Article Text

This article has a correction. Please see:

Recent advances in understanding haemochromatosis: a transition state
Free
  1. K J H Robson1,
  2. A T Merryweather-Clarke1,
  3. E Cadet2,
  4. V Viprakasit1,3,
  5. M G Zaahl4,
  6. J J Pointon5,
  7. D J Weatherall1,
  8. J Rochette2
  1. 1MRC Molecular Haematology Unit, Weatherall Institute of Molecular Medicine, Headley Way, Oxford, OX3 9DS, UK
  2. 2Génétique Médicale-CHU, Faculté de Médecine, Université de Picardie Jules Verne, UMR-INERIS, Amiens, France
  3. 3Department of Paediatrics, Faculty of Medicine, Siriraj Hospital, Mahidol University, Bangkok, Thailand
  4. 4Department of Genetics, University of Stellenbosch, Stellenbosch, South Africa
  5. 5Institute of Musculoskeletal Sciences, University of Oxford, The Botnar Research Centre, Nuffield Orthopaedic Centre, Windmill Road, Headington, Oxford, OX3 7LD, UK
  1. Correspondence to:
 Kathryn J H Robson
 MRC Molecular Haematology Unit, Weatherall Institute of Molecular Medicine, Headley Way, Oxford, OX3 9DS, UK; kathryn.robsonimm.ox.ac.uk
 Correspondence to:
 Alison T Merryweather-Clarke
 MRC Molecular Haematology Unit, Weatherall Institute of Molecular Medicine, Headley Way, Oxford, OX3 9DS, UK; alison.hammerimm.ox.ac.uk

Abstract

Mutations in the hepcidin gene HAMP and the hemojuvelin gene HJV have recently been shown to result in juvenile haemochromatosis (JH). Hepcidin is an antimicrobial peptide that plays a key role in regulating intestinal iron absorption. Hepcidin levels are reduced in patients with haemochromatosis due to mutations in the HFE and HJV genes. Digenic inheritance of mutations in HFE and HAMP can result in either JH or hereditary haemochromatosis (HH) depending upon the severity of the mutation in HAMP. Here we review these findings and discuss how understanding the different types of haemochromatosis and our increasing knowledge of iron metabolism may help to elucidate the host’s response to infection.

  • ACD, anaemia of chronic disease
  • AD, Alzheimer’s disease
  • β2M, β2-microglobulin
  • Cp−/− mice, ceruloplasmin null mice
  • JH, juvenile haemochromatosis
  • HH, hereditary haemochromatosis
  • IEL, intra-epithelial lymphocytes
  • IL-6, interleukin 6
  • IRE, iron responsive element
  • MHC, major histocompatibility complex
  • NF-κB, nuclear factor κB
  • NRAMP1, natural resistance associated macrophage protein 1
  • Pcm, polycythaemic mice
  • PKAN, pantothenate kinase associated neurodegeneration
  • SLC11A1, solute carrier family 1 member 1
  • SLC11A2, solute carrier family 1 member 2
  • TfR2, transferrin receptor 2
  • TFR2, transferrin receptor 2 gene
  • ferroportin
  • haemochromatosis
  • hemojuvelin
  • hepcidin
  • HFE

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Mutations in the hepcidin gene HAMP and the hemojuvelin gene HJV have recently been shown to result in juvenile haemochromatosis (JH), whereas digenic inheritance of mutations in HFE and HAMP can result in either JH or hereditary haemochromatosis (HH). These findings are discussed below.

