Modulation of Iron Availability at the Host-Pathogen Interface in Phagocytic Cells

JOHN FORBES, STEVEN LAM-YUK-TSEUNG, AND PHILIPPE GROS*

[SECTION EDITORS: FERRIC FANG AND MARTIN KAGNOFF]

June 28, 2006

Department of Biochemistry and Center for the Study of Host Resistance, McGill University, Montreal, Quebec, Canada H3G 1Y6

*Corresponding author. Phone: (514) 398-7291, Fax: (514) 398-2603, E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it

INTRODUCTION

Susceptibility to infectious diseases is determined by many factors, including pathogen virulence determinants, the genetic background of the host, and environmental conditions that can influence expression of the other two factors. Genetic analysis of laboratory mice has been an invaluable tool for the identification of host susceptibility determinants and pathogen-encoded virulence factors under controlled conditions. The discovery of a macrophage-specific metal transporter (Nramp1/ Slc11a1) that affects susceptibility to infection by several unrelated intracellular pathogens has focused attention on metal availability as a critical determinant of infection outcome. The involvement of many bacterial metal transport systems in virulence confirms the importance of divalent metals in the host-pathogen interaction. Here we summarize recent data on iron metabolism in macrophages, with a special emphasis on possible bacteriostatic and bactericidal consequences for intracellular pathogens. This review will include the role of biological chelators and transporters in normal macrophage physiology and antimicrobial defense.

OVERVIEW OF MAMMALIAN IRON METABOLISM

Iron is an essential metal cofactor for many biochemical pathways in mammals. However, excess iron promotes the formation of cytotoxic oxygen derivatives so that systemic iron levels must be tightly regulated. In mammals (Fig. 1), dietary iron entering the duodenum is largely in the form of insoluble Fe³⁺ complexes that must be converted to soluble Fe²⁺ by the cytochrome b-like ferrireductase (Dcytb) (106) prior to capture and transport into enterocytes by Nramp2 (DMT1/Slc11a2), a pH-dependent metal transporter that functions by a proton cotransport mechanism at the apical brush border (23, 44, 65). Fe²⁺ is subsequently exported from enterocytes by another membrane transporter called ferroportin (Ireg1, MTP1, Slc40A1) into the portal bloodstream for systemic distribution via the transferrin (Tf) / Tf receptor (TfR) system (1, 40, 68, 107). This process appears to require the conversion of Fe²⁺ back to Fe³⁺ by the ferroxidase hephaestin, which functions to facilitate iron release from intestinal cells (28, 89, 150). Upon reaching its destination, Tf-Fe³⁺ binds to Tf receptors (TfR) at the cell surface and is internalized via clathrin-dependent endocytosis to reach recycling endosomes (68). From there, iron is released from Tf after acidification of the endosomal lumen by the vacuolar H⁺-ATPase and converted into Fe²⁺ by an as yet unidentified ferrireductase. Fe²⁺ is then transported across the membrane of acidified endosomes into the cytoplasm by Nramp2 to be stored in ferritin (Ft) or transported into the mitochondria for biosynthesis of new heme or iron-sulfur cluster-containing proteins. This process is particularly active in erythroid cells where the need for iron is substantial.

Fig. 1Overview of mammalian iron transport. In mammals, dietary iron entering the duodenum is largely in the form of insoluble Fe³⁺ complexes that must be converted to soluble Fe²⁺ by the cytochrome b-like ferrireductase (Dcytb) prior to capture and transport into enterocytes by Nramp2 (DMT1/Slc11a2), via a pH-dependent proton cotransport mechanism at the apical brush border. Fe²⁺ is subsequently exported from enterocytes by ferroportin (Ireg1, MTP1, Slc40A1) into the portal bloodstream for systemic distribution via the transferrin (Tf)/Tf receptor (TfR) system. This process appears to require the conversion of Fe²⁺ back to Fe³⁺ by the ferroxidase hephaestin, which facilitates intestinal iron release. Upon reaching its destination (in this case a reticulocyte), Tf-Fe³⁺ binds to TfRs at the cell surface and is internalized via clathrin-dependent endocytosis to reach recycling endosomes. From there, iron is released from Tf after acidification of the endosomal lumen by the vacuolar H⁺-ATPase and converted into Fe²⁺ by another ferrireductase. Fe²⁺ is then transported across the membrane of acidified endosomes into the cytoplasm by Nramp2 to be stored in ferritin (Ft) or transported into the mitochondria for biosynthesis of new heme or iron-sulfur cluster-containing proteins. This process occurs in most cell types and is particularly active in erythroid cells, where the need for iron is substantial.

New iron is not only obtained from the diet, but also by recycling from erythrocytes within the reticuloendothelial system (68). Macrophages within the spleen, liver (Kupffer cells), and bone marrow appear critical for the latter process since they are responsible for phagocytosis and destruction of senescent red cells (68). The mechanism of iron recycling by macrophages including iron efflux from erythrocyte-containing phagosomes, iron release from macrophages, and entry into the Tf cycle remain poorly understood. The macrophage-specific iron transporter Nramp1 (see below) acts as a Fe²⁺ efflux pump at the phagosomal membrane and is therefore a candidate for involvement in this process (73). However, mice bearing loss-of-function Nramp1 mutations show no impairment of global iron homeostasis and are not anemic, suggesting that Nramp1 is not essential for this transport activity. Nramp2 is expressed in macrophages and has been reported to associate with phagosomes where it may efflux iron (71), but its functional role has not been demonstrated in this context. Ferroportin is expressed in macrophages and widely believed to play a critical role in iron recycling by these cells (1, 38, 40, 84, 85).

