- Open Access
Iron behaving badly: inappropriate iron chelation as a major contributor to the aetiology of vascular and other progressive inflammatory and degenerative diseases
BMC Medical Genomics volume 2, Article number: 2 (2009)
The production of peroxide and superoxide is an inevitable consequence of aerobic metabolism, and while these particular 'reactive oxygen species' (ROSs) can exhibit a number of biological effects, they are not of themselves excessively reactive and thus they are not especially damaging at physiological concentrations. However, their reactions with poorly liganded iron species can lead to the catalytic production of the very reactive and dangerous hydroxyl radical, which is exceptionally damaging, and a major cause of chronic inflammation.
We review the considerable and wide-ranging evidence for the involvement of this combination of (su)peroxide and poorly liganded iron in a large number of physiological and indeed pathological processes and inflammatory disorders, especially those involving the progressive degradation of cellular and organismal performance. These diseases share a great many similarities and thus might be considered to have a common cause (i.e. iron-catalysed free radical and especially hydroxyl radical generation).
The studies reviewed include those focused on a series of cardiovascular, metabolic and neurological diseases, where iron can be found at the sites of plaques and lesions, as well as studies showing the significance of iron to aging and longevity. The effective chelation of iron by natural or synthetic ligands is thus of major physiological (and potentially therapeutic) importance. As systems properties, we need to recognise that physiological observables have multiple molecular causes, and studying them in isolation leads to inconsistent patterns of apparent causality when it is the simultaneous combination of multiple factors that is responsible.
This explains, for instance, the decidedly mixed effects of antioxidants that have been observed, since in some circumstances (especially the presence of poorly liganded iron) molecules that are nominally antioxidants can actually act as pro-oxidants. The reduction of redox stress thus requires suitable levels of both antioxidants and effective iron chelators. Some polyphenolic antioxidants may serve both roles.
Understanding the exact speciation and liganding of iron in all its states is thus crucial to separating its various pro- and anti-inflammatory activities. Redox stress, innate immunity and pro- (and some anti-)inflammatory cytokines are linked in particular via signalling pathways involving NF-kappaB and p38, with the oxidative roles of iron here seemingly involved upstream of the IkappaB kinase (IKK) reaction. In a number of cases it is possible to identify mechanisms by which ROSs and poorly liganded iron act synergistically and autocatalytically, leading to 'runaway' reactions that are hard to control unless one tackles multiple sites of action simultaneously. Some molecules such as statins and erythropoietin, not traditionally associated with anti-inflammatory activity, do indeed have 'pleiotropic' anti-inflammatory effects that may be of benefit here.
Overall we argue, by synthesising a widely dispersed literature, that the role of poorly liganded iron has been rather underappreciated in the past, and that in combination with peroxide and superoxide its activity underpins the behaviour of a great many physiological processes that degrade over time. Understanding these requires an integrative, systems-level approach that may lead to novel therapeutic targets.
Background and preamble
The 'balkanisation' of the literature is in part due to the amount of it (some 25,000 journals with presently 2.5 million peer-reviewed papers per year, i.e. ~5 per minute ), with a number http://www.nlm.nih.gov/bsd/medline_cit_counts_yr_pub.html increasing by something approaching 2 per minute at PubMed/Medline alone. In addition, the disconnect between the papers in the literature (usually as pdf files) and the metadata describing them (author, journal, year, pages, etc) is acute and badly needs filling . Without solving this problem, and without automation of the processes of reading, interpreting and exploiting this literature and its metadata in a digital format, we cannot make use of the existing tools for text mining and natural language processing (e.g. [3–5]), for joining disparate concepts , for literature-based discovery (e.g. [7–11], and for studies of bibliometrics [12, 13], literature dynamics , knowledge domains , detecting republication  and so on. Until we recognise these possibilities we are unlikely to seek to realise them.
