In the liver and elsewhere, many aspects of iron metabolism are regulated by a recently discovered 25-amino acid polypeptide called hepcidin 207241245271289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327 that acts in part as a negative regulator of iron efflux 328 by causing the internalisation of ferroportin 329330331332333. Hepcidin is produced, partly under the regulation of a receptor called hemojuvelin (e.g. 334), via an 84-aa precursor called pre-pro-hepcidin and a 60 mer called pro-hepcidin 304335336 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 337338.

Strikingly, anaemia and anoxia both suppress hepcidin production 245339340 (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 245341342343344345. 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 346. STAT3 is another positive effector of hepcidin expression 347348, and ROSs inhibit this effect 349, 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 350.

Overall there is a complex interplay between positive and negative regulation and the organismal distribution of iron caused by changes in hepcidin concentration 351, with in many cases the hypoxic response (decreased hepcidin) seeming to dominate that due to inflammation (increased hepcidin) even when iron levels are high 352353. Specifically, lowered hepcidin causes hyperferraemia. Hepcidin is also activated by p53 354, and may play a role in the degradation of atherosclerotic plaques 355. 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 356, they have normally been studied separately.

Lipocalins 357 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 358. Synonyms include lipocalin 2, siderocalin, Lcn2, α2-microglobulin-related protein or neu-related lipocalin (in rats) 359360 and (in mice) 24p3 or uterocalin 361. 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 362 (and see 206) 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 363364365366367). A companion paper 368 showed that the iron-delivering activity was expressed in mammalian cells. The structure of NGAL is now known 369 and one of its interaction partners is a matrix metalloproteinase 370 to which it can presumably donate a metal ion and in the complex decrease its degradation 371.

The finding that NGAL was one of the most highly expressed proteins following ischaemia-reperfusion injury in kidney cells 372373374, 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 375376377378379380381382383384385386387388389, and perhaps as a therapeutic 375. Devireddy and colleagues 390 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 391. In humans the megalin receptor can bind siderocalin (and its siderophore payload) and mediate its intracellular uptake 392. Oxidative stress can also induce its expression 393, and it is protective against it 394.

Exogenously administered NGAL also markedly upregulates heme oxygenase-1, a proven multifunctional protective agent in experimental Acute Kidney Injury (AKI) that is thought to work by limiting iron uptake, promoting intracellular iron release, enhancing production of antioxidants such as biliverdin and carbon monoxide, and inducing the cell cycle regulatory protein p21 279395396. Because of this multifaceted protective action, NGAL has emerged as a prime therapeutic target in ischaemic AKI 379.

Structural and direct binding studies have suggested that siderocalin tends (although not exclusively) to bind catecholate-type ligands, rather than hydroxamate- or carboxylate-based siderophores, at least when tested with microbially derived siderophores 362363365 (but cf. 369 for claims, disputed 360 and not now accepted, as to the binding of bacterially derived formyl peptides!). The role of NGAL, as a siderophore-binding agent, is thus consistent with the widespread recognition that iron-induced radical generation is intimately involved in a variety of renal and other diseases 397398. However, while it is certainly the case that siderocalin can reduce the virulence of bacteria when it binds the relevant bacterial siderophores 362363364365366367 and that bacteria can 'evade' this by synthesising siderophores that siderocalin cannot bind (e.g. 186187399400401), it is questionable whether the only role of siderocalin lies in fact in its antibacterial activity. Rather we would suggest that its main role is in sequestrating iron via a human siderophore to stop inappropriately liganded iron from producing damaging oxygen radicals. Consistent with this iron-liganding role for human biology is the fact that the tissue most highly expressing NGAL under normal conditions is bone marrow 360402, the site of erythropoiesis. The liganding can be extensive; as Goetz and colleagues 362 note, "During inflammation, concentrations of NGAL can increase to levels, with concentrations approaching 20–30 nM in the serum 403, presumably adequate to bind all available iron as ferric siderophore complexes".

Significant changes in NGAL expression have also been observed, for instance, during kinase-mediated signalling 404405, in cardiovascular disease 406407408409 and in cancer 410411412.

These findings on the kidney and the role of NGAL, together with the important knowledge that its chief ligand is probably an unknown human siderophore (Figs 2, 4), thus lead us to consider the role of this system (and unliganded iron generally) in a whole series of other diseases that all share many characteristics of oxidative stress and inflammation (see also 413). A similar thesis, albeit with comparatively little stress on iron, is the leitmotif of Finch's recent detailed monograph 163. The theme of these sections is thus to stress the fact that while the role of ROSs in general in such syndromes has been pointed up previously, that of iron as a major culprit has not so generally yet been stressed, notwithstanding that there is in fact a great deal of pertinent literature that we here highlight as the focus of this review.