Because there is no known excretion system for iron in mammals, regulation of systemic iron levels occurs by controlling intestinal iron absorption and by expanding or mobilizing body stores (68). Iron is stored in Ft within hepatocytes and tissue macrophages, and the level of these stores is monitored and maintained by a "stores regulator" (68). When systemic iron demand in response to erythropoiesis exceeds what is available in storage pools, an "erythroid regulator" up-regulates intestinal iron uptake to compensate and replenish the stores. Of direct relevance to the host response to infection is the "inflammatory regulator." During infection or inflammation, marked retention of iron within tissue macrophages, along with reduced intestinal iron absorption, appears to restrict the amount of iron available to pathogens and thereby promote resistance to infection. Hepcidin, a small peptide produced by hepatocytes, has been shown to be a major regulator of these iron stores. In response to iron overload, hepatic expression of hepcidin is induced (124). Hepcidin-deficient (USF2^−/−) mice develop tissue iron overload with iron depletion in splenic and liver macrophages, whereas hepcidin-overexpressing mice are anemic with marked iron accumulation in macrophages (114, 115). Hepcidin also appears to be an acute phase-response peptide induced directly by various inflammatory stimuli including bacterial lipopolysaccharide (LPS) (5, 99, 116, 135, 157). The mechanisms governing these regulatory networks are highly complex and, in general, beyond the scope of this review. For a detailed account of mammalian iron metabolism please refer to an excellent recent review (68).

THE MACROPHAGE IRON TRANSPORTERS

Phagosomal Nramp1 (Slc11a1)

The Ity/Lsh/Bcg locus on mouse chromosome 1 has been known for many years as a key determinant of susceptibility to infection by a number of intracellular bacteria/parasites, including Mycobacterium, Leishmania, and Salmonella (136). Phenotypically susceptible mice (Bcg^s) permit rapid microbial replication within the spleen, liver, and lungs, whereas resistant (Bcg^r) animals are able to suppress this replication. Infections by avirulent bacteria, such as Mycobacterium bovis (BCG), are eventually cleared by both Bcg^r and Bcg^s mice, but Bcg^s mice rapidly succumb to infection by virulent pathogens such as Salmonella enterica serovar Typhimurium. Early experiments with mutant mouse strains and explanted cells infected in vitro established that the mature macrophage was the cell type responsible for the phenotypic expression of the genetic difference at Bcg/Ity/Lsh (61, 96, 97). Molecular characterization of the Bcg/Ity/Lsh locus by positional cloning identified the "natural resistance associated macrophage protein 1" (Nramp1, now known as solute carrier family 11 member 1 [Slc11a1]: OMIM 600266) as the gene affected by Ity/Lsh/Bcg. This was confirmed by systematic sequence analysis of the Nramp1 gene in Bcg^r and Bcg^s mouse strains, as well as through the creation of transgenic animals bearing loss-of-function (knockout) or gain-of-function alleles at the Nramp1 locus (60, 147). The pleiotropic effect of Nramp1 mutations on susceptibility to infection with unrelated pathogens suggested a critical role for this protein in innate immunity.

Nramp1mRNA is expressed in monocytes and macrophages as well as in polymorphonuclear leukocytes (58, 148). Nramp1 is transcribed from a TATA boxless promoter containing consensus binding sites for the macrophage and B-cell-specific transcription factor Pu.1 and the general transcription factors Ap1-3 and Sp1, as well as response elements for LPS and gamma interferon (IFN-γ) (59). Nramp1 mRNA expression can be stimulated in macrophages by mycobacterial infection, treatment with LPS and IFN-γ, or inflammatory stimuli in vivo (58, 59, 161). Macrophage-specific regulation occurs via binding of the IRF-8, Miz-1, and Pu.1 transcription factors to the Nramp1 promoter region (4, 16, 17). IRF-8 is also involved in regulating the expression of inflammatory cytokines interleukin-12 (IL-12), IL-18, and IL-1, and it plays a role in controlling the production of reactive oxygen (phox) and nitrogen (iNOS) species by macrophages, thereby linking Nramp1 expression to the inflammatory response of macrophages against microbes (4). Nramp1 transcription can also be induced in response to oxidative stress, possibly mediated by the Sp1 transcription factor, as occurs during glutathione depletion or when the cytoplasm is loaded with redox-active iron salts (9, 158). c-Myc represses Nramp1 transcription through competitive inhibition of Miz-1 binding (16, 17, 93).

Mature Nramp1 protein is found in the membrane fraction of macrophages as a 90- to 100-kDa phosphoglycoprotein (149). Sequence analysis indicates 12 putative transmembrane (TM) domains and a large glycosylated extracytoplasmic loop for which the location has recently been confirmed by topological studies (26, 48, 122). Susceptibility to bacterial/parasitic infection in Bcg^s mouse strains is caused by a single Gly¹⁶⁹ to Asp¹⁶⁹ amino acid substitution in the predicted TM4 of Nramp1, which leads to a complete loss of the mature protein in macrophages (148, 149). This mutation appears to cause misfolding and retention of Nramp1 within the endoplasmic reticulum leading to its eventual degradation (155). Wild-type Nramp1 protein is localized to the membrane of Lamp-1-positive lysosomes or late-endosomes in primary macrophages or macrophage-derived cell lines (64), and in the tertiary, gelatinase-positive granules of neutrophils. During phagocytosis, Nramp1-containing lysosomal vesicles are incorporated into, and remain associated with, the membranes of phagosomes following uptake of latex particles (64) or live Salmonella, Leishmania, Mycobacterium, or Yersinia organisms (35, 57, 134).