The present article (and see  for a preprint) serves to show some of the benefits than can accrue from a more overarching view of the otherwise highly disparate literature in a particular domain (see also ), but was done 'the hard way', i.e. with a few bibliographic and bibliometric tools but without the kind of automation implied above. For the record, the main tools used (see a review in ) were Web of Knowledge and Scopus for literature and citation searching, supplemented by Google Scholar. Some use was also made of ARROWSMITH [6, 19, 20] and GOPubMed , as well as various workflows in the Taverna environment [22–26], including the BioAID_DiseaseDiscovery workflow http://www.myexperiment.org/workflows/72 written by Marco Roos. Citations and attendant metadata were stored in Endnote (latterly version X).
Even under 'normal' conditions, as well as during ischaemia when tissue oxygenation levels are low, the redox poise of the mitochondrial respiratory chain is such that the normally complete four-electron reduction of dioxygen to water is also accompanied by the production, at considerable rates (ca 1–4% of O2 reduced), of partially reduced forms of dioxygen such as hydrogen peroxide and superoxide (e.g. [27–45]). These 1- and 2-electron reductions of O2 are necessarily exacerbated when the redox poise of the b-type cytochromes is low, for instance when substrate supplies are in excess or when the terminal electron acceptor O2 is abnormally low due to hypoxia or ischaemia. Various other oxygenases, oxidases and peroxidases can also lead directly to the production of such 'reduced' forms of dioxygen in vivo (e.g. [46–48]), with H2O2 from xanthine oxidase being especially implicated in ischaemia/reperfusion injury (e.g. [47, 49–54]). These molecules (peroxide and superoxide) can cause or contribute to various kinds of oxidative stress. However, this is mainly not in fact because they can react directly with tissue components themselves, since they are comparatively non-toxic, cells have well-known means of dealing with them , and they are even used in cellular signalling (e.g. [56–60]). Much more importantly, it is because they can react with other particular species to create far more reactive and damaging products such as hydroxyl radicals, with all these agents nevertheless being known collectively (and indiscriminately) as reactive oxygen species (ROSs). Possibly the commonest means by which such much more damaging species, in particular the hydroxyl radical, are created is by reaction with unliganded or incompletely liganded iron ions [61–63]. The themes of this review are thus (i) that it is this combination of poorly liganded iron species, coupled to the natural production of ROSs, that is especially damaging, (ii) that the role of iron has received far less attention than has the general concept of ROSs, albeit the large literature that we review, and (iii) that this basic combination underpins a great many (and often similar) physiological changes leading to a variety of disease manifestations, and in particular those where the development of the disease is manifestly progressive and degenerative.
An overview of the structure of the review is given in Fig 1, in the form of a 'mind map' . The main literature review for this meta-analysis was completed on June 30th, 2008, with some updates being added following the refereeing process.
Some relevant chemistry of iron and reduced forms of oxygen
While superoxide and peroxide are the proximate forms of incomplete O2 reduction in biology, a reaction catalysed by the enzyme superoxide dismutase  serves to equilibrate superoxide and peroxide:
Arguably the most important reaction of hydrogen peroxide with (free or poorly liganded) Fe(II) is the Fenton reaction , leading to the very reactive and damaging hydroxyl radical (OH•)
Superoxide can also react with ferric iron in the Haber-Weiss reaction  to produce Fe(II) again, thereby effecting redox cycling:
Ascorbate can replace O2 •- within the cell for reducing the Fe(III) to Fe(II) . Further reactions, that are not the real focus here, follow from the ability of hydroxyl radicals and indeed Fe(n) directly to interact with many biological macro- and small molecules, especially including DNA, proteins and unsaturated lipids. Thus [69–73], Fe(II) and certain Fe(II) chelates react with lipid hydroperoxides (ROOH), as they do with hydrogen peroxide, splitting the O--O bond. This gives RO•, an alkoxyl radical, which can also abstract H• from polyunsaturated fatty acids and from hydroperoxides. The resulting peroxyl radicals ROO• can continue propagation of lipid peroxidation. Oxidative stress also leads to considerable DNA damage [74–76] and to the polymerisation and denaturation of proteins [77–79] and proteolipids that can together form insoluble structures typically known as lipofucsin (see e.g. [80, 81]) or indeed plaques. Such plaques can also entrap the catalysts of their formation, and thereby point them up. Some of the evidence for these is described below. Many small molecule metabolic markers for this kind of oxidative stress induced by the hydroxyl radical and other 'reactive oxygen species' (ROSs) are known [43, 82–89], and include 8-oxo-guanine [90–94], 8-hydroxy guanine , 8-hydroxy-2'-deoxy-guanosine [96, 97], 8-oxo-GTP , 4-hydroxy-2-hexenal , 4-hydroxy-nonenal , 4-hydroperoxy-2-nonenal, various isoprostanes [101–107], 7-keto-cholesterol , many other cholesterol derivatives , malondialdehyde , neopterin , nitrotyrosine [112–115] and thymidine glycol [116, 117]. Note that the trivial names in common use for this kind of metabolite are not helpful and may even be ambiguous or misleading, and it is desirable (e.g. ) to refer to such molecules using terminology that relates them either to molecules identified in persistent curated datbases  such as ChEBI  or KEGG , or better to describe them via database-independent encodings such as SMILES  or InChI [123–128] strings. (There are other oxidative markers that may be less direct, such as the ratio of 6-keto-prostaglandin F1α to thromboxane B2 , but these are not our focus here.)