Another important disease that shares many of the same properties (or at least sequelae) of renal impairment, and may have the same fundamental aetiology, is pre-eclampsia. This is the most significant cause of morbidity and mortality in pregnant women 414. The chief clinical manifestations at time of diagnosis are a raised blood pressure (hypertension) 415 and proteinuria, together with raised creatinine, consistent with the

reversible

combination

We note that it is quite common nevertheless for iron to be prescribed during pregnancy, especially during its latter stages 526527, and that this does of course lead to oxidative stress 528529.

Oxidative stress is caused both by the initial rate of production of superoxide and the rate of their conversion into OH^• radicals. The former can be induced by hypoxic conditions such as occur at high altitude, and one prediction, that is borne out 487530, is that PE should therefore be more prevalent at high altitude. Erythropoietin may be a marker for oxidative stress in pre-eclampsia 531.

Regarding the second stage, predictions include that PE should be more common in those suffering from diseases of iron metabolism. Although such mothers are of course less well a priori, this prediction is borne out for α-thalassemia 532533 although not, interestingly, for haemochromatosis 534. We note in this context that thalassaemia not only predisposes towards PE but is known in general to cause hepcidin to decrease and NGAL to increase 352353356535, with consequent and inevitable iron dysregulation.

Another prediction is then that hepcidin should be changed in pre-eclampsia. Although no serum measurements have been reported to date, it is of extreme interest that – while they took it to be an antimicrobial peptide rather than an iron regulator – a recent study by Knox and colleagues of placental gene expression in a mouse model of PE showed that hepcidin expression

increased

Finally, we note that NGAL is significantly implicated in pregnancy, and was even named uterocalin in mice to reflect its high expression in the uterus 361537538539. A very recent study 540 suggests that it may be a useful second trimester biomarker for pre-eclampsia.

Type 2 diabetes and insulin resistance are known complications of pregnancy (e.g. 541542543544545), and also predispose towards PE. In a similar vein, various types of pregnancy-related intrauterine growth restriction predispose towards diabetes in later life 546547, pointing up the progressive nature of these syndromes. Metabolic biomarkers for the one can thus be predictive of the other 548, consistent with a common cause. Certainly ROSs are known to play a substantive role in both insulin resistance 549550551552553554555556 and in a variety of diabetic sequelae 95557558559, and mitochondrial dysfunction may be an early step in this 560. Some anti-diabetic drugs, such as the 'glitazones' that are considered to act on Peroxisome Proliferator Activated Receptor (PPAR)γ, may also act by decreasing ROS production (e.g. 561562563564565), and are even active aganst cerebral ischaemia and stroke 566567568569. As with most if not all of the other diseases we review here, studies of pro-inflammatory markers (such as TNF-α, IL-1 and C-reactive protein 570) during the development of diabetes show its aetiology to be inflammatory in nature 553571572573574575576577578579580581582583584585586587588589590. Iron 'excess' is also a known feature of gestational diabetes 591592593, and is a clear risk factor for the disease even in 'normal' populations 594595596597598599600601602603, and indeed diabetes is a classical consequence of iron overloading as seen in hereditary haemochromatosis 604. Serum ferritin and body iron stores are strongly associated with diabetes 603605606607608609, including prospectively 610, while changes in visfatin are also intimately involved in changes of iron metabolism (with

pro

Non-transferrin-bound iron is also considerably elevated in type 2 diabetes 621, and this too is exacerbated by vitamin C. Iron metabolism is substantially deranged in type 2 diabetes and the metabolism of glucose (a reducing sugar) interacts significantly with iron metabolism 598. Iron is also strongly implicated in non-alcoholic steatohepatitis, considered an early marker of insulin resistance 622623624. Well-known diabetic complications include retinopathies, and it is noteworthy that elevated levels of ferritin can lead to cataract formation 625626.

Although some of its origins may be pre-natal 547, many of the features of these diseases are also seen in the (so-called) Metabolic Syndrome 627628629630631. Thus, serum ferritin is also related to insulin resistance 606632633 and iron levels are raised 624634635. Of course diabetes and the Metabolic Syndrome are also closely coupled, so it is reasonable that features observed in the one may be observed during the development of the other. The metabolic syndrome is also an independent indicator for chronic kidney disease 636 and may be related to liver steatosis 637. Metabolic disorders of this type too are closely intertwined with inflammation 575581587638, that is of course stimulated by ROSs whose generation is increased by high-fat diets 639. Thus, our role here is to point up the existence of a considerable body of more-than-circumstantial evidence that here too the progressive and damaging nature of these diseases may be caused, in part, by inappropriately chelated iron.

