Iron and calcium are required for general cellular functions, as well as for specific neuronal-related activities. However, a pathological increase in their levels favours oxidative stress and mitochondrial damage, leading to neuronal death. Neurodegeneration can thus be determined by alterations in ionic homoeostasis and/or pro-oxidative–antioxidative equilibrium, two conditions that vary significantly in different kinds of brain cell and also with aging. In the present review, we re-evaluate recent data on NTBI (non-transferrin bound iron) uptake that suggest a strict interplay with the mechanisms of calcium control. In particular, we focus on the use of common entry pathways and on the way cytosolic calcium can modulate iron entry and determine its intracellular accumulation.
The control of iron in brain is still a debated issue, with open questions and controversial results. Several aspects make research in this field difficult: the complexity of iron transport through the blood–brain barrier; the presence of different cell types in the CNS (central nervous system) (neurons and macro- and micro-glial cells); distinctive mechanisms governing iron fluxes in specific brain areas; and, finally, an age-dependent control of the expression and activity of molecules involved in iron homoeostasis. A large body of evidence indicates that iron passes the luminal side of brain capillary endothelial cells bound to Tf (transferrin), whereas very little is known about the machinery responsible for the transport through the abluminal side [1–3]. Once across the blood–brain barrier, the concentration of iron exceeds the Tf-binding capacity of the interstitial fluids . This residual iron pool is known as non-Tf-bound iron (NTBI) and is constituted by free iron as well as iron bound to carrier molecules specifically released by the end-feet of astrocytes, such as citrate, ascorbate and ATP , or by transporter proteins such as albumin, lactoferrin and p97 . Most of NTBI circulates in the brain interstitium as Fe3+ and is predominantly bound to lactoferrin, a glycoprotein highly expressed in the substantia nigra, or to p97 (also called melanotransferrin), which is present as both a soluble and a cell-surface GPI (glycosylphosphatidylinositol)-anchored isoform .
The uptake of NTBI constitutes an important iron source for different cell types in the brain: it is involved in the control of intracellular iron homoeostasis and it is suggested to account for iron accumulation in pathological brains [1,4]. Noticeably, increasing levels of the soluble form of p97 are suggested to correlate with Alzheimer's disease progression . Several mechanisms participate in NTBI uptake into neurons, but even though several molecules involved in its regulation have been characterized, their expression and function in the CNS are not yet fully elucidated.
The overall picture is complicated further by the finding that another key player, calcium, can not only regulate NTBI uptake, but also share with it some influx pathways [6a]. This is particularly intriguing, since a fine control of intracellular calcium concentration is required for proper synaptic activity [6a], but its disregulation is a hallmark of a number of stressful neuronal conditions (e.g. excitotoxicity ). In the light of the above considerations, it is not unlikely that the interplay between mechanisms governing calcium and iron concentrations might play a crucial role in some neurodegenerative processes.
Role of calcium influx on DMT1 (divalent metal transporter 1)-mediated iron uptake
DMT1 (the product of the gene SLC11A2 or NRAMP2/DCT1) was identified in 1997 as the first mammalian membrane iron transporter [8,9]. It was soon recognized as being essential for intestinal absorption of Fe2+ and for the endosome-to-cytosol export of Tf-imported iron. As expected, it is expressed on the endosomal membrane as well as on the plasma membrane, where DMT1 can mediate the uptake of a wide range of bivalent metal ions, including Fe2+, Mn2+, Co2+, Cd2+, Cu2+ and Zn2+ , by a proton gradient-dependent transport. Four variant mRNA transcripts arise from the mammalian SLC11A2 gene. They generate four isoforms that differ both at the N-terminus (starting from either exon 1A or exon 1B) and at the C-terminus, in which also part of the 3′-UTR (untranslated region) is changed, thereby including or not including a candidate iron-responsive element [11,12]. The distribution of the different isoforms in the brain is far from established. In situ hybridization of DMT1 mRNA in developing rat brain showed localization in striatum, cortex, hippocampus and cerebellum , whereas immunocytochemistry analysis revealed DMT1 expression in striatum, cerebellum and thalamus, as well as vascular cells throughout the brain and ependymal cells in the third ventricle . Noticeably, different kinds of cell were DMT1-positive in the various regions: mainly astrocytes and endothelial cells in striatum; neurons in the thalamus; and mostly endothelial cells, but also glial end-feet, in the cortext . Although the results from this as well as other studies [15–17] clearly demonstrate the expression of DMT1 in astrocytes, other studies present conflicting evidence .