HLA LINKED HAEMOCHROMATOSIS AND HFE

In his monograph in 1935 Sheldon suggested that haemochromatosis might have a genetic basis.1 It was later found to be an autosomal recessive disease.2 After Marcel Simon demonstrated linkage of the haemochromatosis locus to the MHC class 1 molecule HLA-A3,3 it was presumed that hereditary haemochromatosis (HH) was a simple, autosomal recessive disorder restricted to people of north west European origin. In 1996, Feder and colleagues identified the HFE gene and showed that over 80% of patients were homozygous for the C282Y mutation in this gene.4 Other groups have subsequently confirmed that the majority of haemochromatosis patients are homozygous for the C282Y mutation in HFE.5–7 Among the remainder, over 75% of those who carry a single copy of the C282Y mutation are also compound heterozygotes for the more prevalent but mild H63D mutation in HFE (reviewed by Merryweather-Clarke et al8). Population studies have demonstrated that the C282Y mutation is restricted to people of north west European origin.8,9 Other, rarer HFE mutations and variants have been reported (fig 1)(reviewed by Pointon et al10), the majority of which have been found in conjunction with the C282Y mutation. However, a Vietnamese patient has been described who is homozygous for a novel splice site mutation.11 Another mutation appears to result in an autosomal dominant form of haemochromatosis (478delC)(Pointon et al, manuscript in preparation). It should be noted that Hfe−/− mice have a much more severe phenotype than HfeCys/Cys mice12 and this may help explain the severity of iron loading in these patients. Haemochromatosis due to mutations in HFE is known as type 1 haemochromatosis (OMIM 235200). More males than females are affected. Onset is usually in the fourth decade for men and in the fifth for women.

Figure 1

 Representation of the HFE/β2-microglobulin complex adapted from Feder et al4 illustrating the relative positions of the different amino acid mutations and sequence variants in HFE.4,10,42,156–170 The approximate positions of the two splice site mutations relative to the appropriate exons have also been included.11,171 The R6S mutation is in the signal sequence.172 The HFE protein is 343 amino acids long. The mature protein is 321 amino acids comprising three extracellular domains, α1, α2, and α3, with a single membrane spanning domain and a short cytoplasmic tail. It interacts with β2-microglobulin (β2M) that is essential for the correct assembly of the HFE/β2M complex. Unlike classical class I molecules there is no peptide binding groove between the α1 and α2 domains. The α3 domain is key to the interaction with β2M. The four invariant cysteine residues form two disulphide bridges, one in the α2 domain and one in the α3. It is the C282Y mutation found in the majority of HH patients that disrupts this second disulphide bridge. The correct assembly of HFE/β2M results in cell surface presentation. The C282Y form of HFE becomes stuck in the endoplasmic reticulum and middle Golgi.173,174

The original paper describing the cloning of the HFE gene reported that the mRNA encoding HFE was most abundant in the liver, and that low levels of HFE mRNA were found in many other tissues.4 Subsequently, there has been a series of studies describing the use of antibodies raised against HFE peptides or bacterially expressed HFE for the localization of HFE in different tissues. Histochemical techniques have localized HFE to a number of different cell types, including Kupffer cells,13 the gastrointestinal tract,14 placenta,15 epithelial cells and macrophages and monocytes,16 and crypt cells of the small intestine17 although some reports conflict. A variety of monoclonal and polyclonal anti-peptide antibodies have been used in these studies. One problem with the work on human material has been the lack of quantitation of HFE in the tissues examined. This has been overcome in a very recent study that examined the expression of iron metabolism related mRNAs in the rat using quantitative PCR, in situ hybridization, and western blot analysis.18 The study showed that HFE is expressed primarily in hepatocytes, and at 10-fold lower levels in Kupffer cells.18

The cloning of the HFE gene paved the way for the identification of non-HFE related forms of haemochromatosis, due to identification of mutations in a number of other genes.19–23 Analysis of these genes suggests that the pathway controlling the availability and uptake of iron is more complex that we had once thought. A list of genes with mutations leading to iron storage diseases together with their chromosomal locations, corresponding proteins, and previous names, is shown in table 1. Neonatal haemochromatosis may also have a hereditary component.24

Table 1

 Examples of some of the genes in which mutations give rise to inherited forms of iron overload including haemochromatosis

OTHER INHERITED FORMS OF HAEMOCHROMATOSIS AND IRON OVERLOAD INCLUDING JUVENILE HAEMOCHROMATOSIS AND FERROPORTIN DISEASE

Juvenile haemochromatosis (JH), or type 2 haemochromatosis, is an autosomal recessive disease due to mutations in either the hepcidin (HAMP)22 or hemojuvelin (HJV)23 genes.

The clinical manifestations of JH and HH have been compared in detail.25 Equal numbers of males and females are affected by JH, and cardiac involvement and hypogonadism are frequent in JH.

