HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Rolfs, A. Articles by Hediger, M. A. Search for Related Content PubMed PubMed Citation Articles by Rolfs, A. Articles by Hediger, M. A. Related Collections Review articles

Metal ion transporters in mammals: structure, function and pathological implications

Andreas Rolfs and Matthias A. Hediger

MS 9332 Received 2 March 1999; accepted after revision 29 April 1999.

Despite the importance of metal ions in several catalytic functions, there has been, until recently, little molecular information available on the mechanisms whereby metal ions are actively taken up by mammalian cells. The classical concept for iron uptake into mammalian cells has been the endocytosis of transferrin-bound Fe³⁺ by the transferrin receptor. Studies with hypotransferrinaemic mice revealed that in the intestine mucosal transferrin is derived from the plasma and that its presence is not required in the intestinal lumen for dietary iron absorption. This suggests that, at least in the intestine, other non-receptor-mediated uptake systems exist. The molecular identification of metal ion transporters is of great importance, in particular since an increasing number of human diseases are thought to be related to disturbances in metal ion homeostasis, including metal ion overload and deficiency disorders (i.e. anaemia, haemochromatosis, Menkes disease, Wilson's disease), and neurodegenerative diseases (i.e. Alzheimer's, Friedreich's ataxia and Parkinson's diseases). Furthermore, susceptibilities to mycobacterial infections are caused by metal ion transporter defects. The pathological implications of disturbed metal ion homeostasis confirm the vital roles these metal ions play in the catalytic function of many enzymes, in gene regulation (zinc-finger proteins), and in free radical homeostasis. Recent insights have significantly advanced our knowledge of how metal ions are taken up or released by mammalian cells. The purpose of this review is to summarize these advances and to give an overview on the growing number of mammalian metal ion transporters.

Functional role of iron

Iron is required in all organisms for growth and crucial metabolic pathways. The redox potential of Fe²⁺/Fe³⁺ favours its use in a number of protein complexes, especially those involved in electron transfer. A number of proteins require iron for activity in the form of haeme or iron-sulfur clusters to transfer electrons. Iron complexes are not only necessary in the electron transport chain to supply cells with energy, but they are also affected by oxygen radicals (O₂^-·), and free Fe²⁺ is part of the Fenton reaction to generate reactive oxygen species (Henle & Linn, 1997). Therefore, the maintenance of iron homeostasis in the body as well as in the cells must be balanced, to supply enough iron for the metabolism, and to avoid excessive, toxic levels. Regulation of iron uptake also depends on the state of oxygenation. Studies of duodenal brush-border membranes in rat indicate that iron absorption is increased during chronic hypoxia (O'Riordan et al. 1997).

In the presence of oxygen, ferric iron (Fe³⁺) is the favoured species, but in the organism ferrous iron (Fe²⁺) is required. The uptake, and transport, of iron under physiological conditions requires special mechanisms, because Fe³⁺ has a very low solubility at neutral pH in oxygenated fluids (< 10^-17 mol l^-1 at pH 7·4; Harford, 1994). In daily diet two distinct forms of iron are present, namely non-haeme iron (Fe³⁺) and haeme iron. The rate-limiting step of iron uptake appears to be in the intestine, where high amounts of iron present in the diet have to be absorbed. In mammals, the best-studied uptake mechanism of iron is the process of transferrin receptor-mediated endocytosis (van Eijk & de Jong, 1992; Harford, 1994; Richardson & Ponka, 1997). However, there are two observations that indicate that this is not the pathway by which iron is taken up into the body. First, apo-transferrin is not available in the intestinal lumen, except from biliary excretion (Green et al. 1968; Iancu et al. 1995), which is insufficient to account for dietary iron absorption. Second, experiments with brush-border membrane vesicles suggested that other, non-receptor-mediated iron uptake systems exist in the intestine (Eastham et al. 1977; Teichmann & Stremmel, 1990). The acidic pH in the proximal intestine and/or the reduced pH of 6·0 in the unstirred layer close to the external surface of the intestinal brush-border membrane help to solubilize Fe²⁺, which is rendered in its reduced form by ascorbate, and a ferrireductase (Wien & Van Campen, 1991; Raja et al. 1992; Dorey et al. 1993; Inman et al. 1994; Jordan & Kaplan, 1994; Han et al. 1995; Umbreit et al. 1996).