"As previously pointed out by Booth et al. 640, 100% of the increase in the prevalence of Type 2 diabetes and obesity in the United States during the latter half of the 20th century must be attributed to a changing environment interacting with genes, because 0% of the human genome has changed during this time period." 629

It is well known that there has been a staggering increase in the prevalence of obesity, diabetes, and especially type 2 diabetes, in the last 50 years or so, and that this increase is expected to continue (e.g. 641642643 and http://www.who.int/diabetes/). Equally, it is now well known that obesity, metabolic syndrome, diabetes and cardiovascular diseases are all more or less related to each other 643, and the question arises here as to whether dysfunctional iron metabolism might be a feature of each of them. In the case of obesity per se, however, we see no major evidence as yet for a causative role of deranged iron metabolism or chelation in causing obesity. Indeed, while they are related 644, what little evidence there is 645646 suggests that the converse may be true, i.e. that changes in iron metabolism might be consequent upon obesity (possibly via peroxide generation 639). Importantly, considerable evidence suggests that obesity and inflammation are significantly related 163486575581642647648649650651652653654655656657658659660661662663664, not least because adipocytes produce and release various adipokines including pro-inflammatory cytokines such as IL-6 and TNF-α 575649650665666667668669670671. It is likely that it is the

combination

As well its significance in pre-eclampsia (see above), hypertension is a well known risk factor for many cardiovascular and related disease (e.g. 677), and there is considerable evidence that its underlying cause is inflammatory in nature 678679680681682683684685686, is related to the metabolic syndrome and obesity (e.g. 648687688689690691), and may be mediated mainly via ROSs 692. There is evidence that some of its sequelae may be mediated via iron 693694.

It is well known that elevated iron stores can predispose to coronary artery disease and thence myocardial infarction. The 'iron hypothesis' of the benefits of some iron depletion due to menstruation was devised to account for the lowering of heart-disease risk in young women (that disappears in those post-menopause) and was proposed by Jerome Sullivan in 1981 695696697698 (and see also 699700). (In this sense, the lack of menstruation during pregnancy would predispose to a comparative abundance of iron, as is indeed found – see above.) It is of particular interest that the well-known adverse vascular effects of homocysteine (in inhibiting flow-mediated dilatation) are in fact iron-dependent 701702703, and that reducing homocysteine (e.g. by folate supplementation) in the absence of lowering iron has shown no clinical benefit to date 704, thereby suggestion iron mediation. By contrast, iron stores represent an established risk factor for cardiovascular disease 705.

Of course many factors such as lipid levels, stress, smoking and so are well-known risk factors for cardiovascular, coronary artery disease and related diseases. Indeed kidney disease is well established as a risk factor for cardiovascular disease 706707708 (and indeed stroke 709), all consistent with their having in part a common cause – we believe inflammation). Our purpose here, within the spirit of this review, is to indicate the evidence for the involvement of inappropriately chelated iron in cardiovascular diseases too. There is no doubt that the iron-mediated causal chain of ischaemia → (su)peroxide → OH^• radical formation occurs during the development of heart disease, especially during reperfusion injury 710711712713, and suitable iron chelators inhibit this 714715 (see also 716717, and for thalassaemia 718). Iron is also involved in the protection that can be produced by ischaemic preconditioning 719720. Erythropoietin, a hormone with multiple effects that may involve iron metabolism, is also protective 721722.

The sequelae of heart failure are complex, and involve a

chronic and continuing

It is next on the formation of atherosclerotic plaques that our attention is here focussed.

Atherosclerosis is a progressive

inflammatory disease

seventeen times greater

Statins, typically developed on the basis of their ability to inhibit the enzyme HMG-CoA redutase and thus decrease serum cholesterol, are well established to have benefits in terms of decreasing the adverse events of various types of cardiovascular disease 804, albeit that in many populations (e.g. 805806807) cholesterol alone is a poor predictor of cardiovascular disease, especially in the normal range. However, a known target of statins different from HMGCoA reductase is the β₂-integrin leukocyte function antigen-1 (LFA-1) 808809 and in this context, it is important to note that the clinical benefits of the statins are certainly

not

It has been pointed out that many measures of iron stress are inappropriate, since it is only the redox-active form of iron that is likely involved in oxidative stress. Serum ferritin is considered by some to be the most reliable marker of iron status in general 894, although it is not well correlated with iron distribution in the heart 895, for instance. What is clear, however, from the above is that deranged iron metabolism is intimately and causally involved in the formation of atherosclerotic lesions, and that appropriate iron chelation can help both to prevent and to reverse this.

Iron status is also closely involved in other chronic vascular diseases, and in the behaviour of wounds 896897898899.

Stroke is caused by ischaemia, leading to inflammation 900 and to the formation of ROS and other damaging free radicals 901 in the brain (which is high in metal ions 902), and is exacerbated by existing inflammation – see e.g. 903904. Thus, another prediction is that iron excess should also aggravate the sequelae of stroke, and that appropriate chelation or free radical trapping agents should mitigate these effects. These predictions are indeed borne out 138905906907908909910911912913914. It is also of considerable interest that plasma NGAL levels are increased in stroke 406; it is noteworthy that this can be seen in plasma despite the localised origin of the disease.