Overall, it is in fact well established that the interactions between 'iron' sensu lato and partly reduced forms of oxygen can lead to the production of the very damaging hydroxyl radical (e.g. [43, 130–139]), and that this radical in particular probably underpins or mediates many higher-level manifestations of tissue damage, disease, organ failure and ultimately death[36, 137, 140–143]. While the role of ROSs generally in these processes has been widely discussed, the general recognition of the importance of inadequately liganded iron in each of them has perhaps been less than fully appreciated. One of our tasks here will therefore be to stress this role of 'iron', and to assess the various means of chelating 'iron' such that it does not in fact do this. (Throughout we use 'iron' to refer to forms of Fe(n, n > 0) with unspecified ligands, though we absolutely stress that it is the exact speciation and liganding that determines the reactivity of 'iron' in catalysing reactions such as that of hydroxyl radical formation, and indeed its bioavailability generally – inadequate liganding of iron in the required forms can be a cause of anaemia even if the total amount of 'iron' is plentiful.)
For completeness we note the reactions catalysed by superoxide dismutase
and by catalase
These together, were their activity in the relevant locations sufficiently great, might serve to remove (su)peroxide from cells completely.
In addition to reactive oxygen species there are ions such as the perferryl ion (Fe-O)  and reactive nitrogen species [60, 145–147]. These latter are mainly formed from the natural radical NO, an important inflammatory mediator , with peroxynitrite production (from the reaction of NO and superoxide) [46, 149–154] leading to nitrotyrosine , or nitro-fatty acid [155, 156] or protein cystein nitrosylation [157, 158] being a common means of their detection downstream. Other toxic products of the reactions of NO include NO2, N2O3, and S-nitrosothiols , and the sequelae of some of these may also involve iron .
Overall, we recognise that these kinds of inflammatory, oxidative stress-related reactions are accumulative and somewhat irreversible , that they are consequently age-related, and (see [162–165] and later), and that most diseases and causes of mortality that are prevalent in the developed world are in this sense largely manifestations of this kind of aging.
Ligands and siderophores
As well as the reactions described above, ferrous ions will react with oxygen under aerobic conditions to produce ferric ions, and in natural environments there is little to stop this. Consequently, and because these reflect fundamental physicochemical properties of such ions, the problems of both solubility and toxicity were faced by bacteria (and indeed fungi [166–169]) long ago in evolution, and were solved by their creation and excretion of (mainly ferric-)iron chelators known as siderophores [170–189] (and for haemophores see ). These typically have extremely tight binding constants (Kf > 1030 ) and can solubilise and sequester iron such that it can be internalised via suitable transporter molecules within the bacterial plasma membrane . Bacterial and fungal siderophores usually form hexadentate octahedral complexes with ferric iron and typically employ hydroxamates, α-hydroxycarboxylates and catechols as extremely effective Fe3+ ligands . Since bacterial growth requires iron, it is unsurprising that siderophores are effectively virulence factors (e.g. [174, 193–196]). While upwards of 500 microbial siderophores have been identified , with new ones still appearing (some via genomic analyses, e.g. ), and with the most common one in medical use, desferrioxamine or DFO, being such a bacterial product (see below), it is an astonishing fact that no human siderophore has been chemically identified, even though such activities were detected nearly 30 years ago [198, 199] (see also [200–205]). As noted by Kaplan , "a discovery that mammals produce siderophores would lead to an epochal change in the paradigm of mammalian iron homeostasis." To this end, some recent events have begun to change matters, and our overall knowledge of the regulation of iron metabolism, considerably.