Interestingly, the process of transferrin receptor-mediated endocytosis, thought to be the principal way of uptake of Fe³⁺ into non-intestinal cells, did not lead to an explanation of how iron can cross the endosomal membrane. Studies on the process of transferrin receptor-mediated endocytosis led to the observation that these endosomes need to be acidified. The low endosomal pH is necessary for release of iron from transferrin. Furthermore, the transfer from the endosomes into the cytosol requires the activity of an ferrireductase as well as an Fe²⁺ transporter, because iron exists primarily as Fe²⁺ in the cytosol (Dautry-Varsat, 1986; van Eijk & de Jong, 1992).

The recently cloned plant ferrireductase (Robinson et al. 1999) may shed light on eukaryotic iron acquisition. However, a mammalian homologue has not yet been identified. In yeast the combination of a ferrireductase/-oxidase reaction probably makes iron accessible for uptake from stable organic complexes (Askwith & Kaplan, 1998; Dancis, 1998).

Evidence for two types of iron transporters

The activity of more than one transporter for iron in the mammalian intestine was already promoted in 1977 by studies of Eastham and colleagues (Eastham et al. 1977) who found that vesicles from the brush-border or basolateral membrane show different kinetics of Fe²⁺ uptake. These findings were supported by those of Teichmann & Stremmel (1990) on Fe³⁺ uptake in intestinal vesicles and by studies of mice suffering from microcytic hypochromic anaemia. This anaemia is indistinguishable from iron deficiency anaemia but is unresponsive to increased dietary iron. In microcytic anaemia (mk) mice a transporter located in the brush-border membrane of microvillus cells was thought to be mutated, thereby blocking iron uptake into these cells. In sex-linked anaemia (sla) mice iron uptake into the microvillus cells is detectable, but the iron is not released into the serum (Anderson et al. 1998). These studies indicate two types of transporter proteins, located in the brush-border and basolateral membranes, respectively (Fig. 1).

		View larger version [in this window] [in a new window]
Figure 1. Iron transport in intestinal (villus) cells Dietary iron (Fe3+) in the lumen is taken up at the apical side by DCT1 after its reduction to Fe2+. In the cytosol, Fe2+ is bound to low molecular weight complexes, or proteins, or stored by binding to ferritin. At the basolateral side, Fe2+ is released from villus cells into the plasma by a transporter which requires a multicopper oxidase (the sla protein). The released Fe3+ is instantly transferred to transferrin.

In recent years, various iron transporters have been identified for lower organisms, including the Fe²⁺ transporters feoB from E. coli (Kammler et al. 1993), FET4 from yeast (Dix et al. 1994), the plant IRT1 (Eide et al. 1996), and the Fe³⁺ transporter FTR1 from yeast (Stearman et al. 1996). However, until recently the transporters which mediate direct uptake of iron and metal ions into mammalian cells remained elusive.

Functional cloning of rat DCT1

Using expression cloning with Xenopus laevis oocytes, our laboratory isolated the divalent cation/metal ion transporter (DCT1) from a duodenal cDNA library, prepared from mRNA from rats fed a low-iron diet (Gunshin et al. 1997). DCT1 was isolated by screening this library using a radiotracer assay of ⁵⁵Fe²⁺ uptake in Xenopus oocytes. The isolated cDNA clone encodes a 561 amino acid polypeptide, which, when expressed in oocytes, increases the uptake of ⁵⁵Fe²⁺ more than 200-fold compared with control (water injected) oocytes. The amino acid sequence predicts 12 membrane spanning domains, a glycosylated extracellular loop, a consensus transport motif (CTM) in the fourth intracellular loop, and a topology with N- and C-termini in the cytosol (see Fig. 3). The CTM is thought to interact with ATP coupling subunits, distinct from the nucleotide binding fold of ABC transporters. However, ATP depletion in oocytes did not affect DCT1-mediated iron uptake. DCT1 expression was found to be widespread in rat based on in situ hybridization and Northern blot analysis. Comparison of DCT1 expression in tissues from rats fed a normal diet or an iron-deficient diet for 3 weeks showed that iron deprivation triggers a strong increase of DCT1 mRNA levels in intestine and to some extent in all tissues examined. The finding of an iron-responsive element (IRE) in the 3' untranslated region of the cloned DCT1 cDNA indicates regulation of DCT1 at the RNA stability level, analogous to the transferrin receptor mRNA. Recent studies in our laboratory on the function of the DCT1-IRE support this hypothesis and suggest that the iron-response element binding protein 1 (IRP1) binds to this IRE (M. A. Hediger, unpublished data).