More studies will be required to reach a unanimous and comprehensive view of cellular localization, and also expression and functional regulation of the metal transporter in the CNS. A recent study, from Snyder's group , first explored an intracellular pathway responsible for DMT1 regulation and iron uptake in neuronal cells. In the proposed model, the activation of nNOS (neuronal nitric oxide synthase) by NMDA (N-methyl-D-aspartate)-mediated calcium influx promotes S-nitrosylation of Dexras1, a brain-enriched member of the Ras family. This modification should allow the interaction of Dexras1 with DMT1 via an adaptor protein, PAP7 (peripheral benzodiazepine receptor-associated protein), thereby promoting a positive modulation of the transporter, with an ensuing increase in Fe2+ uptake into the cells. This mechanism was proposed to exacerbate neuronal injury during acute or chronic insults, with massive calcium influx through the NMDA receptors favouring intracellular iron accumulation. However, despite the evidence of an interaction among Dexras1, PAP7 and DMT1, a direct demonstration of modulation of Fe2+ entry via this pathway is still lacking. The absence of calcium-modulated Fe2+ uptake in nNOS-knockout mice was presented in support of this view, but this result cannot exclude the contribution of different molecules or other mechanisms. For instance, iron could enter through a VOCC (voltage-operated calcium channel), a possibility proposed recently  and discussed below. Since nitric oxide was reported to positively modulate the expression of VOCC subunits  as well as their channel current [21,22], the effect of the absence of nNOS expression on iron uptake is open to explanations alternative to DMT1 involvement. Accordingly, further confirmation of the role of DMT1 in neurons is needed by approaches such as silencing by siRNA (small interfering RNA), or the use of neuronal cultures from animal models, in which the iron-transport capacity of DMT1 is severely impaired by spontaneous mutations (Belgrade rats or heterozygous mutant microcytic mice). It should also be pointed out that, in the work by Cheah et al. , iron-uptake experiments were performed at pH 5.5 to increase the transporter activity. However, DMT1 is able to mediate Fe2+ transport also in the absence of a proton gradient ; moreover, specific DMT1 isoforms differently control iron uptake as a function of pH . Given these premises, it will be important to verify whether a DMT1-mediated iron-entry mechanism is also observed under more physiological conditions (i.e. at pH close to 7) or whether it is restricted to pathological situations only.
A different interplay between DMT1 and calcium was proposed by Xu et al. . They investigated the significance of DMT1 mutations on iron transport in Belgrade rats and in two strains of microcytic mice [8,26]. A common picture of severe systemic iron deficiency and anaemia was observed in all of these animal models, and it was ascribed to the same amino acid substitution G185R, which impairs Fe2+ transport, but confers calcium permeability on DMT1. The authors propose a model in which DMT1 is a proton channel with low permeability for bivalent metal ions; they suggest that the increased calcium permeability in G185R mutants might give some selective advantage favouring this common mutation.
On the basis of these premises, the function of DMT1 was addressed further by Ludwiczek et al.  by investigating the capacity of calcium channel blockers to modulate DMT1-mediated iron uptake. Their results indicate that the dihydropyridine-type blocker nifedipine increases Fe2+ entry 10–100-fold. In the studies in vivo, this is reflected by an increase in both iron accumulation in duodenal tissue and iron excretion at the renal level. Even though the site of action of nifedipine was not defined, it was proposed to prolong the activity of DMT1 via a mechanism independent of the reduction of calcium influx. However, the use of nifedipine as a pharmacological tool able to positively modulate DMT1 activity is made less attractive by a non-specific effect of its photodegraded by-products. Indeed, our results (I. Pelizzoni, unpublished work) and those of others [27,28] show that nifedipine rapidly undergoes degradation, becoming an iron ionophore, thereby allowing iron entry by a DMT1-independent mechanism. Altogether, the issue of DMT1-mediated iron entry in vivo is certainly relevant, but still open to a number of interpretations.
NTBI uptake by pathways common to calcium influx
The VOCC blockers were also reported to have opposite effect on iron uptake. In primary haemochromatosis and secondary iron overload, two conditions that cause cardiomyopathies and increased mortality, an NTBI influx through L-type calcium channels was reported to become the dominant pathway for Fe2+ uptake into the myocardium . This mechanism is particularly harmful since it is not inhibited by iron accumulation so that continuous iron entry is permitted. As expected, the use of L-type calcium channel blockers reduced the iron overload and resulted in cardioprotection.
Gaasch et al.  investigated whether a similar mechanism operates in neuronal cells. In their study, NGF (nerve growth factor)-differentiated rat PC12 cells were treated with high K+, and the influx of radioactive calcium or Fe2+ was measured. Their findings fully confirmed the expectations: the depolarization promoted calcium as well as Fe2+ influx; extracellular Fe2+ inhibited calcium entry in a dose-dependent manner; and the uptake of both cations was prevented by the calcium channel blocker nimodipine.