Human Nramp1 mRNA is expressed in spleen, lung, and peripheral blood leukocytes (27). Studies in primary cells and in model cell lines (HL-60) show that Nramp1 mRNA expression is particularly abundant in neutrophils and increases upon migration of these cells to tissues (27). Immunoblotting studies of the cellular and subcellular localization of Nramp1 protein in fractioned primary neutrophil membranes using anti-Nramp1 antibodies together with granule-specific markers show that Nramp1 is primarily within tertiary granules positive for the matrix enzyme gelatinase and membrane subunits of the vacuolar H⁺-ATPase (21). Immunogold cryoelectron microscopy (EM) has confirmed that most Nramp1-positive granules (75%) are positive for gelatinase. These granules can be triggered for exocytosis by treatment of neutrophils with phorbol myristate acetate. Immunofluorescence studies of Candida albicans-infected neutrophils show that Nramp1 can be recruited from tertiary granules to the phagosomal membrane during phagocytosis (21). These results suggest that human Nramp1 may play an important role in the antimicrobial and inflammatory actions of both macrophages and neutrophils.

In humans, the Nramp1 gene maps to chromosome 2q35 near the gene encoding the IL-8 receptor gene (25). Multiple genomic DNA sequence polymorphisms have been identified in or near Nramp1 (14, 20, 95, 98). Nramp1 polymorphic variants such as these have been used in genetic linkage and association studies to demonstrate an association with several human infectious and inflammatory diseases (15, 125). Associations of Nramp1 polymorphisms with infectious diseases such as tuberculosis and leprosy, and inflammatory conditions such as rheumatoid arthritis, diabetes, inflammatory bowel disease, and multiple sclerosis, have been established. Nramp1 polymorphisms were found to be risk factors for pediatric tuberculosis disease (105) and are associated with susceptibility to nontuberculous (e.g., Mycobacterium avium complex) mycobacterial lung disease (86). Note that while Nramp1 alleles have been associated with human tuberculosis (and other diseases), in the mouse model Nramp1 has a greater effect on susceptibility/resistance to less virulent mycobacteria such as M. avium and BCG than to the virulent human pathogen Mycobacterium tuberculosis (103, 108, 109, 118).

The Ubiquitous Nramp2 (Slc11a2)

Nramp2(OMIM 600523) also known as DMT1, DCT1, and now designated SLC11a2, shares 64% amino acid identity (78% overall similarity) with the predicted polypeptide sequence of Nramp1(63). Like Nramp1, Nramp2 is a heavily glycosylated membrane protein with a molecular mass of 90–100 kDa (62). In mice, Nramp2 is ubiquitously expressed but is particularly abundant in the intestine, kidney, and erythroid compartment (23, 63). There are four splice variants of Nramp2 that are thought to confer specific tissue regulation. Two variants result from splicing of an additional 5' exon (70). Exon 1B-containing mRNA is expressed ubiquitously, whereas Exon 1A mRNA appears to be restricted to the kidney and duodenum. These splice variants remain to be confirmed at the protein level. Alternative splicing of the 3' exon generates two additional transcripts, which encode distinct C termini that can be differentiated with isoform-specific antibodies (23, 94). Isoform I contains an iron response element (IRE) in its 3'-untranslated region (3'-UTR). This isoform is expressed at the duodenum brush border where it is responsible for Tf-independent uptake of dietary iron from the intestinal lumen and can be induced by dietary iron restriction. Isoform II does not contain an IRE and is not responsive to deprivation of nutritional iron. Like Nramp1, Nramp2 also appears to be up-regulated in mouse macrophages, bronchial epithelial cells, and human monocytes by inflammatory cytokines or infection (101, 153, 161).

Detailed analysis of the subcellular localization of Nramp2 in reticulocytes and transfected epithelial cells shows that the majority of Nramp2 (isoform II) is present in acidic TfR-positive recycling endosomes, where it acts to efflux Tf-delivered iron into the cytosol (24, 43, 62, 143). This endomembrane Nramp2 pool is in dynamic equilibrium with a smaller plasma membrane (PM) pool, and Nramp2 cycles between the two pools using phosphatidyl inositol-3-kinase-regulated exocytosis and Clathrin-dependent endocytic pathways (122, 143). In addition, Nramp2 has been shown to associate with Latex-bead- or red-blood-cell-containing phagosomes formed in J774 or RAW264.7 macrophages and with spermatozoids in testicle-derived phagocytes (Sertoli cell line TM4) (62, 71). Based on this colocalization it has been proposed that Nramp2 may participate in recycling of iron from senescent red blood cells or sperm cells after phagocytosis by specialized macrophages. This hypothesis is supported by studies showing that mk mice exhibit both microcytic anemia and reduced fertility due to impaired spermatogenesis (24, 44, 130). However, Nramp2 function in the phagosomal membrane remains to be fully characterized.

A Gly185Arg amino acid substitution in Nramp2 causes microcytic anemia in both mk mice and Belgrade rats (43, 44). Microcytic anemia in the mk and Belgrade mutants is associated with impaired iron acquisition at the intestinal brush border, as well as impaired iron metabolism in peripheral tissues that cannot be corrected by parenteral or intravenous iron administration (119). At the molecular level, the Gly185Arg mutation is associated with reduced transport activity (90, 139), defective intracellular trafficking (22), and impairment of both intestinal iron absorption and endosomal iron efflux (24, 43, 44, 83). When expressed in epithelial LLC-PK1 cells, Nramp2^G185R has been shown to be inefficiently processed, resulting in accumulation of the immature core glycosylated form (144). Most of this protein is retained in the endoplasmic reticulum, where it is degraded by a proteosome-dependent mechanism. A small proportion of Nramp2^G185R localizes normally but is rapidly degraded at the plasma membrane and exhibits greatly reduced specific transport activity. Therefore, the microcytic anemia phenotype caused by the mk mutation in mice is due to multiple defects in the function and biosynthesis of Nramp2. Recently, a mutation in human Nramp2/DMT1 was identified in a patient with severe microcytic anemia (110). The mutation at the 3' end of exon 12 appears to have the dual effect of altering the coding sequence of the protein (E399D) and favoring skipping of exon 12 during processing of the pre-mRNA. The E399D substitution alone was shown to have no measured effect on DMT1 (92, 127), but complete deletion of exon 12 resulted in defective DMT1 protein expression, transport function, and subcellular localization (127). Therefore, quantitative reduction of functional DMT1 protein by exon skipping appeared to be responsible for disease in this patient. In contrast to the mk and Belgrade animal models, the human mutation was associated with iron overload, suggesting either mutation-specific effects or significant physiological differences between humans and the rodent models of DMT1 deficiency.