Electrophysiological studies of DCT1 expressed in oocytes revealed that the divalent cation transporter is electrogenic, with Fe²⁺ evoking currents of up to 1000 nA and an apparent K_m for Fe²⁺ of 6 µM. These currents were both voltage- and pH-dependent, indicating that an acid pH is necessary for functional transport of iron. Furthermore, it was shown by these experiments that DCT1 transports not only Fe²⁺, but also Zn²⁺, Mn²⁺, Cu²⁺, Co²⁺, Cd²⁺ and Pb²⁺.

The cloning and characterization of DCT1 provided the first demonstration of an active, cellular uptake mechanism for divalent cations (including Fe²⁺) in mammalian cells. The functional properties of DCT1 show the need of an acidic environment, as found in the proximal duodenum, but also in the transferrin receptor-mediated endosomes. Localization studies by immunofluorescence and confocal microscopy show the presence of DCT1 primarily in recycling endosomes and also in the plasma membrane (Gruenheid et al. 1999). DCT1 turns out to be the rat homologue of the previously identified mouse and human Nramp2 and these proteins are homologous (70 % identical) to Nramp1, the natural resistance-associated macrophage protein (Vidal et al. 1993, 1995b). The biological functions of the Nramp proteins remained unknown, until DCT1 was identified. The name Nramp was given to these proteins because a defect in Nramp1 leads to lack of resistance to infections of macrophages (Vidal et al. 1995b; Gruenheid et al. 1997).

Positional cloning of Nramp2 (DCT1) in mouse and rat

Independently of our cloning of DCT1, Andrews and colleagues (Fleming et al. 1997) identified mouse Nramp2/DCT1 as an iron transporter, by using a genetic approach. That approach stems from their search for the gene for microcytic anaemia (mk) in mice. The severe microcytic hypochromic anaemia is indistinguishable from iron deficiency anaemia but is unresponsive to increased dietary iron. Iron injections, intended to circumvent the intestinal block, did not reverse the anaemia, suggesting a block of iron entry into red blood cell precursors. The locus for mk was found to be linked to an area of the mouse genome where the Nramp2/DCT1 gene was found earlier (Gruenheid et al. 1995; Vidal et al. 1995a). By using RT-PCR analysis for comparison of Nramp2/DCT1 cDNA in wild-type and mk mice, a single point-mutation, substituting arginine for glycine (G185R) within transmembrane domain 4 of mk-Nramp2/DCT1 was found (Fleming et al. 1997). Given the capacity of Nramp2/DCT1 to transport several other divalent heavy metal ions besides iron, the phenotype of the mk/mk mouse might be explained in terms of deficiencies of other metal ions such as Zn²⁺, Mn²⁺ and Cu²⁺ (Chua & Morgan, 1997).

Similar to mk mice, the Belgrade (b) rats have a severe hypochromic, microcytic anaemia that is associated with defects in erythroid iron utilization and intestinal iron uptake in the apical membrane of the enterocytes. Andrews and colleagues have reported that the b rats also have the G185R mutation in transmembrane domain 4 (Fleming et al. 1998). Expression of this mutant protein in eukaryotic cells did not result in stimulation of iron uptake (Su et al. 1998). It is likely that a subset of human patients with congenital anaemia also harbour mutations in DCT1. Indeed, isolated families with an apparently autosomal recessive inherited iron-deficiency anaemia, unresponsive to iron therapy, have been described (Hartman & Barker, 1996). The affected individuals have a phenotype reminiscent of that of the b rat and mk mouse.

