Human iron transporters manage iron carefully because tissues need iron for critical functions, but too much iron increases the risk of reactive oxygen species. Iron acquisition occurs in the duodenum via divalent metal transporter (DMT1) and ferroportin. Iron trafficking depends largely on the transferrin cycle. Nevertheless, non-digestive tissues have a variety of other iron transporters that may render DMT1 modestly redundant, and DMT1 levels exceed those needed for the just-mentioned tasks. This review begins to consider why and also describes advances after 2008 that begin to address this challenge.
KeywordsDivalent metal transporter (DMT1) Ferroportin Isoforms Transferrin Ferric reductase Iron-responsive element (IRE) Iron regulatory protein (IRP)
Because the author recently had the opportunity to address a closely related topic , this review will largely be an update on that one with the emphasis on exploring issues that have arisen since then or were overlooked before or are specific for human iron transporters, again trying to encourage researchers to move to improve our understanding rather than to follow fashion. It will repeat information only where essential for understanding of a topic. New recent relevant reviews include coverage of iron acquisition and storage , impaired iron homeostasis in Parkinsonism , normal iron homeostasis from a molecular viewpoint , normal and pathological iron homeostasis in the eye , hereditary hemochromatosis , hepcidin’s role as a regulator  and recycling of iron primarily by macrophage .
Human cells and tissues manage iron carefully. Nearly all of them need iron to supply redox functions in many pathways in both heme and non-heme proteins. Iron, however, helps to generate reactive oxygen species (ROS) via the same capability, thus the need for careful management.
Duodenal iron acquisition
Earlier, we  called attention to the lack of recent research on a speculative role of mucins interacting with DMT1 in the cellular acquisition of iron from the lumen of the duodenum. A few months later, coculture of cells that model enterocytes with the ones that model goblet (mucin-secreting) cells was described . Such cocultures appear to predict iron bioavailability for foods  and look like a system where new insights can develop. Perhaps they can also encourage a new level of insight into how phytates, polyphenols, and other inhibitors render iron-rich plant foods like spinach into poor sources of iron.
The Tf cycle: iron trafficking into erythroid cells as well as most other tissues
In the previous review , we considered ferritin (ft) in a separate section, but the recent discovery that a form of ft can enter cells via the human TfR1  places ft as an alternative substrate for the Tf cycle. Ft normally consists of 24 subunits of two types called H and L to distinguish the heavier chain from the lighter. The proportion of H and L ft varies. Li et al. show that TfR1 is a receptor also for H ft but not for L ft and that binding leads to H ft entering endosomes and lysosomes. Potentially, H ft can be a source of iron for cells and tissues via this route. The authors clearly distinguish this pathway from that for H ft reliant on T-cell immunoglobulin-domain and mucin-domain protein 2 (TIM-2) in mice [3, 7, 17]. They state that murine TfR1 does not bind human H ft nor murine H ft, suggesting that the mechanism for receptor recognition of H ft diverged between humans and mice. The properties imply that one should look for how well the pathway actually contributes to iron trafficking in human mutants with atransferrinemia or hypotransferrinemia, but that the similar mouse mutants will not be of value in such an appraisal.
Macrophage and Nramp1
Nramp1 (Natural resistance–associated macrophage protein 1) has its highest expression in macrophage as its name implies  and citations therein. There has now been an important advance that helps to define its functions and direction of transport . The investigators showed that loss of either DMT1 or Nramp1 leads to a minor deficit in recycling hemoglobin after erythrophagocytosis. Simultaneous loss of both transporters, however, severely impairs iron recycling. Their work shows that Nramp1 plays a role in the phagosome similar to the one that DMT1 plays in the endosome: Exit of iron toward its destination of recovery for another round of hemoglobin formation. Their data also bear on the controversy on whether Nramp1 is a proton symporter like DMT1 or an antiporter because the function of Nramp1 is now demonstrably partially redundant to DMT1 , implying that Nramp1 is a proton symporter that provides a measure of resistance to many infections by depleting the phagosome of iron and perhaps manganese.
Several aspects of iron transporter properties and regulation in the liver have been covered before  and in citations therein. There is value now in updating two aspects: efforts to modulate transporters and the role of the second Tf receptor (TfR2).
