appears to be sensitive enough to detect iron deficiency before iron-restricted erythropoiesis is clinically apparent. sTfR and the sTfR-F index are both decreased in patients with iron overload.

Iron Deficiency