A variety of other studies have shown the beneficial treatments in stroke models of anti-inflammatory and antioxidant treatment, i.e. treatments that lower the amount of ROSs (e.g. 915916917918919920921922), as well as of preconditioning 923. Given its role in iron metabolism, it is of considerable interest that erythropoietin also seems to be very effective in protecting against brain ischaemia/reperfusion injury and stroke 924925926927928929930931932933934935936937938939, by a mechanism independent of erythropoiesis 940941942, and one that appears to involve anti-inflammatory activity 943.

Oxidative stress and inflammation are early events of neurodegenerative diseases 920944945946947948949950951952953954955956957 such as Alzheimer's disease (e.g. 958959960961962963964965966967968969970971972973), where plaque formation precedes neurodegeneration 974. Iron (and in some cases copper) is also

strongly

Indeed Thompson and colleagues comment 136 that "The underlying pathogenic event in oxidative stress is cellular iron mismanagement" and stress that "Multiple lines of evidence implicate redox-active transition metals, such as iron and copper, as mediators of oxidative stress and ROS production in neurodegenerative diseases". There is also ample evidence for its presence in the plaques characteristic of Alzheimer's disease 1004101310271088, just as in those of atherosclerosis (see above). Note too that iron can catalyse the oxidation of dopamine to a quinine form that can bind covalently to and then aggregate proteins 1089. Kostoff 1090 has used a very interesting literature-based discovery approach to highlight the role of oxidative stress in the development of Parkinson's disease.

Other papers highlight the role of iron in multiple sclerosis 89999110911092109310941095109610971098 and in prion diseases 96710991100. However, a particularly clear example of iron-mediated neurodegeneration is given by the sequelae consequent upon lesions in a protein known as frataxin involved in the disease Friedreich's ataxia (FA).

A number of repiratory chain components contain non-heme iron, and the question arises as to how they acquire it 1101. Frataxin is a mitochondrial iron chaperone protein 11021103110411051106110711081109, involved in the safe insertion of Fe(II) during the production of Fe-S centres in the mitochondrial respiratory chain 1110. As are some other aspects of iron metabolism 1111, it is highly conserved in eukaryotes from yeast to humans 11121113, a fact that made the unravelling of its function considerably easier 1114111511161117111811191120112111221123. Friedreich's ataxia (FA) is a neurodegenerative disorder that arises from a genetic deficit of frataxin activity, whether by a missense mutation or, much more commonly, via the addition of GAA trinucleotide repeats 11071124112511261127. As well as the neurodegeneration and measurable iron deposition, clinical symptoms include cardiac hypertrophy 1128 and (pre-)diabetes 1129, consistent with the general thesis described here that all are in part manifestations of iron dysregulation, and in which suitable chelation may be beneficial 11301131 (but cf. 1132).

ROSs are undoubtedly involved in FA 11331134, specifically via Fenton chemistry 11351136, since the attenuation of H₂O₂ production (but not of superoxide) 1110 ameliorates the disease 1137. The deficit in frataxin causes both an increase in ROS (H₂O₂) production via the mitochondrial electron transport deficiency 1138 as well as a dysregulation in iron metabolism, potentially a very damaging synergistic

combination

ALS is another progressive inflammatory 1142 disease in which motor neuron death causes irreversible wasting of skeletal muscles. It has largely defied efforts to uncover the genetic basis of any predisposition 1143, save for a very clear association with defects in a Cu/Zn superoxide dismutase 33114411451146 that can obviously lead to an increase in the steady-state levels of superoxide (and hence hydroxyl radical formation). There is also significant evidence for the involvement of iron 11471148. Drug therapies have to date shown rather limited benefits, and more in mouse models of Cu/Zn SOD deficiencies than in humans, though iron chelation therapy does not seem to have figured heavily, and it is recognised that combination therapies might offer better prognoses 1149.

Aging or senescence is defined as a decline in performance and fitness with advancing age 1150. Iron stores tend to increase with age 11511152115311541155, partly due to dietary reasons 1156 (and see 11571158), as does anaemia 11591160. So too does the expression of NGAL/Lcn2/siderocalin, a process that can be reversed by melatonin 1161. Mainstream theories of aging 1631651162116311641165116611671168116911701171117211731174117511761177117811791180 recognise the relationship between progressive inflammation, cellular damage and repair and the higher-level manifestations of the aging process, and ('the free radical theory of aging' 11631181) ROSs are of course strongly implicated as partial contributors to the aging process (e.g. 314398011631182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206). Needless to say, not least because of the low steady-state net rate of generation of the various ROSs 1207, few studies have managed to be very specific mechanistically 1208, but it should be clear that all ROSs are not created equal and we need here to concentrate mainly on the 'nasty' parts of ROS metabolism, and in particular on the hydroxyl radical as generated via poorly liganded iron and on peroxynitrite, and to have the greatest effects we need to inhibit both their generation and their reactivity (see Systems Biology section, below). The iron content of cells also increases as cells age normally 1209. As many diseases increase with age, probably via mechanisms highlighted herein, treating aging can thereby treat disease 162, and it is important to recognise that most 'diseases' are in fact consequences of aging (despite the considerably greater historical focus on the former).