Weiss’s group  have presented data showing that nifedipine, a calcium channel blocker, stimulates DMT1 transport activity. They also reported that nifedipine mobilizes iron from the liver of mice with primary and secondary iron overload and enhances urinary iron excretion. Another result was that nifedipine led to loss of serum iron in a fashion that related to DMT1 genotype in +/+ and +/mk mice where the mk mutation is G185R in DMT1. These results offer nifedipine as a means of modulating iron overload in a fashion that would be off-label use of a USA FDA-approved drug. We  have not been able to reproduce the stimulation of DMT1 by nifedipine and suggest that the ability of photodegraded nifedipine to serve as an iron channel [19, 44, 45] may account for some of their observations. Unfortunately, their statistical analysis  supporting the argument that the DMT1 mutation diminishes the loss of serum iron is flawed and the postulated role of DMT1 in supporting exit of iron would make DMT1 join Fpn as an iron exporter. We do have unpublished data (M. Garrick, L. Zhao, B. Hagerty, S. Gadersohi, A. Ghio, L. Garrick and B. Mackenzie 2009) that suggest that nifedipine treatment does decrease serum iron in rodents in a fashion that does not relate to the DMT1 genotype, so nifedipine could still be of interest for removing iron. Nifedipine’s mechanism for doing so, however, and whether it does diminish liver iron both require more studies.
Although stimulation of DMT1 activity remains elusive, there are multiple reports of potential inhibitors. Ebselen, a seleno compound that may act via redox activity, clearly inhibits DMT1 . Several polysulfonated dyes are also inhibitors .
Most published approaches to Fpn are based on its regulation by hepcidin and other regulators. In this context, the critical involvement of the liver in TfR2 (and HFE) modulation of hepcidin has just been confirmed  in mice. Another study  examines the role of two isoforms of TfR2 mRNA. The α isoform is longer and encodes a full-length TfR2, while the β isoform is shorter. By generating a mouse that lacks the α isoform but retains as a knock-in the β isoform, this group suggests that its product is also involved in regulating Fpn transcription. Their results confirm the already identified function of the α isoform in hepcidin activation but also suggest that β TfR2 is more specifically involved in splenic Fpn function (iron efflux).
Contributions from transporter isoforms
Two human iron transporters have two or more isoforms with properties that insure that the isoforms act in different circumstances. These are DMT1 and Fpn.
The 4 isoforms of DMT1 mRNA (1A/+IRE, 1A/−IRE, 1B/+IRE, and 1B/−IRE) also encode related but distinct proteins (Fig. 4b). Potentially, differences between the N-terminus that starts in exon 1A and the one that starts in exon 2 and differences between the 18 C-terminal amino acid residues of the +IRE form and the 25 of the −IRE could affect localization of the transporter within the cell, thus where it functions and even where it turns over. Existing data do suggest the presence of such distinctions [5, 15, 24, 26, 38, 39, 48, 49].
Recently, Rouault’s laboratory  described alternate transcripts for Fpn (Fig. 4c). It is the only known human iron export transporter, as such it regulates the exit of iron from enterocytes into circulation. Their findings resolve a paradox about the regulation of Fpn. Given that the only previously recognized mRNA isoform has an IRE in the 5′ UTR, IRPs binding to that IRE during iron deficiency would shut off synthesis of Fpn, preventing iron absorption when absorption is needed most. Zhang et al. have established that this paradox applies only to one isoform of Fpn mRNA where transcription initiates from its promoter upstream of exon 1a. For some other mRNA isoforms, transcription initiates farther upstream so that the first exon is 1b where splicing to the middle or beyond of exon 1a leads to omission of the IRE. The transcripts that start with exon 1b will then not be down-regulated by iron deficiency. There is a choice of 3 possible splice acceptor sites for the 1a–b splicing, but it is unclear whether this choice has any effect on Fpn gene expression as all 3 lead to Fpn mRNA lacking an IRE. These same transcripts are also well expressed in erythroid differentiation, supporting the postulate  that iron homeostasis in these cells has privileged regulation to handle the heavy demand for iron. The Fpn amino acid sequence predicted by all of the mRNA isoforms is invariant, indicating that the targeting of their products is the same.
Neural tissue and related cell types
Because we previously reviewed this area  and there have been both an even more recently focused review  and another broader one that preceded , this section focuses on recent relevant data and a relationship that has been overlooked by many. Salazar et al.  have shown that the +IRE form of DMT1 is increased in the substantia nigra of Parkinson’s disease patients. Given that both microcytic mice (gene symbol mk) and Belgrade rats (gene symbol b) have a G185R mutation that impairs DMT1 transport activity [10, 11], they could determine whether similar increases in DMT1 were part of an inflammatory etiology of neurodegeneration in two rodent models of Parkinson’s disease or were part of a protective response. They used 1-methyl-4-phenyl-1, 2, 3, 6-tetrahydropyridine (MPTP) to induce Parkinsonism in mice. Homozygous mk/mk mice retained more tyrosine hydroxylase–positive neurons from the substantia nigra than control +/mk or +/+ mice after MPTP intoxication. Stereotactical injection of 6-hydroxydopamine (6-OHDA) into the hippocampus leads to retrograde transport of the 6-OHDA and loss of neurons from the substantia nigra. Similarly, b/b rats lost fewer tyrosine hydroxylase–positive neurons than control +/b or +/+ rats. Thus, DMT1 up-regulation in response to inflammation is likely to be part of a vicious cycle that contributes to neurodegeneration.