Metal Transport by Nramp Proteins

The Nramp family of proteins is an ancient family of divalent metal transporters, with orthologues in almost every species examined, spanning mammals, birds, insects, plants, fish, yeast, and bacteria (26, 47, 91). Nramp mutants in model organisms, such as Drosophila melanogaster and Saccharomyces cerevisiae, can be complemented by expressing their mammalian counterparts, demonstrating that sequence conservation in the Nramp family also reflects functional conservation (47, 91). The yeast S. cerevisiae has three members of the Nramp family: SMF1, 2, and 3 (126). Studies in intact yeast cells (126, 140) or by electrophysiological measurements in Xenopus laevis oocytes (29, 131) have revealed an acidic pH-dependent, divalent metal transport mechanism with broad substrate specificity (Mn²⁺, Cd²⁺, Cu²⁺) for Smf1p and Smf2p, while Smf3p seems to be more selective for Fe²⁺. Similarly, Nramp proteins (designated MntH) from Escherichia coli, S. enterica serovar Typhimurium, and M. tuberculosis have been shown to be multispecific, acidic pH-dependent, divalent metal uptake transporters with a preference for Mn²⁺ ions (2, 79, 104).

Transport studies in X. laevis oocytes or transfected mammalian cells (48, 65, 90, 122, 141) have shown that Nramp2 expressed at the cell surface is capable of transporting many divalent metals, including Fe²⁺, Zn²⁺, Cd²⁺, Mn²⁺, Cu²⁺, and Co²⁺, but not Ca²⁺ or Mg²⁺, across the plasma membrane. Metal transport is saturable and pH dependent, with maximal activity under acidic conditions (48, 65, 122, 141). Nramp2 has a high affinity for iron (Fe²⁺), with half-maximal activity in the low micromolar range (65). Iron can only be transported in its reduced form and remains accessible to intracellular chelators following transport, suggesting that it is transported directly into the cytoplasmic labile iron pool (122). Metal transport by Nramp2 is accompanied by an inward proton current (65), suggesting that Nramp2 is an active, proton/divalent metal symporter utilizing a proton electrochemical gradient as its energy source. However, the exact stoichiometry of cation:proton for cotransport appears to be variable, a phenomenon that remains to be explained (65, 102).

Although biochemical studies and sequence comparisons suggested that evolutionarily distant but related Nramp proteins share a common transport mechanism, Nramp1 at the phagosomal membrane, in particular, has been difficult to study, and therefore its mechanism of transport and substrate specificity have been controversial. Early studies reported increased Nramp1-dependent accumulation/binding of radioisotopic Fe²⁺ into isolated phagosomes containing either Latex beads or M. avium (87, 88, 165). This Fe²⁺ accumulation was abrogated by the addition of anti-Nramp1 antibodies suggesting that Nramp1 might transport cytoplasmic Fe²⁺ into phagosomes. A potential limitation of these studies was that phagosomal metal binding (whether Nramp1 dependent or independent) was not distinguished from transmembrane metal transport. A later study using Nramp1 mRNA injected into X. laevis oocytes found Zn²⁺-dependent inward currents to be induced over controls at an alkaline extracellular pH, which was suggestive of metal uptake against the experimental proton gradient (56). Based on additional pH-dependent transport studies using radioisotopic Zn²⁺, these authors concluded that Nramp1 might transport cytoplasmic metals into the phagosome by a proton/divalent-metal antiport mechanism. They proposed that increased phagosomal Fe²⁺ could stimulate oxygen radical production in situ via the Fenton reaction, resulting in increased bactericidal activity (56, 87, 88, 165). However, Nramp1-dependent phagosomal metal influx by an antiport mechanism would require a transport mechanism distinct from Nramp2, both with respect to the direction of transport and utilization of the transmembrane pH gradient.

Real-time microfluorescence imaging of zymosan particles chemically coupled with a metal-sensitive fluorophor (Fura-FF6) was used in another study to monitor divalent-metal flux across the membrane of individual phagosomes formed in live Nramp1^+/+ or Nramp1^−/− peritoneal macrophages (73). Following exposure of thapsigargin-permeabilized macrophages to high Mn²⁺ concentrations, Nramp1^+/+ phagosomes accumulated less Mn²⁺ than did Nramp1^−/− controls. Similarly, Nramp1^+/+ phagosomes released more Mn²⁺ from preloaded Fura-FF6-zymosan particles than their Nramp1^−/− counterparts. Dissipation of the phagosomal proton gradient by using concanamycin (a vacuolar H⁺-ATPase inhibitor) eliminated Nramp1-dependent Mn²⁺ transport, suggesting that Nramp1 functions in a pH-dependent fashion to efflux Mn²⁺ ions from acidified phagosomes down the proton gradient maintained by the vacuolar H⁺-ATPase. Independent biochemical studies support this model (7, 8).