JH is a more severe form of iron loading disorder than HH, being characterised by an earlier age of onset, usually before the age of 30. A particularly severe form of JH is caused by autosomal inheritance of mutations in HAMP(OMIM 606464).22 The HAMP gene maps to 19q13 and comprises three exons; the processed hepcidin peptide is coded entirely by exon 3.26 The first two JH mutations reported disrupt the reading frame for exon 3.22 The first, 166C→T, results in a premature termination codon R56X, and the second is a frameshift mutation (93delG) that alters the reading frame of exon 3.22 The hepcidin peptide has four disulphide bonds,27 the third of which is disrupted by a missense HAMP mutation C70R resulting in JH.28 High levels of hepcidin result in anaemia29,30 and hyperferraemia.31–35

A second gene, HJV, has recently been identified that is involved in the same pathway as hepcidin. HJV mutations also result in JH, with an autosomal recessive mode of inheritance.23HJV maps to 1q2136 and encodes hemojuvelin (OMIM 608374).23 JH due to mutations in HJV appears to be less severe than that due to HAMP mutations.22 As is seen with mutations in the transferrin receptor 2 gene (TFR2),37–43 there is striking allele heterogeneity in JH due to mutations in HJV, with the majority of the mutations being private.23,44–46 Most mutations generate premature termination codons or are missense substitutions affecting conserved amino acid residues. The 24 mutations described to date are summarized in table 2. The protein sequence of hemojuvelin has no hallmarks that explain its role in iron metabolism. Its expression is restricted to liver, heart, and skeletal muscle. Hemojuvelin is predicted to be a transmembrane protein, having both an Arg-Gly-Asp (RGD) motif and a partial von Willebrand factor type D domain. Many RGD motifs occur in cell surface proteins that interact with integrins in protein–cell or cell–cell interactions.47 Hemojuvelin undergoes alternative splicing,23 the smallest of the spliced variants lacking the RGD motif. The longest isoform encodes a protein of 426 amino acids and shows sequence similarity with the repulsive guidance molecule.23 Urinary hepcidin levels are depressed in individuals with HJV linked JH, suggesting that hemojuvelin modulates hepcidin expression.23 Similar observations have been made for haemochromatosis expressing C282Y homozygotes.33,35HAMP mRNA levels in the liver of HH patients are inappropriately low for body iron stores and urinary levels are correspondingly low.33,34

Table 2

 HJV mutations that have been identified in patients with JH

In southern Italy, where a minority of haemochromatosis patients are C282Y homozygotes, genome-wide screening of affected families led to the identification of another rare form of haemochromatosis caused by autosomal inheritance of mutations in the gene for TFR2.19 This is known as haemochromatosis type 3 (OMIM 604250) and is due to a number of private mutations.37–43 The first mutation, Y250X, was identified in two Sicilian families who were homozygous for a C→G transversion at nucleotide 750 in the coding sequence of the mRNA.19 A second nonsense mutation, missense and frameshift mutations have now been reported: E60X,37 M172K,37 R455Q,39 Q690P,38 V221I,42 and the deletion of AVAQ (residues 594–597).39,40,43 The clinical phenotype is similar to that of type 1 haemochromatosis, with iron deposition in the liver having a periportal distribution. However, the clinical features vary depending upon the reported mutation.37–43

Interestingly, transferrin receptor 2 (TfR2) is expressed in hepatocytes48 and in duodenal crypt cells17 as well as in platelets.49 Griffiths and Cox have demonstrated that HFE and TFR2 colocalise in duodenal crypt cells, and that within the crypt areas the distribution of TFR2 differed from that of transferrin receptor 1.17 Little or no HFE was present in the duodenal crypt cells of an HH patient and unlike the observed intracellular localisation of wildtype HFE, TFR2 localised to the basolateral surface of the duodenal crypt cell. These authors suggest that the interaction of HFE with TFR2 within this vesicular compartment was stimulated by the presence of holotransferrin (iron loaded transferrin).17 TFR2 plays a role in cellular iron transport although it has a 25-fold lower affinity for transferrin relative to that of TFR1.50

Type 4 haemochromatosis (OMIM 606069) is an autosomal dominant form of iron overload associated with mutations in the gene encoding ferroportin/IREG1/MTP1 (SLC40A1, previously SLC11A3).20,21,51 The first description of this form of haemochromatosis was in a large kindred from the Solomon Islands.52 Mutations in SLC40A1 have a wide geographical distribution.53–58 One in particular, 162delVal, has been found in six unrelated families.59–62 Ferroportin is a membrane spanning transporter that exports iron from cells and is predicted to have nine or more transmembrane domains. Ferroportin is predominantly found in Kupffer cells and the basolateral membrane of the duodenal enterocyte.63–65 It is also expressed in reticulo-endothelial macrophages, placental syncytiotrophoblasts, and hepatocytes.