Nramp1 cloning and pathophysiology

Nramp1 (natural resistance-associated macrophage protein 1) was found by its impaired function in macrophages in response to infections with various species of mycobacteria and other intracellular parasites (Atkinson & Barton, 1998). The Nramp1 gene was identified by positional cloning, and found to be almost exclusively expressed in macrophages (Vidal et al. 1993, 1995b). The amino acid sequence is highly homologous to DCT1/Nramp2 (73 %), suggesting a similar metal ion transport function for Nramp1 in macrophages. Expression of Nramp1 in oocytes suggested a lower affinity for iron compared with DCT1 (Gunshin et al. 1997). However, expression levels were low in these experiments and further studies are required to elucidate the functional characteristics of Nramp1. In experiments in which Nramp1 was expressed in monkey COS-1 cells it was shown that iron uptake was not increased, but there was a reduced cellular iron content (Atkinson & Barton, 1998). Localization studies show that Nramp1 protein is present in the lysosomal compartment of macrophages, and in phagosomal membranes during phagocytosis (Gruenheid et al. 1997). Its localization does not overlap with that of DCT1/Nramp2, which is also expressed in these cells (Gruenheid et al. 1999). Given this localization, together with the observed metal ion transport, Nramp1 might play a role in resistance to infections by depleting the phagosome of Fe²⁺, Mn²⁺ or other essential divalent metal ions. Recently, it has been reported that tuberculosis infections in West Africans are linked to polymorphism in the Nramp1 gene (Bellamy et al. 1998), indicating that the successful development of treatment strategies for mycobacterial infections requires evaluation of these polymorphisms.

Additional iron transporters

Polarized epithelial cells have two different iron transporters, one in the apical and one in the basolateral membrane (Fig. 1). This is supported by findings in mice with sex-linked anaemia, sla mice. These mice suffer from microcytic anaemia similar to mk mice. mk animals have a disturbed uptake of iron in the intestine, whereas the export out of the intestine is normal. In contrast, the sla animals show the opposite disorder. Their uptake of iron into the villus cells appears normal, and is probably mediated by DCT1, but the release into the blood is compromised. Recent attempts to clone the basolateral transporter indicate that it is composed of at least two subunits, one necessary for Fe²⁺ transport, and the other one for oxidation of Fe²⁺ to Fe³⁺ (Anderson et al. 1998; McKie et al. 1998; Vulpe et al. 1999). The recently cloned hephaestin, deficient in sla mice, is a membrane-bound homologue of ceruloplasmin, and probably functions as multi-copper ferrioxidase. The protein contains only one transmembrane domain, indicating that hephaestin is not the basolateral iron transporter itself. Most probably it interacts with the transporter and facilitates release of iron into the blood by its ferrioxidase activity (Vulpe et al. 1999).

A stimulator of iron transport (SFT) has also been isolated (Gutierrez et al. 1997, 1998). This protein was reported to increase the uptake of iron into Xenopus oocytes, but it is not clear whether this is due to transport or (as indicated by the name) to stimulation of an endogenous transport protein, and whether Fe²⁺ or Fe³⁺ is the substrate in vivo (Yu & Wessling-Resnick, 1998a,b).

In addition to these iron transporters, frataxin, which is defective in the mitochondria of patients with Friedreich's ataxia has been isolated and shown to be an iron transport exit mechanism for mitochondria (Campuzano et al. 1996; Babcock et al. 1997; Kispal et al. 1997; Koutnikova et al. 1997; Wilson & Roof, 1997; Radisky et al. 1999). Its defect leads to iron accumulation in the myocardium of patients. Whether the human ABC7, a homologue of the yeast mitochondrial ABC-type iron transporter Atm1p, functions in addition to frataxin as an iron exporter in mitochondria has yet to be proven (Kispal et al. 1997; Csere et al. 1998).

Sensing body iron, iron transport and hereditary haemochromatosis

One of the most common genetic disorders in man, hereditary haemochromatosis (HH), is caused by the mutation of a MHC II protein, HFE. Recently it was reported that the HFE protein interacts with the transferrin receptor (TfR), and in the case of HH that this interaction is disrupted, giving rise to the long-lasting iron overload in the body (Parkkila et al. 1997; Feder et al. 1998; Gross et al. 1998; Zhou et al. 1998).