Mammalian iron metabolism
The total body iron in an adult male is 3000 to 4000 mg and the daily iron requirement for erythropoiesis, the major 'sink', is about 20 mg . However, the loss of iron in a typical adult male is very small [208, 209] and can be met by absorbing just 1 – 2 mg of iron per day [210, 211]. The careful conservation and recycling of iron – mainly from degrading erythrocytes – is in fact essential because typical human diets contain just enough iron to replace the small losses, although when dietary iron is more abundant, absorption must be (and is) attenuated since higher levels than necessary lead to iron overload and many distressing sequelae contingent on the radical production described above.
A variety of aspects of mammalian iron metabolism have been reviewed in detail elsewhere (e.g. [134, 139, 195, 212–241]), including a series on 'iron imports' [242–248], and for our present purposes (Fig 2) mainly involves the intestinal (mainly duodenal) uptake of Fe(II) (produced from Fe(III) using a luminal ferrireductase) via a divalent metal ion transporter DMT1/DCT1/NRAMP [249, 250] and its subsequent binding as Fe(III) to transferrin (Tf). The intestinal uptake of haem (heme) occurs via the heme carrier protein-1 (HCP1)  and it is thereby internalized, while the iron in heme is liberated by heme oxygenase-1 (HO1) [252–254]. Haem is synthesised in many tissues, especially liver and erythroid cells . Vesicular routes of intestinal transfer may also occur [256, 257]. Low MW cytoplasmic chelators such as citrate can bind iron fairly weakly and thereby contribute to a labile iron pool (LIP) in the cytoplasm and especially the lysosomes and mitochondria (see [258–262]), while ferritin  too can bind cytoplasmic iron (via a chaperone ) and is seen as a good overall marker of iron status [265–267]. Iron(II) is subsequently exported through the basolateral membrane of the enterocyte by ferroportin-1 (FPN1) [268–270]. Ferroportin may also contribute to uptake in enterocytes . Fe(III) may then be produced by hephaestin (Hp)  before it is bound by transferrin (Tf), which is the main but not sole means of binding Fe(III) when it is transported through the circulation, with major iron storage taking place in the liver. Similar processes occur in the peripheral tissues, with significant transfer of iron from transferrin occurring via the transferrin receptor .
'Free' haem appears in the circulation (it may have a signalling role ) and elsewhere largely because of erythrocyte degradation, and it can also greatly amplify the cellular damage caused by ROSs , and its degradation pathway via haem oxygenase [276, 277] to biliverdin and then using biliverdin reductase to form bilirubin generates 'free' (and potentially redox-active) iron. It would appear, not least because biliverdin has powerful antioxidant properties, that haem oxygenase is more protective than damaging [253, 278–282], even though one of the products of its reaction is Fe that must eventually be liganded (or e.g. incorporated into ferritin). (Another product is the gas CO, that has been proposed as a measure of oxidative stress in the lung .)
All of the above obviously ignores both some important aspects of the speciation and liganding of iron, as well as the tissue distribution of the specific proteins involved – for which latter global information will shortly emerge  (http://www.proteinatlas.org/ and see later). It also ignores any discussion on the genetic regulation of iron metabolism (e.g. [285–288]), which is not our main focus.
However, one molecule in particular, hepcidin, has recently emerged as a 'master regulator' of regulation at the physiological level, and we describe some of these new developments.