The majority of ferroportin mutations localize to the external face of the protein in the model of Devalia et al,59 many of them mapping to the extracellular loop between transmembrane domains 3 and 4 (fig 2). It has been suggested that ferroportin expressed in the membrane of macrophages may interact with caeruloplasmin, a ferroxidase, to provide Fe3+ for transferrin.66,67 The accumulation of iron by macrophages in patients with type 4 haemochromatosis suggests some loss of ferroportin function, possibly through a failure to interact with another protein such as caeruloplasmin. This is consistent with the phenotype of the ceruloplasmin null (Cp−/−) mouse in which iron accumulates in the Kupffer cells68 in a manner similar to that seen in patients with ferroportin mutations. Hephaestin plays an important role in intestinal iron absorption and is predicted to be a ferroxidase based on significant sequence identity to the serum multi-copper ferroxidase, ceruloplasmin.69 A recent report has shown that, in the rat, hephaestin is expressed both in the enterocyte and the macrophage, suggesting that caeruloplasmin may not be required as the ferroxidase in macrophages.18 Other mutations elsewhere in ferroportin may have the reverse effect, resulting in loss of regulation and enhanced export of iron from the macrophage (Viprakasit et al, unpublished observations). Disruption of ferroportin gene regulation causes dynamic alterations in iron homeostasis and erythropoiesis in polycythaemic mice.70 A 58 bp microdeletion in the Slc40a1 promoter, identified in radiation induced polycythaemic mice (Pcm), altered the transcription start sites and eliminated the iron responsive element (IRE) in the 5′ untranslated region.70 This resulted in increased duodenal and hepatic ferroportin protein levels during early postnatal development. Pcm mutant mice were iron deficient at birth and went on to develop reticuloendothelial iron overload as young adults.70 They also showed an erythropoietin dependent polycythaemia in heterozygotes and a hypochromic, microcytic anaemia in homozygotes.70 It was also observed that the defects in erythropoiesis were transient and that the delayed upregulation of hepcidin during postnatal development correlated with increased ferroportin levels and polycythaemia in Pcm heterozygotes.70

Figure 2

 Schematic representation of the structure of ferroportin, based on the model proposed by Devalia et al.59 The filled circles correspond to the cysteine residues. The relative positions of the reported mutations are indicated by open circles.20,21,53–62,95 A third mutation at amino acid position 144 has just been reported associated with parenchymal loading and cirrhosis.175 We include two novel unpublished mutations, C326Y (Viprakasit et al, manuscript in preparation) and D270V (Zaahl et al, manuscript in preparation). The length in amino acids of the predicted loops and the position of three potential glycosylation sites (Y) are indicated.

These findings will no doubt modify current models of how HFE mutations result in iron loading.67,71–73 In the model proposed by Townsend and Drakesmith HFE has two mutually exclusive activities in cells: inhibition of both uptake and release of iron from cells.71 The balance between transferrin saturation and transferrin receptor concentration would determine which of these two functions predominates. The inhibition of iron release may involve ferroportin. HFE present in the duodenal crypt cells and the reticuloendothelial system interprets the body’s need for iron and thus controls iron absorption by the gut enterocyte and its subsequent distribution. Hence mutations in HFE would be predicted to result in over-absorption of dietary iron or its distribution.