As alluded to above, DCT1 mRNA levels in the intestine are tightly controlled by serum iron levels. Intestinal crypt cells express HFE and TfR (Anderson et al. 1994; Wood & Han, 1998), whereas mature villus cells express DCT1, but not HFE (Anderson et al. 1994; Wood & Han, 1998). These findings led us to speculate that HFE and TfR together sense serum iron in crypt cells. We propose that HFE and TfR in crypt cells regulate the expression of the proteins involved in iron absorption in villus cells, including DCT1, via the IRE/IRP system (Fig. 2) (Waheed et al. 1999). In patients suffering from HH, this regulation may be disturbed in crypt cells, leading to a higher expression rate of proteins involved in iron absorption in mature villus and in increased intestinal iron absorption. This hypothesis is supported by studies of hypotransferrinaemic mice (Simpson et al. 1991). In these animals a hyperabsorption of iron is observed, in conjunction with a reduced level of transferrin expression (1-2 % of normal) (Buys et al. 1991). A recent report by Feder and coauthors (Roy et al. 1999) showed a direct effect of HFE on Tf-mediated, but not on non-Tf-mediated, iron uptake in a non-polarized cell line. In addition, Andrews and colleagues showed that, in heterozygous TfR knock-out mice, reduced levels of TfR expression give rise to microcytic, hypochromic anaemia, due to impaired iron uptake by maturating erythrocytes (Levy et al. 1999). The iron content in the liver and spleen of these mice was reduced, probably due to reduced intestinal iron absorption, and possibly as a consequence of disturbed iron sensing by TfR-HFE interaction in crypt cells. This concept is consistent with recent findings by Sly and colleagues addressing regulation of intestinal iron absorption in HFE knock-out mice. These mice were shown to have increased duodenal DCT1/Nramp2 levels, despite high serum iron concentration (Fleming et al. 1999).

		View larger version [in this window] [in a new window]
Figure 2. Hypothetical model of iron sensing and uptake in the intestine and liver Two Fe3+ bound to Tf (Fe2TF) are taken up via the Tf-TfR cycle in the intestinal crypt cells as well as in other cells of the body such as hepatocytes. Intact HFE and TfR are required in crypt cells for iron absorption from the blood, thereby serving as a sensor of body iron. We hypothesize that the amount of iron absorbed in crypt cells determines the stability of DCT1 mRNA along the crypt-to-villus axis, and thereby the amount of DCT1 protein expressed in mature villus cells. A similar regulatory mechanism might exist for other proteins involved in intestinal iron absorption. In patients with atransferrinaemia or haemochromatosis this sensing is predicted to be disturbed, leading to increased iron uptake.

DCT1 as a metal ion transporter with broad selectivity

The finding that DCT1 mediates active cellular uptake of not only Fe²⁺, but also Zn²⁺, Mn²⁺, Cu²⁺, Co²⁺, Ni²⁺ and the toxic metal ions Pb²⁺ and Cd²⁺, was surprising. The existence of a common intestinal absorptive mechanism for a range of metals has important nutritional implications and emphasizes the potential interplay among these essential trace minerals at the level of their absorption. For example, dietary copper deficiency can cause microcytic anaemia, indistinguishable from iron-deficiency microcytic anaemia, suggesting that a defect in DCT1 may also lead to anaemia due to defective copper uptake. In addition, mk/mk mice also suffer from skin lesions reminiscent of defective zinc uptake. b rats are also reported to have widespread abnormalities in manganese metabolism, including impaired duodenal and reticulocyte uptake (Chua & Morgan, 1997; Savigni & Morgan, 1998). Thus, consistent with the substrate range which we determined for DCT1, the metabolism of a variety of metal ions including iron, manganese, cobalt (Savigni & Morgan, 1998) and zinc are deranged due to the DCT1 mutation in the mk and b alleles, underlining the importance of DCT1 in the metabolism of these metals.

The similar affinity of transferrin for other metal ions (Zn²⁺, Mn²⁺, Cu²⁺ and Al³⁺) suggests that this mechanism mediates uptake of a variety of other metal ions. Thus, in non-intestinal tissues, the TfR-DCT1 endocytotic uptake pathway might allow uptake of not only iron but a range of other metal ions.

Uptake of manganese was previously thought to be related to iron uptake (Chua et al. 1996). At least three mechanisms were described based on studies of erythroid cells, one using manganese-transferrin, one involving a high-affinity Mn²⁺ transporter, and one involving a low-affinity Mn²⁺ transporter (Chua et al. 1996). It is likely that the previously reported high-affinity transporter for Mn²⁺ in erythroid cells involves TfR and DCT1, but this still needs verification (Savigni & Morgan, 1998).