One issue of aging is not that just it happens but that it can manifest in a series of essentially undesirable physiological changes, referred to as frailty 1210, in which ROSs have also been strongly implicated. Indeed, there are many parts of physiology and metabolism that lose functionality during aging (e.g. the cardiovascular system 1211 and of course cognitive function 1212), and iron metabolism is known to change considerably as humans age 9801213, with anaemia a typical accompaniment of aging 1214. The question then arises as to how much of this deranged iron metabolism is causal in accelerated aging, and this is not easy to state at this time. However, lowering iron does increase the lifespan of Drosophila 1215 and yeast 1105. At all events, the purpose of this rather brief section is to point out to researchers in aging, frailty and gerontology generally the relevance of inadequately controlled iron metabolism as a major part of ROS-induced injuries that may accelerate the aging process.

Although aging and longevity are not of course the same thing, and longevity is not a 'disease', studies of aging are often performed with the intention of improving our understanding of longevity 1216, and certainly longevity is linked to age-related disease 1217. However, the longevity of many organisms can be varied by the manipulation of any number of diseases or processes. This said, caloric or dietary restriction is a well-known contributor to longevity (indeed the only reliable one in pretty well all species 121812191220122112221223 – although possibly not H. sapiens 1180), and appears to act at the root of the processes involved 1224. Caloric restriction appears to be associated with a considerably lowered rate of production of ROSs and accrual of ROS-induced damage 12071225122612271228, and this is to be expected on general grounds in that a restriction of substrate supply will make the redox poise of mitochondria 1229 more oxiding and thereby minimise the amount of 1- and 2-electron reductions of O₂ to form peroxide and superoxide. It is therefore also of great interest that caloric restriction also benefits iron status 1230 and that it is this improved iron status that in part promotes longevity 1231. Caloric restriction has also been shown to effect a differential stress response between normal and tumour cells 1232, although this study did not look at relative iron status.

Antioxidants can also influence lifespan. Thus (rather high doses of) melatonin extended the lifespan (and stress resistance) of Drosophila 1191123312341235 while that of Caenorhadbitis elegans could be extended by mimics 1236 of SOD and catalase 1187, and by a variety of antioxidant and other pharmacological agents 1237.

As an example, let us consider C. elegans. Mutants with a decreased activity of the insulin-like growth factor signalling pathway (e.g. daf2 mutants that have a greater amount of the DAF16 FOXO-like transcription factor) can live for nearly twice as long as wild types 12381239 and produce more catalase, superoxide dismutase (sod-3) 1240 and glutathione-S-transferase.

Overall, it is the potent combination of oxidative stress, already leading to damaging peroxides and radicals, and its catalysis and further reactions caused by inappropriately chelated iron, that causes a 'double whammy'. Indeed, iron, copper and H₂O₂ have been referred to as the 'toxic triad' 1032. While there is comparatively little that we can do about the production of superoxide and peroxide, we can (by pharmacological or dietary means) try and improve the speciation of iron ions.

One disease whose aetiology is well known to be bound up with ROSs is rheumatoid arthritis (RA) 36124112421243124412451246. What is known of the role of iron metabolism? Generally an overall low iron status – anaemia – is a characteristic of rheumatoid arthritis 12471248124912501251, whereas by contrast iron is elevated in the synovial fluid of arthritic joints 1252125312541255. This suggests a significant derangement of iron metabolism in RA as well, and a mechanism 1256125712581259 in which elevated superoxide liberates free iron from ferritin in synovial fluid (and elsewhere 1260), thereby catalysing further the damaging production of hydroxyl radicals. This autocatalytic process (Fig 6) is, even in principle, especially destructive (and may account for species differences in sensitivity to iron loading 1261). Note that erythrocytes when oxidized can also release free iron 524. Natural antioxidants such as vitamin E are also lowered 1262. There is some evidence that appropriate iron chelators can ameliorate the symptoms of RA 1263, though membrane-impermeant chelators such as desferroxamine cannot 1264.

One interesting feature of RA is that in 75% of women it is strongly ameliorated during pregnancy 126512661267; although the multifactorial nature of this observation has made a mechanistic interpretation difficult, from everything that we have seen so far it would be surprising if changes in iron metabolism were not strongly involved.