Other rare forms of inherited disorders of iron metabolism have been reported. These include mutations in the IRE of H-ferritin,74 and mutations in the IRE of L-ferritin which result in the hyperferritinaemia-cataract syndrome.42,54,75–78 This syndrome has been reviewed recently.79 It is associated with low serum iron and elevated serum ferritin levels. Acaeruloplasminaemia results from mutations in caeruloplasmin.80,81 This disease is characterised by iron accumulation in the brain and visceral organs; the involvement of the central nervous system distinguishes it from other inherited iron storage disorders. Acaeruloplasminaemia is associated with anaemia and low serum iron levels (for reviews see Gitlin82 and Miyajima et al83). A total of 21 mutations in the caeruloplasmin gene have been identified in 24 families worldwide.82 Neurological problems such as ataxia and dementia are common, suggesting that iron efflux from storage sites within the CNS requires caeruloplasmin. The latter has a direct role in mobilising iron from parenchymal tissues by oxidising Fe2+ to Fe3+, enabling it to bind to circulating apo-transferrin.83

POPULATION GENETICS AND SCREENING

The genetics of haemochromatosis is complicated by incomplete penetrance and digenic inheritance. Studies of various mouse models suggest that the inheritance of iron loading is more complex than was initially believed. A number of human population studies have demonstrated that there is a direct relationship between the HFE C282Y genotype and transferrin saturation.84–86 Transferrin saturation reflects the amount of transport iron that is available to cells via uptake through the transferrin receptor cycle.87 An unexplained raised transferrin saturation is usually an early biochemical indicator of haemochromatosis. The normal range has been the subject of debate. For men it is usually between 15–55% and for women 15–50%. The normal serum iron concentration is 10–30 μmol/l. Patients who have haemochromatosis due to mutations in the HFE gene have transferrin saturations above 55% and also have a raised serum ferritin concentration (men >300 μg/l, women >200 μg/l).88 However, not all C282Y homozygotes will show signs of iron loading. The age of onset is variable and is on average a decade later in women compared to men due to blood loss during menstruation and pregnancy. The presenting features are variable and lead to difficulties in diagnosis. Disease penetrance in haemochromatosis has been the subject of much recent discussion in the literature.89–93

It appears that disease is associated with a raised serum ferritin concentration, suggesting that there are genetic modifiers that contribute to increased serum ferritin concentrations in the presence of a raised transferrin saturation. This is supported by two observations. Firstly, mutations in ferroportin result in a phenotype that differs from that of HFE haemochromatosis. Patients with ferroportin mutations have a raised serum ferritin in the absence of a raised transferrin saturation. They rapidly become anaemic when treated by venesection unless they are also heterozygous for the H63D variant of HFE, and tend to have microcytic red cells. Iron loading in the liver is initially restricted to Kupffer cells, unlike that of HFE haemochromatosis in which iron is found predominantly in hepatocytes, with a periportal distribution; Kupffer cells are spared until the late phase of the disease progression. As a consequence, type 4 haemochromatosis has been referred to as “ferroportin disease”.94,95 Patients with HFE haemochromatosis tolerate venesection to mobilise their iron stores and their red cells tend to be macrocytic.96,97

The second observation suggesting that genetic modifiers influence serum ferritin concentration in the presence of increased transferrin saturation comes from work with mouse models of haemochromatosis. The degree of iron loading varies depending on the genetic background of the mouse.98,99 Results presented at Bioiron 2003 suggested that there are at least four independent loci contributing to iron loading in the mouse model.100

Unfortunately, the phenotypic markers for haemochromatosis lack specificity and sensitivity. Transferrin saturation and serum ferritin levels are best evaluated together. A raised serum ferritin is not unique to a patient with haemochromatosis. Serum ferritin concentrations can be increased and transferrin saturation decreased in both acute and chronic disease.101 Due to variability in the age of onset of disease in HFE haemochromatosis, biochemical tests need to be repeated on a regular basis to identify those in the presymptomatic phase. Genetic screening for haemochromatosis is not currently deemed an acceptable alternative to phenotypic screening because of the incomplete penetrance of the C282Y homozygous state. The acceptability of screening must also be considered, although it may be more acceptable to screen for haemochromatosis than is currently thought (Cadet et al, submitted). Genetic screening for haemochromatosis may become more valuable when disease penetrance is better understood and it may be more appropriate in populations with a higher degree of penetrance. Reports from South Wales and California suggest low penetrance whereas much higher penetrance is seen in the Picardy region of northern France.89,91,93,102,103 Environmental as well as genetic factors may help explain some of these differences.