To gain direct access to Nramp1for biochemical studies, we inserted an influenza hemagglutinin (HA) epitope into the TM7/8-loop of Nramp1 (Nramp1HA), which resulted in ectopic targeting of the recombinant protein to the PM of transfected CHO cells (48). Differential accessibility of the HA or N-terminal epitopes to extracellular antibodies in live cells revealed that the TM7/8 loop, which contains several predicted N-linked glycosylation sites (26), is extracellular and that the N terminus is cytoplasmic when Nramp1HA is expressed at the PM. Thus, the membrane organization of Nramp1HA at the PM is identical with the orientation established previously for Nramp2HA (122). The topology of the two proteins is also very similar to that of the E. coli MntH protein as determined by using PhoA reporter fusions (34). According to this topology, the TM7/8 loop of Nramp1 and Nramp2 is situated in the lumenal space of phagolysosomes or recycling endosomes, respectively.

Fluorescence-quenching assays and studies with radioisotopic ⁵⁵Fe²⁺ and ⁵⁴Mn²⁺ demonstrated that Nramp1HA expressed at the PM could function as a multispecific divalent-metal transporter (48). Divalent-metal transport by Nramp1HA across the PM was time, temperature, and acidic pH dependent, occurring down a proton gradient similar to that previously shown for Nramp2 in X. laevis oocytes (65) or transfected mammalian cells (90, 122, 141), as well as Nramp family transporters from other species (29, 65, 79, 131). The apparent affinity of Nramp1HA (and Nramp2HA) for Fe²⁺ and Mn²⁺ (48) is in the low micromolar range, consistent with the concentration of chelatable Fe²⁺ (0.2 to 1 μM) or Mn²⁺ (<4 μM) in cells or tissues (19, 76). Together, these results strongly support a model in which Nramp1 removes divalent metals from the lumen of acidified phagosomes transporting them into the cytoplasm to limit the concentration of phagosomal metal ions.

These studies have convincingly established that Nramp1 and Nramp2 are functionally equivalent multispecific divalent-metal transporters both using a proton-coupled cotransport mechanism (48). The major physiological difference between Nramp1 and 2 appears to be their distinct subcellular sites of transport and tissue specificity. Nramp2 functions at the intestinal brush border to take up dietary iron and in other tissues transports Tf-delivered iron from the acidified lumen of recycling endosomes into the cytoplasm. The phagocyte-specific Nramp1 functions in an analogous fashion to remove divalent metals from the acidified phagosome. In both cases, luminal endomembrane acidification is generated by the vacuolar H⁺-ATPase that provides the proton gradient required for metal efflux (6, 67, 73). For a more complete discussion of Nramp protein biochemistry please refer to a recent review (102).

Ferroportin

Ferroportin (also called Fpn1, Ireg1, Mtp1, annotated as Slc40a1) is the prototypical member of a newly identified protein family. The ferroportin gene encodes an integral membrane protein containing 10 putative TM domains and is expressed in the liver, spleen, and within bone marrow macrophages (1, 40, 156). In duodenal mucosal cells, ferroportin is localized to the basolateral surface where it effluxes dietary iron across the basal membrane and into the circulation after uptake of iron from the intestinal lumen by Nramp2 at the apical surface (1, 40, 107). Both ferroportin and Nramp2 are coregulated in the intestine in response to iron availability and by hepcidin (1, 23, 146, 163, 164)

Ferroportin expression in the liver, spleen, and bone marrow cells appears to be restricted to macrophages (1, 40, 156). Mutant mice bearing a conditional deletion of the ferroportin gene in macrophages show retention of iron by hepatic Kupffer cells and splenic macrophages (41). Likewise, disruption of ferroportin expression by RNA interference results in macrophage iron accumulation and increased synthesis of ferritin, strongly suggesting that ferroportin participates in iron metabolism in macrophages (51). Therefore, ferroportin appears to act as the primary iron efflux system in macrophages. In LPS-stimulated or Leishmania donovani-infected mice, ferroportin expression in macrophages is down-regulated (156). In vitro studies have shown that IFN-γ and LPS treatment down-regulate ferroportin mRNA in monocytes (101) and macrophages (99) (F. Canonne-Hergaux and P. Gros, unpublished data). Hepcidin is induced by LPS in mouse spleens and splenic macrophage in vitro (99) and appears to mediate the LPS-induced down-regulation of ferroportin in the intestine (157) and in splenic macrophages (99), suggesting that inflammatory agents may regulate iron metabolism through modulation of ferroportin expression. Mutations in the human ferroportin gene cause an autosomal-dominant form of iron overload distinct from hemochromotosis, which is characterized by iron accumulation in the reticuloendothelial system (39, 111, 117, 123, 152).

Immunofluorescence experiments have detected ferroportin at the PM and within intracellular vesicles of macrophages (1, 37, 156). Ferroportin was originally localized to intracellular compartments of macrophages but not the PM, leading to the suggestion that cellular iron efflux by this protein may require redistribution of the protein to the PM in response to intracellular iron (85). This concept is supported by a recent study (37) showing that iron treatment or phagocytosis of red blood cells (erythrophagocytosis) resulted in strong localization of ferroportin to the PM of macrophage. Iron treatment caused an overall up-regulation of ferroportin protein levels, thus it remains unclear if PM localization of ferroportin is due to redistribution or increased expression. Treatment of macrophage with hepcidin was shown to cause a dramatic reduction in ferroportin protein levels and induce its rapid internalization and degradation (37, 84). Similarly, a ferroportin green fluorescent protein fusion (Fp-GFP) was shown to localize to the plasma membrane of HeLa or HEK293 cell lines in the absence of hepcidin (113). Direct binding of hepcidin to Fp-GFP at the cell surface caused the internalization and degradation of the protein with a concomitant decrease in iron efflux measured by changes in intracellular Ft levels. The authors of this study propose a homeostatic loop in which the levels of secreted hepcidin are regulated by iron, with subsequent control of ferroportin levels by modulation of both transcription and cell surface expression (113). In Kupffer cells, intracellular ferroportin vesicles are positive for hemosiderin (1), suggesting an additional role for ferroportin in intracellular iron trafficking.