An interesting related feature is the 'restless legs syndrome' 126812691270, that is often associated with iron deficiency and pregnancy. Serum transferrin receptor seems to be a rather sensitive measure of this iron deficiency 27312711272. There are relationships with other syndromes that we discuss here, such as cardiovascular disease 1273, but in the present review it is of especial interest that a dysregulation of iron metabolism appears to play a significant role 1274

Lupus, or Systemic Lupus Erythematosus (SLE) 12751276 describes a syndrome, somewhat related to arthritis and rheumatism, of broadly auto-immunological or inflammatory origin, and with a large variety of manifestations (e.g. 127712781279128012811282128312841285), but characterised in particular by fatigue 1286, often as a result of anaemia. This of course points to a certain level of iron dysregulation (of any number of causes), and there is certain some evidence for this 12741287. Thus, while anaemia is a feature of the disease, serum ferritin may be raised in SLE 1288, and some of the usual lipid markers of oxidative stress (that can be a result of hydroxyl radical production catalysed by poorly liganded iron) are also present 1289.

There is also a very interesting linkage between SLE and vitamin D metabolism 1290, something that has also come up in relation to the statins and atherosclerosis (8321291, but cf. 1292), and indeed it there is evidence that statins may be of benefit in the treatment and even reversal of lupus 12931294. Indeed, from a much more general point of view, there is precedent for these kinds of linkages being used to uncover unknown mediators in disease states that may be worth pursuing (e.g. 6192012951296).

Asthma is a well-known inflammatory disease, and has been linked with ROS generation 1297 catalysed by iron 276129812991300.

By definition, IBD such as Crohn's disease and ulcerative colitis are inflammatory diseases, and while the inflammation and ROS production are well established here, their origins are somewhat uncertain 1301130213031304. They are frequently accompanied by anaemia, implying a derangement in iron absorption and/or metabolism 13051306, and very probably absorption 13071308. The anaemia may be monitored by ferritin and transferrin receptor levels, and its correction is possible by iron supplementation plus erythropoietin 1309131013111312131313141315. One may suppose that some of the issues here relate to iron speciation, that is usually not measured in these studies.

Age-related macular degeneration (AMD) 1316 is now the leading cause of blindness and visual disability in the elderly in developed countries 1317131813191320. Many components of atherosclerotic plaques have also been demonstrated in drusen 1321, a characteristic of AMD and, as here, it is reasonable to propose a common mechanism of pathogenesis between AMD and atherosclerosis. Retinal iron levels increase with age 1322, iron is significantly implicated in AMD 1323132413251326132713281329, and iron chelation may help to reverse the process 1330. Dietary antioxidants are also protective 1331. The source of the iron appears to be excess angiogenesis and leakage from blood vessels catalysed by VEGF, and a PEGylated aptamer 133213331334 against VEGF (pegaptanib) or a monoclonal antibody (ranibizumab) have shown significant promise in the treatment of macular degeneration 133513361337133813391340. Plausibly a combination therapy with one of these plus a suitable iron chelator might be even more effective.

Psoriasis is an anflammatory disease in which the production of free radicals and ROSs are strongly implicated 134113421343. Here too there is clear evidence for the involvement of a deranged iron metabolism 134313441345. Early attempts at therapy with a series of unusual iron chelators (that unfortunately had side effects) 1346 do not seem to have been followed up.

Gout is another important inflammatory disease, characterised by the accumulation of uric acid 1347. There is considerable evidence that this too is a disease of iron overload, and that uric acid accumulation – as both an antioxidant and an iron chelator 1348 – is in response to the iron overload 13491350 and with highly beneficial remission of gouty symptoms occurring on depletion of iron by phlebotomy 1351.

It is known that with chronic excess, either iron or alcohol alone may individually injure the liver and other organs, and that in combination, each exaggerates the adverse effects of the other. Specifically, in alcoholic liver disease, both iron and alcohol contribute to the production of hepatic fibrosis 1352135313541355135613571358. Iron overload is well known to lead to hepatotoxicity 1359136013611362 and liver cancer 13631364, and lowering or chelating it is protective 13651366. Hepcidin may be involved here 1367.

Chronic obstructive pulmonary disease (COPD) is a progressive and chronic disease which is characterised by an apparently inexorable decline in respiratory function, exercise capacity, and health status. It is also characterised by periods in which the symptoms are considerably exacerbated 1368136913701371. Such an acute exacerbation of COPD (AECOPD) is defined 1372 as "a sustained worsening of the patient's condition from the stable state and beyond normal day to day variations, that is acute in onset and necessitates a change in regular medication in a patient with underlying COPD". Smoking is a major source of free radicals (and indeed metals 1373), and is a major cause of COPD. Consequently, there is considerable evidence for the evidence for the involvement of inflammation and ROSs in both the 'stable' and 'exacerbated' stages 13741375137613771378137913801381.