In the future, screening for the C282Y mutation may have other advantages. Although the C282Y allele is not a high risk factor for diseases such as Alzheimer’s or cardiovascular disease in isolation, in women who smoke or who have hypertension, it increases the risk of developing cardiovascular disease.104 The risk of myocardial infarction and atherosclerosis is related to iron stores.105 The HFE C282Y mutation is also a risk factor for Alzheimer’s disease (AD) when present with the apolipoprotein E4 allele, a risk that is increased in the presence of the C2 allele of transferrin.106 Over 30 polymorphisms have been reported in the human transferrin gene.107 Of these, the C2 allele has a slightly reduced affinity for iron. If genetic screening for susceptibility to AD becomes a reality it may be worth including both HFE C282Y and TF C2 or routine monitoring of serum iron levels as part of a disease prevention programme. Increased circulating iron may induce the oxidative damage to lipids implicated in the genesis of common diseases such as cardiovascular disease and AD, both of which involve plaque formation. Iron misregulation is known to play a role in pantothenate kinase associated neurodegeneration (PKAN; previously known as Hallervorden-Spatz disease),108 and has been implicated in other neurodegenerative disorders.109–112

GENETIC MODIFIERS AND DIGENIC INHERITANCE

Clues as to the complexity of factors resulting in iron loading disorders have come from recent work investigating haemochromatosis patients who carry only one copy of the HFE C282Y mutation. We and others have found that some of them are heterozygous for mutations in HAMP.42,113,114 This suggests that haemochromatosis can result from digenic inheritance of mutations in HFE and HAMP. The severity of disease correlates with the type of HAMP mutation.113 We identified two different mutations in HAMP, the more severe being a four nucleotide deletion, ATGG, at the 3′ end of exon 2 (Met50del IVS2+1(−G)) which removes the splice site consensus GT but creates a second splice site consensus leading to a different open reading frame in exon 3.113HAMP encodes hepcidin, which is synthesized in the liver as a pre-propeptide and then processed to produce hepcidin. As the mature peptide is entirely encoded by exon 3,32 no hepcidin is expressed by the Met50del IVS2+1(−G) mutant allele, and heterozygosity for this and the C282Y mutation in HFE results in JH.113 Such patients presumably have half the functional levels of HFE and hepcidin, which act synergistically to produce disease. Double heterozygosity for the reported missense HAMP mutations (fig 3) and the HFE C282Y mutation results in a milder disease that resembles the classic adult onset HFE haemochromatosis.42,113,114 Consistent with these observations, hepcidin has been shown to act as a genetic modifier in the mouse model for haemochromatosis.115

Figure 3

 Schematic representation of the structure of hepcidin.27 The eight cysteine residues are indicated by black circles. The positions of the missense mutations G71D (black background) and C70R (dark grey) are indicated. The only amino acid with an acidic side chain in the mature peptide, D60 (white background) is also indicated. The remaining amino acids are represented by light grey filled circles.

Several groups have suggested that the ancestral haplotype carrying the C282Y mutation is associated with more severe disease.116–118 Others have not been able to confirm this observation.119 The ancestral haplotype extends into the major histocompatibility complex (MHC) class I region. The extended MHC region includes genes involved in modulating the immune response, one of which is tumour necrosis factor α. There are number of tumour necrosis factor α polymorphisms two which have been studied in relation to expression of HH.120 These findings suggest that tumour necrosis factor α plays a role in HH by modulating the severity of the liver damage.120 One of the problems encountered in identifying the HFE gene was that it occurred on an extended haplotype.4 It is not clear whether other genes on this extended haplotype have provided a selective advantage. It has been proposed that the MHC class I genes may themselves be genetic modifiers and so explain the extended linkage disequilibrium.121

We have reported that the 16189 variant of mitochondrial DNA may also act as a modifier.122 Its frequency in the normal UK population is 8.8%, increasing to 14.4% in disease expressing C282Y homozygotes and decreasing to 3.2% in non-expressing homozygotes.122

Studies such as these need independent confirmation in a second population. Sampling error can sometimes lead to false disease associations.