Other metal ion uptake transporters

Thus far, several divalent cation transporters with an affinity for single metal ions have been identified in lower organisms. Most of these are thought to be mono-specific. They include the Fe²⁺ transporters feoB from E. coli (Kammler et al. 1993), FET4 from yeast (Dix et al. 1994), the plant IRT1 (Eide et al. 1996), and the Fe³⁺ transporter FTR1 from yeast (Stearman et al. 1996). The first Zn²⁺ transporters ZRT1 and 2 were isolated from yeast (Zhao & Eide, 1996a,b), and a Zn²⁺-translocating P-type ATPase in E. coli has been identified (Rensing et al. 1997), which was recently reported to function as a Pb²⁺-transporting P-type ATPase (Rensing et al. 1998). A manganese transporter belonging to the ABC transporter superfamily has been found in a cyanobacterial mutant strain (Sambongi et al. 1997). The Zn²⁺ transporters ZIP1 to 4 were identified in Arabidopsis (Grotz et al. 1998). SMF1 to 3 are yeast homologues of the mammalian Nramp proteins, and recent studies indicated that SMF1 is expressed in the yeast plasma membrane where it mediates Zn²⁺ and Mn²⁺ uptake (Supek et al. 1996). Studies in our laboratory indicate that SMF1 isoforms expressed in Xenopus oocytes mediate Fe²⁺ transport (X.-Z. Chen & M. A. Hediger, unpublished observations).

Recently, a human gene involved in high-affinity cellular uptake of copper (CTR-1) has been identified by complementation in yeast (Zhou & Gitschier, 1997). In yeast, three copper uptake proteins are described, CTR-1, CTR-2, and CTR-3. Whereas CTR-1 and CTR-3 are high-affinity copper transporters, CTR-2 is thought to be a low-affinity transporter (Askwith & Kaplan, 1998; Eide, 1998). Using CTR-1- and -3-deficient yeast, a mammalian cDNA library was screened for clones that restore copper uptake into these cells. Using this functional complementation approach, human CTR-1, which showed a low level of homology to yeast CTR-1, was isolated. The amino acid sequence of hCTR-1 predicts three transmembrane domains, analogous to the yeast CTR-1. The extracellular N-terminus has several histidine, serine and methionine residues, which are thought to be important for binding of copper ions in bacterial copper ATPases. A functional analysis of hCTR-1 has not yet been reported. The human CTR-1 sequence has been used for homology screening, giving rise to a putative second human copper transporter, hCTR-2. Its cDNA codes for a similar putative transporter, with pronounced differences in the N-terminus (Zhou & Gitschier, 1997). Clearly, more studies are required to establish the function, distribution and physiological relevance of the CTR transporters in human. The structure of these transporters is of particular interest since they only have three transmembrane domains.

There are a number of metal ion transporters identified in yeast, plant and bacteria, for which mammalian homologues have not yet been identified (Paulsen & Saier, 1997; Eng et al. 1998; Grotz et al. 1998). Thus, much work will be required to fully elucidate the molecular identity and properties of these metal ion transporters.

Metal ion export transporters

Several divalent cation export systems have been reported recently. These include the Zn²⁺ export systems ZnT-1 to ZnT-4 (Palmiter & Findley, 1995; Huang & Gitschier, 1997; Palmiter et al. 1996a,b; review articles: Huang, 1997; McMahon & Cousins, 1998), and two highly homologous P-type copper ATPase exporters, the Menkes (MNK) and Wilson's (WND) disease proteins (Mercer et al. 1993; Tanzi et al. 1993; Vulpe et al. 1993; Petrukhin et al. 1994). With respect to the ZnT family members, a naturally occurring gene defect of ZnT-4 in mouse has been reported, leading to the phenotype called 'lethal milk' (Huang & Gitschier, 1997). Targeted gene disruption of ZnT-3 did not lead to a specific phenotype. This was of great interest, since the protein is localized on synaptic vesicles of zinc-containing neurons in the hippocampus of wild-type animals, and is thereby associated with zinc transport into these vesicles (Wenzel et al. 1997; Cole et al. 1999).