Expression of ferroportin in X. laevis oocytes and mammalian cells causes cellular efflux of iron and depletion of intracellular Ft-bound iron (1, 40, 113, 156). In macrophages, ferroportin is up-regulated by iron, copper, and erythrophagocytosis, and stimulates the efflux of iron delivered by erythrophagocytosis (30, 37, 38, 84, 85). Treatment of ferroportin-overexpressing J774 macrophages with hepcidin caused a reduction in ferroportin protein levels with a corresponding reduction of iron efflux after erythrophagocytosis (84). These experiments support a physiological role of ferroportin in recycling iron from old or damaged red cells within macrophages. Erythrophagocytosis in macrophages was reported to result in induction of ferroportin and Nramp1 (twofold), but not Nramp2, mRNA levels (85). Nramp1 has previously been reported to stimulate the release of iron from phagocytosed Tf/immune complexes and to reduce iron deposition in the red pulp in response to iron overload (13, 112). These data are consistent with a model in which Nramp1 may transport iron into the cytoplasm after its release from iron-containing complexes or particles phagocytosed by macrophages for subsequent efflux by ferroportin. Taken together, these results suggest that Nramp1 and ferroportin may be functionally linked during erythrophagocytosis in macrophages, although note that there are no reports in the literature to date of hematological defects or anemia associated with Nramp1 in mice. Likewise, there have been no reports of immunological defects directly associated with mutations in ferroportin, suggesting the possible existence of redundant iron transport mechanisms.

IRON RESTRICTION IN MACROPHAGES AND BACTERIAL PATHOGENESIS

Effect of Nramp1 on Intracellular Mycobacteria

It has long been established that mycobacteria survive within macrophages by preventing the maturation of phagosomes into fully bactericidal phagolysosomes (31, 32, 33, 129, 132, 137, 138) (Fig. 2). Permissive Nramp1^−/− mycobacterial phagosomes exhibit impaired acidification due to a decreased number of vacuolar H⁺-ATPase pumps at the phagosomal membrane, as well as reduced acquisition of lysosomal markers, such as Lamp1 and the acid hydrolase Cathepsin D (129, 132, 137, 138). These phagosomes remain positive for sorting/early endosome markers, such as Rab5 and the TfR, and are accessible to exogenous Tf. Inhibition of maturation of Nramp1^−/− mycobacterial phagosomes is an active process that only takes place with live, metabolically active mycobacteria. Nramp1 recruitment to the membrane of M. avium-containing phagosomes (in Nramp1^+/+ cells) can prevent this mycobacterial-induced inhibition of maturation, resulting in increased bacterial damage, increased acidification, and enhanced fusion with lysosomes, with greater bacteriostasis than Nramp1^−/− phagosomes (49, 67) (Fig. 2). In addition, Nramp1^+/+ macrophages produce more nitric oxide (NO) in response to IFN-γ stimulation or mycobacterial infection than congenic Nramp1^−/− controls, possibly contributing to the inhibition of bacterial growth (10, 11, 50).

Fig. 2Model of macrophage metal transport processes during mycobacterial infection. Mycobacteria phagocytosed by macrophages are initially enclosed within a plasma membrane-derived phagosome. In Nramp1^−/− macrophages, these phagosomes remain immature, displaying characteristics of early endosomes (EEs): positivity for TfR and retained access to Tf-bound iron. These immature phagosomes are permissive for bacterial replication. In Nramp1^+/+ macrophages, mycobacterial phagosomes fuse with Lamp1, H⁺-ATPase, Cathepsin D, and Nramp1-positive late endosomes (LE)/lysosomes and mature into acidified phagolysosomes. Such phagosomes are nonpermissive for bacterial replication, and mycobacteria internalized under these conditions are destroyed. The mycobacterial inhibition of phagosome maturation is an active and metal-dependent process. Nramp1-mediated efflux of divalent metals, such as iron and manganese, at the phagosomal membrane antagonizes this process. Nramp1 has an analogous effect in modulating the fusogenic properties and antibacterial properties of SCV. Ferroportin is inducible by iron or copper and mediates the efflux of iron recycled by macrophages from phagocytosed red blood cells (erythrophagocytosis). During infection and inflammation ferroportin is repressed, contributing to the retention of iron within the cytosol during the establishment of "anemia of chronic disease" (ACD).

One possible explanation for these findings is that the active inhibition of phagosome maturation by mycobacteria is a metal-dependent process that can be counteracted by Nramp1-mediated metal efflux from the phagosomal lumen. The ability of Nramp1 to antagonize mycobacterial-directed inhibition of phagosome maturation has been linked to iron levels (80). Rab5 is a cytosolic GTPase involved in regulation of endosome-phagosome fusion that may be utilized by mycobacteria to arrest phagosome maturation (32, 80). A dominant-negative Rab5 mutant was shown to force the maturation of M. avium-containing phagosomes into mature phagolysosomes in Nramp1^−/− bone marrow macrophages, with a concomitant increase in bacteriostatic activity (80). Phagosomal maturation could be inhibited by preloading the macrophage with iron-citrate, suggesting that iron promotes the maturation block. Thus, Nramp1-induced depletion of phagosomal iron could impair the ability of mycobacteria to modulate phagolysosomal fusion.

Inhibition of phagosomal maturation by mycobacteria in Nramp1^−/− macrophages can be overcome by activation with IFN-γ (132), a process that may involve the induction of autophagy (66). Activation-dependent maturation of Nramp1^−/− mycobacterial phagosomes is associated with decreased acquisition of exogenous Tf through the early endosomal pathway, which may enhance iron limitation in the phagosome (120, 132). Elemental analysis of Nramp1^−/− macrophage phagosomes containing M. tuberculosis or M. avium detected sequestered iron delivered by exogenous Tf, which was abrogated by addition of either IFN-γ or tumor necrosis factor alpha (TNF-α) prior to infection (151). Avirulent Mycobacterium smegmatis or a siderophore-deficient M. tuberculosis mutant did not accumulate iron from Tf under the same conditions.