IRON, HOST DEFENCE, AND INFECTION

The complex and controversial literature on the roles of iron deficiency or overload in susceptibility to infection is reviewed elsewhere.121,123,124 It is not clear whether, in evolutionary terms, genetic diversity which favours increased iron absorption would have been advantageous. However, it is generally accepted that the hypoferraemia that occurs in acute infections, reflecting redistribution of iron rather than iron deficiency, may be an important mechanism for limiting iron availability for micro-organisms or parasites.

It is uncertain at present how mutations in some of the proteins described in this review result in iron overload. It is clear, however, that they are likely to be involved in a novel pathway of iron metabolism. That hepcidin is an acute phase protein33 and HFE an MHC class I-like molecule4 suggests that other players in the pathway may well be involved in a host defence pathway that limits iron availability and restricts growth of invading pathogens.

Two groups identified hepcidin as an antimicrobial peptide.26,125 Two separate approaches identified hepcidin as having a key role in iron homeostasis.31,32 Loss of hepcidin expression in the mouse leads to severe iron overload31 and in humans a severe form JH is due to autosomal recessive inheritance of mutations in HAMP.22 The human gene encodes a pre-propeptide of 84 amino acids,26 which after cleavage of the signal sequence yields a propeptide of 60 amino acids that is cleaved further to give the 25 and 22 amino acid forms of hepcidin which is found in serum125 and urine.26 It is these 22 and 25 amino acid forms that have antibacterial26 and antifungal activities.125 Hepcidin is a member of a family of proteins which includes the defensins, some of which have been shown to be involved in inflammation. Hepcidin is a cysteine-rich, cationic antimicrobial peptide and is a type II acute phase protein.33

Hepcidin regulates intestinal iron absorption.126 Elegant work by Nemeth and colleagues has shown that interleukin 6 (IL-6) mediates hypoferraemia of inflammation by inducing synthesis of hepcidin.127 This is a key finding as it helps explain the anaemia of chronic disease (ACD, also called the anaemia of inflammation). ACD is a hypoproliferative anaemia commonly found in patients with chronic infections, cancer, injury, and inflammatory disorders.128,129 Clinically, patients with ACD have a low serum iron (hyposideraemia) and reduced erythropoiesis due to low levels of erythropoietin. Inflammation increases retention of iron by the macrophage and restricts duodenal iron uptake. Hepcidin is key to this response. Mice lacking Hfe mount a general inflammatory response after injection of lipopolysaccharide but lack the appropriate hepcidin response and so do not develop hyposideraemia.130 Understanding the basis of ACD may help in developing new treatments.

HFE has a structure similar to MHC class I molecules.131 Importantly, these molecules form a heterodimer with β2-microglobulin (β2-M). When mutations in HFE were first proposed to be responsible for HH, supporting evidence came from the β2-M null mouse.132 The iron loading observed in these mice recapitulated the iron loading observed in HH. Mice lacking both Hfe and β2-M have significantly greater iron overload that mice lacking either gene alone.133

Other proteins involved in iron metabolism include DMT1, the divalent metal ion transporter,134 which was originally identified as a gene with homology to natural resistance associated macrophage protein 1 (NRAMP1) and so given the name NRAMP2.135NRAMP1 is now classified as solute carrier family 1 member 1 (SLC11A1) and NRAMP2 as solute carrier family 1 member 2 (SLC11A2). SLC11A1 is expressed in spleen, liver, and lungs, but is most abundant in monocytes/macrophages and polymorphonuclear leukocytes.136,137 The G169D mutation in the murine gene, Slc11a1, is associated with susceptibility to infection with intracellular parasites such as Leishmania, Mycobacteria, and Salmonella.138 SLC11A1 has a role in the transport of divalent metals and so restricts the availability of divalent metal ions to phagocytosed pathogens in the phagosome.139 Within the phagolysosome SLC11A1 is predicted to function in a pH dependent fashion, transporting divalent ions from the lumen of the phagolysosome thus restricting the growth of the pathogen.140

The iron transporter SLC11A2 has a major role in the transport of iron across the brush border into the enterocyte of the duodenum. It is also involved in transporting the iron released from transferrin in the lumen of recycling endosomes into the cytoplasm.141 Mutations in SLC11A2 result in anaemia.134