The Menkes disease protein (MNK) is a P-type ATPase Cu²⁺ transporter. The MNK cDNA was the first mammalian heavy metal ion transporter cDNA to be cloned. In patients suffering from Menkes disease, the transport of Cu²⁺ is defective, leading to damage of certain tissues, neurodegeneration, and to death in early childhood. The cloning of the Menkes disease gene (MNK) was performed in part by positional cloning (Mercer et al. 1993; Chelly et al. 1993) and exon trapping (Vulpe et al. 1993). The cDNA (8·5 kb) predicts a protein of 1500 amino acids with eight transmembrane domains, six putative copper binding sites within a 600 amino acid N-terminal domain, an ATP-binding motif, a phosphatase domain, and several invariant amino acid residues in the proposed cation transduction channel (Vulpe & Packman, 1995; Lutsenko et al. 1997). Mutations in the MNK gene which cause Menkes disease give rise to accumulation of copper in cell cultures, and lead to a deficiency of enzymes that need copper for their activity. In patients, Cu²⁺ accumulates predominantly in the intestine and kidney, due to (re)absorption of Cu²⁺ in these organs. MNK is expressed in all tissues except the liver, where its function is probably mediated by the Wilson protein.

The Wilson's disease protein (WND) was thought to be related to a defect in the blood Cu²⁺ transport protein ceruloplasmin, since in this disease the amount of ceruloplasmin and its Cu²⁺ content are reduced. The clinical picture of Wilson's disease is hepatic cirrhosis and neuronal degradation in early childhood, due to the lack of Cu²⁺ release from the bile, and Cu²⁺ overload of hepatic cells. However, cloning of the WND gene showed that it encodes a Cu²⁺-ATPase consisting of 1465 amino acids, found to be expressed in liver and kidney (Tanzi et al. 1993). The sequence predicts eight transmembrane domains, six Cu²⁺ binding domains, and the characteristic P-type ATPase features. WND was cloned in part by positional cloning, and in part by its homology to MNK (54 % homology of the proteins). Experiments with hepatic cell lines revealed only low expression of WND on the plasma membrane. Most of its expression was found in the trans-Golgi network (Nagano et al. 1998), together with ceruloplasmin. This might indicate that Cu²⁺ is not released as a free ion into the blood, but is bound to ceruloplasmin, a copper transport protein thought to be secreted in the bile only after complete Cu²⁺ loading (Chowrimootoo et al. 1996). WND localized in the trans-Golgi network is thought to migrate close to the plasma membrane in a copper-dependent manner. Such a copper-dependent re-localization was also described for the MNK protein (Petris et al. 1996, 1998). Cu²⁺, but not Zn²⁺, Fe²⁺, Cd²⁺ or Co²⁺, was able to stimulate this migration of WND from the Golgi network to the plasma membrane (Hung et al. 1997). This specific effect of copper on the relocalization of WND is interesting, since zinc acetate is very efficient for the treatment of patients suffering from Wilson's disease, with minimal side effects (Brewer et al. 1994; Anderson et al. 1998). Other therapeutic strategies involve the use of Cu²⁺ chelating agents (d-penicillamine and trientine). However, these components show significant side effects. A detailed description of Cu²⁺-ATPases, their pathophysiological impacts, and their interaction with copper chaperons are presented elsewhere in excellent recent articles (Vulpe & Packman, 1995; DiDonato & Sarkar, 1997; Pufahl et al. 1997; Askwith & Kaplan, 1998).

Structure of human metal ion transporters

The diversity of the predicted structures of metal ion transporters from different transporter families is striking. Table 1 summarizes these differences and Fig. 3 shows topology models of selected transporters. Except for the DCT1/NRAMP2 family, metal transporters do not comply with the 12 transmembrane domain (TMs) dogma of transport proteins. The MNK and WND proteins have eight putative TMs consistent with the structure of other ATP-driven transporters. Interestingly, the ZnTs and the SFT have only six TMs and the CTRs only three TMs. The low number of TMs in the CTRs gives rise to the question of whether these transporters function as homo- or heteromultimers, since a single protein would probably not be able to form a pore complex sufficient for the translocation of the metal ion.

Table 1. Mammalian metal ion transporters