Needless to say, there is also considerable evidence for the significance of exposure to iron 1382 (as well as exposure to other toxic metals 1383) in the development of COPD and other lung diseases 1384. Haem oxygenase also appears to be a significant culprit 280, and lung (lavage) iron is increased 1385 while transferrin levels can be considerably lower 1386.

Other lung diseases in which ROS and iron have been implicated include Adult Respiratory Distress Syndrome 1387138813891390 and cystic fibrosis 1391.

Tobacco smoke contains many unpleasant and carcinogenic compounds, and that tobacco smoking is a leading cause of carcinoma of the lung and indeed of other organs has become well known since the pioneering epidemiological studies of Doll, Peto and colleagues (e.g. 1392139313941395). Our purpose here is to point out that many of the particles associated with smoking (and also ingested from other sources) are heavily laden with micro-particulate iron, which, as a major catalyst of hydroxyl radical production, undoubtedly is a substantial contributor as well (see e.g. 1373138413961397139813991400).

In addition to the issues of smoking, the development of cancer can certainly contain an inflammatory component (e.g. 1401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425), and indeed the long-term use of prophylactic anti-inflammatory aspirin lowers colon cancer incidence by 40% (age-adjusted relative risk = 0.6) 1426 (though note that this may have other side-effects 1427). (The well-known association between infectious agents such as H. pylori 142814291430 and e.g. bowel cancer is probably initiated by chronic inflammation.) Given that cells require iron, restricting its supply can also limit the growth of cells, including tumour cells 143114321433143414351436143714381439144014411442. Conversely the iron carrier NGAL is overexpressed in tumours 4104111443, a process mediated via NF-κB 1444 (and see later). Further, the roles of iron, not least in the mutagenic effects of metal-catalysed Fenton chemistry, are also of significance in promoting oncogenesis 14451446144714481449145014511452145314541455145614571458145914601461. Iron chelators 14381462 are thus a doubly attractive component of anti-cancer therapeutics. The mutagenic, carcinogenic and disease-causing actions of asbestos and related fibres may also be due in significant measure to the ability of the Fe(n) that they contain to catalyse hydroxyl radical production 1463146414651466146714681469147014711472147314741475, while there seem to be complex relations between the likelihood of apoptosis and the differential concentrations of superoxide and H₂O₂ 14761477. Overall, it is becoming increasingly widely recognised that anti-inflammatory agents have a role to play in the treatment of cancers; we would suggest that iron chelation may be a useful component of such treatments.

Just as do tumour cells, the malarial parasite Plasmodium falciparum requires considerable iron for growth, and there is evidence that lowering the amount of available iron provides a promising route to antimalarials (e.g. 14781479148014811482148314841485148614871488148914901491149214931494, but cf. 1495). Note in this context that iron-catalysed radical formation is also significantly involved in the antimalarial (i.e. cytotoxic) mode of action of artemisinin 1494149614971498149915001501150215031504, and this reaction is in fact inhibited by iron chelators 1505 such that

a combined artemisinin-chelator therapy would be contraindicated

Lower down the evolutionary scale, and as presaged earlier in the section on bacterial siderophores, microbes require iron for growth, its presence may be limiting even at the scale of global CO₂ fixation 1506150715081509, its excess can in some circumstances 1510 correlate with infectivity or virulence (see above and e.g. 1801931493151115121513151415151516151715181519152015211522152315241525152615271528152915301531), and its chelation in a form not available to bacteria offers a route to at least a bacteriostatic kind of antibiotic or to novel therapies based on the lowering of iron available to microbes by using hepcidin 1532 or NGAL 1533. Iron chelators are also inhibitory to trypanosomes 15341535, and changes in iron metabolism are also associated with viral infections 1536.

It is well known that one consequence of bacterial infection (sepsis) can be septic shock, that this can be mimicked by the Gram-negative bacterial outer membrane component LPS (lipopolysaccharide), and that in the worst cases this leads via multiple organ failure to death. However, whatever LPS does it is quite independent of the present of viable (i.e. growing or culturable – see 15371538) bacteria as the same phenomena leading to multiple organ failure (MOF) are seen without infection 15391540. Consequently, the recognition of a series of symptoms contingent on this initial inflammatory response has led to the development of the idea of a Systemic Inflammatory Response Syndrome (SIRS) 15411542154315441545154615471548154915501551 that leads to the MOF, both via apoptosis 1552 and necrosis 15531554. There is by now little doubt that these phenomena too are associated with the hyperproduction of ROSs 154615511555155615571558155915601561156215631564156515661567156815691570. Circulating free iron is raised in sepsis and related conditions 1400157115721573. Direct assays of oxidant induced cell death indicate that most 'free' iron is concentrated in lysosomes 1574157515761577, that its decompartmentation is substantially involved 1570, and that its chelation can thus prevent cell death 1578157915801581.