Considering that VOCCs are highly expressed in the neurons and that the depolarization necessary for their opening is provided by synaptic-neuronal activity [activation of AMPA (α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid) and NMDA receptors, back propagating action potentials etc.], it can be inferred that iron entry through VOCCs might be, at least partially, responsible for the observed NMDA-stimulated iron uptake. As discussed earlier, the involvement of DMT1 in NMDA-modulated iron entry cannot be proved by the lack of modulation in nNOS-knockout animals, since this condition also affects the expression and activity of VOCCs. In any event, this issue should be addressed in primary neuronal cultures, since NGF-treated PC12 cells do not reproduce the complex molecular phenotype of primary neurons. Moreover, it would be possible to investigate whether this pathway can be regulated by neuronal activity, or by the same mechanisms affecting the calcium entry through VOCCs. Finally, the contribution of the VOCC-mediated NTBI entry to iron accumulation in physiopathological conditions could give important information about NTBI regulation in the brain.
Another pathway for both NTBI uptake and calcium influx could be provided by the activation of TRP (transient receptor potential) channels, first identified in Drosophila photocells . In particular, Mwanjewe and Grover  investigated the role of the members of the canonical subfamily of TRP (TRPC ) that are considered possible candidates for store- and receptor-operated calcium channels. These non-selective cationic channels might therefore account for the calcium influx that is observed after store depletion or phospholipase C activation by G-protein-coupled receptors .
There are seven members in the TRPC family, but only the mRNAs of TRPC3 and TRPC6 are well represented in the brain. The authors exploited the property of PC12 cells to express TRPC6 only when they are differentiated to a neuronal-like phenotype by NGF treatment . NTBI uptake was higher in NGF-treated cells compared with untreated cells, and this difference markedly increased upon treatment with a cell-permeant analogue of DAG (diacylglycerol), which physiologically activates TRPC6 in a PKC (protein kinase C)-independent way. These results were confirmed in HEK-293 (human embryonic kidney) cells overexpressing TRPC6. The influx was apparently similar for Fe2+ and Fe3+, but it is possible that TRPC6 is selective for Fe2+ and that Fe3+ ions underwent reduction by ferric reductases. Again, the contribution of this mechanism in primary neurons has to be investigated.
Iron plays an essential role in the brain. It is involved in a number of general cellular functions (electron transport in the respiratory chain, haem and DNA synthesis, etc. ) and also in specific neuronal activities (from neurotransmitter synthesis to myelination, up to cognitive functions ). However, an excess of iron is toxic to cells, since it can react with hydrogen peroxide, leading to oxidative damage . Therefore calcium and iron share similarities, both being necessary for neuronal function, but also requiring tight control to avoid the triggering of neurodegenerative processes. These similarities in action extend further to the use of common entry pathways and to their mutual reinforcement, with calcium able to favour iron entry and thus accumulation, and iron capable of furthering an increase in calcium by inducing oxidative stress [36,37]. Altogether, it is clear that a tight interplay occurs between these two cations and that they are both responsible for cytotoxic effects in the CNS. However, distinctions must be made. Astrocytes are more resistant than neurons to iron challenge, and neurons appear to decrease their resistance with time in culture (I. Pelizzoni and R. Macco, unpublished work). Moreover, neurons and astrocytes in specific brain areas are known to have diverse tendencies to undergo calcium  or iron overload and to be susceptible to death [39,40]. These observations can be partly accounted for by the different levels of intracellular detoxifying enzymes and scavenger molecules able to protect cells against oxidative stress [35,41], but they can also reflect a different capability to handle the two cations. Noticeably, neurons, more liable than astrocytes to death, are subjected to the convergence of synaptic and extrasynaptic signals [42,43] that promote the activation of calcium channels, thereby favouring not only calcium influx, but also iron entry. Moreover, the same signals can sustain the production of the DAG  necessary for activation of PKC-dependent (e.g. calcium reinforcement) and -independent (e.g. TRPC6 modulation) pathways. In this respect, the release of neurotransmitters by astrocytes might well contribute to the activation of the neuronal receptors , a release that would be enhanced further under conditions in which astrocytes undergo the change in phenotype known as ‘activation’ . Overall, this signalling network might account for NTBI uptake under physiological conditions, although further studies will be necessary to elucidate its relevance.
In conclusion, exciting new findings have recently emerged, adding further complexity to the already intricate picture of the neurodegenerative processes. More studies will be required to fully understand iron influx and calcium elevation as well as reactive oxygen species production and their complex interplay in neuronal and glial cells.
Metal Metabolism: Transport, Development and Neurodegeneration: A Biochemical Society Focused Meeting held at Imperial College London, U.K., 9–10 July 2008. Organized and Edited by David Allsop (Lancaster, U.K.) and Harry McArdle (Rowett Research Institute, Aberdeen, U.K.).
central nervous system
divalent metal transporter 1
nerve growth factor
neuronal nitric oxide synthase
peripheral benzodiazepine receptor-associated protein
protein kinase C
transient receptor potential
voltage-operated calcium channel
Financial support was from the Italian Ministry of Research [PRIN (Progetti di Ricerca di Interesse Nazionale) project 2006054051 to F.G.] and the Italian Telethon Foundation (GGP05141 grant to F.G.).