Further evidence suggesting that there is a pathway involving host defence and iron comes from the work of de Sousa and colleagues.142 They noted that the relative proportions of CD4+ and CD8+ T lymphocytes were abnormal in patients with HH. Those with abnormally high iron stores have CD4:CD8 >3 (normal: 0.61–1.77). The abnormal ratio, due to abnormally low numbers of CD8 cells, remains unchanged after treatment by phlebotomy.143 In patients whose abnormal ratios persist after intensive treatment, the transferrin saturation starts to increase earlier than in those whose CD4:CD8 ratios are closer to the normal range.144 Anomalies of the T cell receptor repertoire have also been described in C282Y carriers.145

Nuclear factor κB (NF-κB) is a transcription factor that plays a pivotal role in transactivation of genes involved in the hepatic acute phase response and innate and adaptive immunity.146 One of the genes that it regulates, IL-6, is known to activate hepcidin.33 As discussed above, hepcidin plays a significant role in iron metabolism.30,32 Endotoxin induced NF-κB activation in macrophages is inhibited by iron chelators, suggesting that intracellular iron plays a signalling role in NF-κB activation.146 Tumour necrosis factor α expression is also induced by NF-κB. A model proposed by Chorney et al73 suggests that HFE may either signal directly to the intra-epithelial lymphocytes (IEL) or induce the expression of cell-surface or secreted molecules that carry out the signalling process. They suggest that iron loaded enterocytes would be viewed by a neighbouring γδ IEL as being oxidatively stressed, in keeping with the role of γδ T cells in responding to tissue damage.147 The tumour necrosis factor α released could lead to an upregulation of ferritin expression and hence increased iron stores.148 This may provide yet another link between genes involved in iron metabolism and host immunity.

The haemoglobin scavenger receptor, CD163, is expressed on monocytes at low levels and at high levels on macrophages.149 IL-6, glucocorticoids, and IL-10 are acute phase mediators that strongly induce CD163 mRNA and its expression at the cell surface.150,151 LPS, IL-4, tumour necrosis factor α, and interferon-γ downregulate its expression.150,151 It has been suggested that haptoglobin-haemoglobin complexes may crosslink CD163 on the macrophage surface and so trigger a cascade that increases the production and secretion of anti-inflammatory cytokines.152 This may provide further clues as to IL-6 production resulting in hepcidin expression and ACD.127 CD163 may also play a role in atherosclerosis,153 again implicating iron in chronic disease.

HFE and hemojuvelin clearly operate upstream of hepcidin. Mice administered lipopolysaccharide have reduced levels of ferroportin mRNA and protein in the liver, spleen, and duodenum.154 Recent data suggest that hepcidin regulates ferroportin expression in the liver and intestine of the rat, suggesting that ferroportin is downstream of hepcidin in this pathway.155 Hepcidin could function by blocking the ferroportin linked iron export from enterocytes and macrophages. Some mutations in ferroportin would be predicted to alter this interaction, resulting in either enhanced or reduced export of iron from macrophages, as discussed above. The hepcidin receptor and the proteins involved in transcriptional control of ferroportin remain to be identified. From our experience there are still patients who have unexplained iron overload and may have mutations in other genes in the hepcidin pathway. Identification of these genes and their disease causing mutations will help fill the gaps in this pathway.

Clearly, the regulation of iron homeostasis and the inflammatory and immune responses are linked in a highly complex interactive system, many facets of which must have come under intense evolutionary pressure and which may show broad homology over many species. Hence they are likely to exhibit wide genetic heterogeneity, the further study of which may have important implications for a better understanding of both disorders of iron metabolism and their effects and the variability of response to infection among different ethnic groups.

Acknowledgments

The authors thank two reviewers for positive and constructive criticism.

REFERENCES

Supplementary materials

  • Publisher and Author Correction
    Please note that there are errors in Figures 1 and 2

    Figure 1
    Two amino acid mutations are incorrectly labelled A176C and R244G, the correct labels are:
    A176V and R224Q

    Figure 2
    One amino acid mutation is incorrectly labelled N114D, the correct label is:
    N144D

    The errors are much regretted.

Footnotes

  • Work in the laboratories of KJHR and JR has been supported by EC contract QLRT-1999-02237.

  • Conflict of interest: none declared.

Linked Articles

  • Correction
    BMJ Publishing Group Ltd