In the liver and elsewhere, many aspects of iron metabolism are regulated by a recently discovered 25-amino acid polypeptide called hepcidin [207, 241, 245, 271, 289–327] that acts in part as a negative regulator of iron efflux  by causing the internalisation of ferroportin [329–333]. Hepcidin is produced, partly under the regulation of a receptor called hemojuvelin (e.g. ), via an 84-aa precursor called pre-pro-hepcidin and a 60 mer called pro-hepcidin [304, 335, 336] although the active agent is considered to be the 25 mer referred to above, and with the inactive precursors appearing not to be useful markers [337, 338].
Strikingly, anaemia and anoxia both suppress hepcidin production [245, 339, 340] (Fig 3), such that just while superoxide production is being enhanced by the anoxia there is more iron being absorbed from the intestine and effluxed into the circulation. In view of the inter-reactivity of superoxide and iron this could be anticipated to enhance free radical formation, leading to a positive feedback loop in which the problems are amplified: ischaemia/anoxia changes Fe(n) distribution leading to differential reactivity with the products of anoxia and thus further free radical production. However, hepcidin is overexpressed in inflammatory disease and is an early inflammatory marker [245, 341–345]. Its expression is positively controlled inter alia by SMAD4, and loss of hepatic SMAD4 is thus associated with dramatically decreased expression of hepcidin in liver and increased duodenal expression of a variety of genes involved in intestinal iron absorption, including Dcytb, DMT1 and ferroportin, leading to iron overload . STAT3 is another positive effector of hepcidin expression [347, 348], and ROSs inhibit this effect , thereby creating a link between ROSs and Fe metabolism. To understand the exact roles of hepcidin in iron metabolism, it is going to be especially important to understand where it is expressed; fortunately, such studies are beginning to emerge .
Overall there is a complex interplay between positive and negative regulation and the organismal distribution of iron caused by changes in hepcidin concentration , with in many cases the hypoxic response (decreased hepcidin) seeming to dominate that due to inflammation (increased hepcidin) even when iron levels are high [352, 353]. Specifically, lowered hepcidin causes hyperferraemia. Hepcidin is also activated by p53 , and may play a role in the degradation of atherosclerotic plaques . Another recently discovered protein that is crucially involved in human iron metabolism is NGAL or siderocalin, and while there is some evidence for their co-regulation , they have normally been studied separately.
NGAL (also known as lipocalin-2 or siderocalin)
Lipocalins  are a diverse group of ligand-binding proteins that share a conserved structure even in the absence of significant sequence conservation. This core structure includes an eight-stranded anti- parallel β barrel that defines a calyx, or cup-shaped structure, enclosing the ligand binding site.
NGAL – neutrophil gelatinase-associated lipocalin – is a 21 kDal glycoprotein first isolated by Kjeldsen and colleagues in 1993 . Synonyms include lipocalin 2, siderocalin, Lcn2, α2-microglobulin-related protein or neu-related lipocalin (in rats) [359, 360] and (in mice) 24p3 or uterocalin . Although lipocalins are well known to be involved in the sequestration and transport of a variety of ligands, the natural ligand of NGAL (as is the case with many lipocalins) was not initially known even in terms of its chemical class. This changed with the seminal paper of Goetz and colleagues  (and see ) who purified recombinant NGAL from a particular strain of E. coli and found that its structure contained a negatively charged ferric siderophore with a subnanomolar dissociation constant that it had extracted from its bacterial host, and that the apo form of this molecule could also act as a potent bacteriostatic agent by sequestering iron (see also [363–367]). A companion paper  showed that the iron-delivering activity was expressed in mammalian cells. The structure of NGAL is now known  and one of its interaction partners is a matrix metalloproteinase  to which it can presumably donate a metal ion and in the complex decrease its degradation .
The finding that NGAL was one of the most highly expressed proteins following ischaemia-reperfusion injury in kidney cells [372–374], and prognostic of kidney damage long before the more traditional marker creatinine was raised significantly, has led to considerable interest in this protein, especially as a marker of renal injury [375–389], and perhaps as a therapeutic . Devireddy and colleagues  identified a receptor that internalizes 24p3, and internalization of iron bound to 24p3 prevents apoptosis. In contrast, internalization of the apo form of 24p3 that does not contain iron led to cellular iron efflux and apoptosis via the proapoptotic protein Bim .