Infection of explanted human macrophages or mice with M. tuberculosis induces expression of mycobacterial genes required for siderophore-mediated iron uptake, and mutations in some of these genes attenuate virulence in vivo (36, 53, 142). In fact, gene expression analysis by microarray of M. tuberculosis isolated from Nramp1^−/− phagosomes reveals that one of the key bacterial responses following bone marrow macrophage activation by IFN-γ is the up-regulation of mycobacterial genes involved in metal acquisition (133), accounting for over half of the activated genes and indicating metal restriction in the intraphagosomal environment even in the absence of Nramp1. Indeed, iron enhances mycobacterial growth in Nramp1^+/+ mice and promotes the development of active tuberculosis (and growth of other pathogens) in humans (54, 128). Likewise, extracellular iron can stimulate intracellular growth of M. avium in both Nramp1^+/+ and Nramp1^−/− macrophages (55), suggesting that excess cellular iron may suppress any advantage conferred by Nramp1 (54, 55).Collectively, these results suggest that iron is required by mycobacteria to arrest phagosomal maturation and survive intracellularly and this process is antagonized by phagosomal Nramp1 (Fig. 2).

Effect of Nramp1 on Intracellular Salmonella

In macrophages, Salmonella can subvert phagosomal maturation to create specialized Salmonella-containing vacuoles (SCVs) in which bacteria can survive and replicate (82). Unlike mycobacterial phagosomes, SCVs formed in either Nramp1^+/+ or Nramp1^−/− macrophages acidify fully and fuse with Lamp1-positive lysosomes. However, in Nramp1^−/− RAW 264.7 permissive macrophages, SCVs do not recruit mannose-6-phosophate receptor (M6PR) positive vesicles and become inaccessible to the endosomal pathway (35). Expression of Nramp1 in RAW264.7 macrophages restores recruitment of M6PR and endosomal markers (EEA1 or fluid phase FITC-dextran) to the SCVs. Therefore, Nramp1 can counteract the ability of Salmonella to shelter itself from the degradative phagosomal pathway. Salmonella carrying gene-specific reporter constructs respond to the presence of Nramp1 in macrophages in vitro and in mice in vivo during infection by increasing expression of Salmonella pathogenicity island 2 (SPI2) virulence genes including ssrA, sseA, and sseJ compared with Nramp1^−/− controls, suggesting that Nramp1 creates a more stressful SCV environment ( 160). Thus, the presence of Nramp1 causes changes in the biochemical and fusogenic properties of SCVs that are sensed and reacted to by the internalized bacilli.

Both Fe²⁺ and Mn²⁺ are essential for Salmonella virulence in vivo and for intracellular replication within cultured macrophages (78, 128). Salmonella possess several high- or low-affinity Fe²⁺/Fe³⁺ and Mn²⁺ transporters, such as fepBCDG, sitA-D, feoABC, and the Nramp homolog mntH (75, 77, 145, 162). Several of these transporters have been shown to contribute to Salmonella virulence in vivo (12, 18, 74, 145). For example, single mutations of feoB or sitA-D reduce virulence of Salmonella in vivo in otherwise susceptible 129sv-Nramp1^−/− mutant mice, while double mutations of sitA-D and feoB or mntH abrogate virulence entirely (18). Independently, it was shown that mntH or sitA single mutants are attenuated for virulence following intraperitoneal infection of C57BL/6-Nramp1^+/+ transgenic mice, while the double mntH sitA mutant is completely avirulent (159). Both mutants were as virulent as wild-type Salmonella following intraperitoneal infection of susceptible Nramp1^−/− C57BL/6 mice, suggesting that bacterial metal transporters are required for virulence under conditions of metal limitation caused by Nramp1. In addition, divalent metal chelation was found to significantly reduce the intracellular replication of wild-type Salmonella and, in particular, of Salmonella mutants impaired for metal acquisition within susceptible Nramp1^−/− macrophages (18). Thus, Nramp1-mediated depletion of metals from the intraphagosomal space appears to have a major impact on intracellular survival and replication of Salmonella. In support of this proposal, mntH and sitA were found to be induced at a greater level following infection of Nramp1^+/+ than of Nramp1^−/− macrophages, suggesting reduced metal availability in Nramp1-positive SCVs (159). Conversely, microarray studies of gene expression in Salmonella recovered from Nramp1^−/−-negative SCVs formed in J774 macrophages revealed an expression profile consistent with a metal-replete intravesicular environment in the absence of Nramp1 (42).

As mentioned above, Nramp1 stimulates both acquisition of M6PR and accessibility of endosomes to SCVs. As with mycobacteria, this process has been linked to intracellular metals (72). Pretreatment of Nramp1^−/− primary macrophages or the RAW264.7 macrophage cell line with the membrane-permeating iron chelators desferroxamine or salicylaldehyde isocotinoyl hydrazone restores the recruitment of M6PR and delivery of the fluid-phase marker rhodamine-dextran to SCVs at levels similar to those seen in macrophages expressing wild type Nramp1. These data suggest that Nramp1-mediated deprivation of iron and possibly of other divalent metals such as Mn²⁺ from the phagosomal space inhibits the ability of Salmonella to block phagosomal maturation. Thus, Nramp1-mediated metal efflux at the phagosomal membrane antagonizes the survival strategy of Salmonella.