The increase in +IRE DMT1 in Parkinson’s disease and these models for Parkinsonism can be accounted for in multiple ways: NF-κB might be involved as it is frequently activated in inflammation, and the DMT1 1B promoter is responsive to this transcription factor [38, 39]. Another possibility is based on the observation that +IRE DMT1 increases. Possibly, IRP activation has occurred to stabilize +IRE DMT1 mRNA, although this response would be despite locally elevated iron levels. An interesting new connection relates to turnover of DMT1 protein. One of the studies of transcriptional regulation of DMT1 [38, 39] also showed that DMT1 isoforms are degraded primarily in the proteasome but also in the lysosome. Recently, we found that 1B DMT1 degradation depends on parkin . Proteasomal degradation depends primarily on the ubiquitination pathway where 3 types of enzymes, E1-E3, are involved. E1 is responsible for ubiquitin activation; the E2 family is responsible for conjugation; while the E3 family ligates ubiquitin to the protein target (here DMT1). Parkin is an E3 ligase; mutants in the PARK-2 gene account for the majority of autosomal recessive juvenile Parkinson’s disease. Our data show that 1A DMT1 is not affected by transient expression of parkin, while +IRE and −IRE 1B DMT1 are degraded. By immuno-depleting 1A DMT1, we verified that the 1B (or 2) forms are degraded by wild type but not by mutant parkin . Although most of our experiments were in neuronal models, lymphocytes from humans who have a PARK-2 mutation revealed that 1B DMT1 levels increased, so parkin affects DMT1 levels in non-neuronal tissue too .
Parkin is not the only E3 ubiquitin ligase that regulates DMT1 turnover; the Nedd4 family member WWP2 also diminishes DMT1 levels . Although members of the Nedd4 family ordinarily interact with substrates via their so-called WW domains, DMT1 does not have any WW domains. Instead, DMT1 interacts with 2 WW domain-interacting proteins called Nedd4 family-interacting proteins 1 and 2 (Ndfip1 and Ndfip2), which are considered to have roles in protein trafficking. Interestingly, Ndfip1 knockout mice have increased DMT1 plus notable hepatic iron deposition, making Ndfip1 relevant to iron homeostasis. Using human neurons, Howitt and collaborators  found that the Ndfip1 is up-regulated and binds to DMT1 in response to Fe and cobalt (Co) exposure. Apparently, this response mediates the degradation of DMT1 to decrease metal entry. Ndfip1 over-expression protects neurons from metal toxicity; and decreasing Ndfip1 by RNAi leads to hypersensitivity to metals. Brains from Ndfip1 knockout mice exposed to Fe accumulate excess Fe within neurons. These data suggest that iron transporters and their regulation are very critically involved in neurodegeneration.
There is another iron transporter connection to neuronal properties that provokes re-examination of some older metabolic data . Carlson et al. created mice with hippocampal ablation of DMT1; local iron deficiency accompanied the loss of DMT1. Spatial memory training induced the increased hippocampal DMT1 expression in normal mice, but those with a hippocampal DMT1 knockout performed less well on spatial memory tasks. Hippocampal neurons are richly populated with N-methyl d-aspartate (NMDA) receptors that may well participate in the learning process. Another group  has shown that brief exposure of primary hippocampal cultures to 50 μM NMDA induces 24 h later increased expression of 1B and +IRE DMT1 mRNA, but not of DMT1−IRE mRNA. Both actinomycin D and the NMDA receptor antagonist, MK-801, block the induction. 1B and +IRE DMT1 mRNA also went up in the hippocampus of rats after spatial memory training. The mice with hippocampal ablation of DMT1 had weakened energy metabolism and diminished glutamatergic neurotransmission . While the basis for weakened energy metabolism is likely to be the connection between local iron deficiency and mitochondrial dysfunction as reviewed previously , there is a surprising but ultimately reasonable linkage between iron transport and glutamate signaling . McGahan et al. start from the recognition that IRP1 when iron replete is cytosolic aconitase, thus one can conclude that this enzyme is regulated by iron status. It catalyzes the reaction citrate → isocitrate. Cytoplasmic isocitrate dehydrogenase can then convert isocitrate → α-ketoglutarate while generating a reducing equivalent as NADPH. The reaction α-ketoglutarate → glutamate where glutamate dehydrogenase reductively aminates the substrate is a source of glutamate. The authors demonstrate that retinal-pigmented epithelial cells and cultured neuronal cells generate glutamate in an iron-dependent fashion. They point out that dysregulation of iron homeostasis and glutamate metabolism frequently occur together in neurodegenerative conditions. Although the excitatory toxicity of the glutamatergic response and the ROS contributions of excess iron usually attract attention from very different groups of researchers, this work points out that they should be considered together, even coordinately.