Tf, Nramp2, and Ferroportin

In humans, global changes in systemic iron metabolism have been observed in response to acute or chronic inflammation (128). These changes include decreased serum Tf-Fe³⁺, decreased intestinal iron uptake, and marked retention of iron in macrophages. This syndrome has sometimes been referred to as the "anemia of chronic disease" (ACD) (154). ACD is thought to represent an adaptive response to limit the bioavailability of iron for pathogen growth during infection. Indeed, treatment of ACD by iron supplementation has been documented to have deleterious effects on the host, especially in immunocompromised patients (128, 154).

Of the macrophage iron transporters, only Nramp1 has been shown to act directly on pathogens as described above. However, like Nramp1, ferroportin and Nramp2 are modulated by immune stimuli in macrophages, suggesting a possible role in host defense through the establishment of ACD. In macrophages, Nramp2 is up-regulated in response to infection while expression of the TfR is simultaneously down-regulated (161). This would be expected to reduce the amount of Tf-Fe³⁺ available to intracellular pathogens within the endosomal system. Ferroportin down-regulation in response to infection/inflammation is thought to contribute to the establishment of ACD (154). This causes iron accumulation within macrophages and enterocytes, leading to systemic anemia and cellular iron restriction, including reduced circulating Tf-Fe³⁺. Macrophage iron accumulation is concomitant with increased Ft expression (154), which may act to sequester excess iron in the cytosolic compartment of macrophages during ACD, limiting its accessibility to pathogens contained within endomembrane compartments (i.e., mycobacterial phagolysosomes and SCV). Together, these metabolic changes in Tf, Ft, Nramp2, and ferroportin during infection may synergize with Nramp1 to limit bioavailability of iron to intracellular microbes and thus play a critical role at the interface of host and pathogen. Similar to ACD, inflammation and infection triggers hypozincemia, to which the zinc import protein Zip14 apparently contributes (100).

Lipocalins

Lipocalins are a family of small proteins that bind hydrophobic ligands and are thought to serve as carriers for the delivery of small molecules to target cells (3). Of this protein family, Lipocalin-2 (Lcn2, siderocalin) has been identified as a component of the innate immune system involved in the acute phase response to bacterial infections (45, 81). In mice, Lcn2 mRNA and protein are induced in the liver and, to a lesser extent, in the spleen in response to E. coli infection or LPS treatment. Induction depends on toll-like receptor 4 (Tlr4) and peaks 24 h after infection along with a marked rise of Lcn2 protein in the serum (45). After intraperitoneal infection with E. coli, Lcn2^−/− knockout mice suffer severe bacteremia, display increased bacterial load in spleen and liver, and have reduced survival compared with Lcn^+/+ controls, demonstrating the antibacterial properties of Lcn2. Lcn2 has been shown to bind with high affinity to enterochelin-like siderophores produced by E. coli and Salmonella, as well as carboxymycobactins produced by Mycobacterium (45, 52, 69) (see elsewhere in EcoSal regarding iron-uptake systems). Lcn2 can directly inhibit enterochelin-dependent growth of cultured E. coli in a dose-dependent manner, an effect reversed by the addition of exogenous enterochelin (45). The antibacterial effect of Lcn2 can be reversed in mice by coinjection of ferrichrome, a siderophore to which Lcn2 does not bind, that can be utilized by E. coli as an alternative iron source. These observations suggest that Lcn2 contributes to bacterial resistance by interfering with enterochelin-like siderophore-dependent iron uptake by bacteria. Although Lcn2 itself acts only on certain types of siderophores, other Lcn2-like anti-siderophore molecules exist (46) or have been predicted to exist (69) and may collectively form a critical line of defense to prevent iron acquisition by invading pathogens.

CONCLUSIONS

Control of iron levels in the intraphagosomal environment constitutes a key battle fought between competing iron acquisition systems of both the host and microbial pathogen. The host transporter Nramp1 may compete directly with bacterial divalent-metal transport systems for the acquisition of divalent metals within the phagosomal space (47), a process further supported by the actions of Nramp2 and ferroportin at the systemic level. The ultimate outcome of these competing interactions influences the ability of pathogens to survive and replicate intracellularly. This seems particularly relevant to the Salmonella, Leishmania, and Mycobacterium spp., in which inactivating mutations in Nramp1 abrogate the natural resistance of macrophages to these pathogens.

Divalent-metals such as Zn²⁺, Cu²⁺, Fe²⁺, and Mn²⁺ are all essential bacterial micronutrients acting as cofactors for many enzymatic reactions (47). Illustrating this point is that under starvation conditions E. coli are able to concentrate intracellular Fe²⁺ and Zn²⁺ approximately 2,000-fold (and other metals including Cu²⁺ and Mn²⁺ to a lesser extent) against the concentration gradient (121). Therefore, depletion of one or more of these divalent-metals from the phagosome by Nramp1 may have a simple and direct bacteriostatic effect by removing a rate-limiting nutrient from the ecological niche of intracellular pathogens (47). Metal depletion in bacterial phagosomes may also enhance the bactericidal arsenal of the macrophage by rendering the pathogen more sensitive to killing by reactive nitrogen and/or oxygen species. Indeed, several antioxidant defenses of microbial pathogens require the presence of iron or manganese, cofactors for catalase and superoxide dismutase. Additionally, Nramp1-mediated depletion of divalent metals from the phagosomal space, either during or before phagocytosis, could affect the expression and/or function of secreted bacterial effector proteins or lipids involved in the modulation of phagosome maturation, thereby having an indirect effect on pathogen survival by promoting the formation of mature bactericidal phagolysosomes. Thus, prokaryotic and mammalian divalent-metal transporters appear to be key determinants of bacterial virulence and host defense during intracellular parasitism. Pharmacological modulation of this interaction may represent a new opportunity for drug discovery and pharmacological intervention in the treatment of infections.