The same group  also uncovered a relationship where the level of iron transported into retinal-pigmented epithelial cells, neurons, or cultured lens epithelial cells would affect the cells’ ability to manage ROS. Glutamate is also a precursor of the cellular reductant glutathione (GSH) and critically involved in making cysteine available for GSH synthesis via a glutamate/cystine antiporter. In this study, the authors showed that iron regulates l-cystine uptake and the downstream production of GSH. Thus, at least under normal circumstances, increased iron importation could lead to improved management of ROS. This series of results and many other issues in iron homeostasis for the eye have been recently reviewed by the same group .
Potential and actual human iron transporters
What it does or might do
Divalent metal (ion) transporter
It is the major duodenal importer of ferrous iron and other metal ions; it is also responsible for exit of ferrous iron from endosomes during the Tf cycle; this transporter also participates in Tf-independent iron entry into cells (non-Tf-bound iron uptake) and possibly in transcytosis of iron across enterocytes
Divalent cation transporter
Natural resistance–associated macrophage protein 2
Solute carrier 11a2
It is the only known cellular iron exporter in mammals; this distinction is physiologically important even if the ability of DMT1 to support endosomal iron exit makes it also a semantic point
Metal transporter protein
Solute carrier 40
Natural resistance–associated macrophage protein 1
Identified initially by the phenotype of resistance to many microbial infections, this transporter probably participates in the exit of iron and manganese from phagosomes
It carries ferric iron in the plasma, lymph, and cerebrospinal fluid behaving as an iron transporter in conjunction with TfR1
Tf receptor 1
It binds Tf to deliver iron into cells with the transport process due to receptor-mediated endocytosis
Tf receptor 2
It may also deliver Tf-bound iron and even iron not bound to Tf to some cells
It is primarily an iron storage protein, but H ft may also participate in iron transport by binding to TfR1
Zrt- and Irt-like
Although it was initially identified as a Zn transporter, Zip 14 is responsible for at least a part of iron not bound to Tf uptake in the liver. More speculatively, Zip8 may also be an iron transporter
Ngal 24p3 lipocalin(-2) uterocalin
Neutrophil gelatinase-associated lipocalin
This protein binds iron bound to an internal siderophore that is yet to be identified. It and its receptor are involved in iron withholding during infection and early kidney development
Some Ca channels may also be routes for iron uptake. In addition, there is evidence that the G185R mutation of DMT1 leads to enhanced behavior as a Ca channel for DMT1
Transient receptor potential mucolipidosis–associated protein
Recently shown to act as a channel for iron exit from endosomes and lysosomes, it is best known as a protein in which mutations can cause mucolipidosis type IV disease
It was postulated to be a form of calreticulin but also a major iron transporter, but whether it is or not is not known
I thank Dr. AL Crumbliss for drawing to my attention that it might be more felicitous to what actually happens to consider that Steap3 reduction could occur before release of iron from transferrin. Dr. Laura Garrick carefully read and critiqued multiple versions of this review.
- 1.Andolfo I, De Falco L et al (2010) Regulation of divalent metal transporter 1 (DMT1) non−IRE isoform by the microRNA Let-7d in erythroid cells. Haematologica 95(8):1244–1252Google Scholar
- 13.Gao J, Chen J et al (2010) Hepatocyte-targeted HFE and TFR2 control hepcidin expression in mice. Blood 115(16):3374–3381Google Scholar
- 33.Mackenzie B, Shawki A et al (2010) Calcium-channel blockers do not affect iron transport mediated by divalent metal-ion transporter-1. Blood 115(20):4148–4149Google Scholar
- 41.Roetto A, Di Cunto F et al (2010) Comparison of three Tfr2-deficient murine models suggests distinct functions for TFR2 alpha and beta isoforms in different tissues. Blood 115(16):3382–3389Google Scholar
- 47.Soe-Lin S, Apte SS et al (2010) Both Nramp1 and DMT1 are necessary for efficient macrophage iron recycling. Exp Hematol 38(8):609–617Google Scholar
- 51.Wareing M, Ferguson CJ et al (2000) In vivo characterization of renal iron transport in the anaesthetized rat. J. Physiol (Lond.) 524(2): 581–586Google Scholar
- 56.Zhang A, Enns C (2009) Molecular mechanisms of normal iron homeostasis. Hematol Am Soc Hematol Educ Progr 2009:207–214Google Scholar