Many circulating inflammatory factors have been identified as important in the development of septic shock, including cytokines such as Tissue Necrosis Factor (TNF) 1582, and cellular responses via the Toll-Like Receptor are clearly involved in this process 15831584. However, we would argue that since antibodies against TNF do not inhibit the sequelae of septic shock such as multiple organ failure, the truly damaging agents are caused elsewhere and are likely to involve the iron-mediated production of damaging hydroxyl radicals (see also 1563).

In this regard, it is especially interesting that the antioxidant melatonin is particularly effective in preventing septic shock 158515861587, and a variety of suitable antioxidants have shown potential here 15731588158915901591, notably in combination with iron chelators 15921593 (and see also 15941595). As with quite a number of the indications given above, a further link with Fe metabolism is seen in the protective effects of erythropoietin 15961597.

A number of reviews (e.g. 143714381672167316741675167616771678) cover aspects of iron chelators that have had or may have utility clinically.

Whitnall and Richardson 1062 list a number of useful features of an experimentally (and clinically) useful iron chelator. Thus, "A compound suitable for the treatment of neurodegenerative disease should possess a number of qualities, namely (1) strong affinity for FeIII, (2) low molecular weight, (3) lipophilicity high enough to accommodate permeation of cell membranes and the BBB, (4) oral activity, and (5) minimal toxicity 1062. Also, partly because there are few trivalent ions other than Fe(III) that the body actually needs, the major synthetic focus has been on the design of FeIII-selective chelators which feature "hard" oxygen donor atoms. Additionally, under aerobic conditions, high-affinity FeIII chelators will tend to chelate FeII to facilitate autoxidation, such that high-affinity FeIII-selective compounds will beneficially bind both FeIII and FeII under most physiological conditions" 1062. (Note that liophilicity per se may not be relevant, as drugs require carriers to cross membranes 18, and promiscuity and off-target effects increase with lipophilicity 1679.)

Desferrioxamines are nonpeptide hydroxamate siderophores composed of alternating dicarboxylic acid and diamine units. linked by amide bonds. They are produced by many Streptomyces species 1680. Desferrioxamine B is a linear (acyclic) substance produced (industrially) by the actinobacterium Streptomyces pilosus 1681, and is widely used as an iron chelator for the prevention and treatment of the effects of iron overload. It is commercially available as desferal (desferrioxamine methane sulphonate), also known as deferoxamine in the USA. It has been very effective in the treatment of a number of diseases, leading to the view that such molecules should have considerable therapeutic potential. A significant disadvantage of DFO is that it does not seem to cross the intestine intact (despite the rather catholic substrate specificity of intestinal peptide transporters 1682168316841685) and must therefore be given intravenously or subcutaneously. By contrast, another chelator known as Deferriprone or L1 does appear to cross cell membranes, but it is only bidentate.

Those with approval for clinical use are few in number and we deal with them first. Table 1 compares them with the 'ideal' properties of a clinically useful iron chelator, while Fig 7 gives the structure of the three most common, viz. desferal (deferoxamine), ferriprox (L1 or deferiprone) and exjade (ICL670 or deferasirox). (Dexrazoxane, a hexadentate chelator, is also marketed 1686.)

Desferal (deferoxamine) is the most used chelator for historical reasons. It is hexadentate but is not orally bioavailable. Ferriprox (deferiprone) is a bidentate ligand (1,2-dimethyl, 3-hydroxypyridin-4-one). It is orally bioavailable although comparatively high doses are required, and it postdates desferal. "Deferiprone has high affinity for iron and interacts with almost all the iron pools at the molecular, cellular, tissue and organ levels. Doses of 50–120 mg/kg/day appear to be effective in bringing patients to negative iron balance" 1687. It can have somewhat better properties than desferal 1688. Finally, Exjade (ICL670) (deferasirox) 16891690169116921693169416951696169716981699170017011702 is the most recent chelator approved for clinical use, and is tridentate. It is orally active, and there is a large bibliography at http://www.exjade.com/utils/bibliography.jsp?disclaimer=Y. The recommended initial daily dose of Exjade is 20 mg/kg body weight.

It is clear from Table 1 that in the time evolution from deferoxamine through deferiprone to deferasirox there has been a noticeable improvement in the general properties of iron chelators, although there are few published data on the quantitative structure-activity relationships of candidate molecules that might allow one to design future ones rationally. What is certainly clear is that there is a trade-off in properties, and that appropriate chelators will keep iron levels intermediate, i.e. not too low and not too high (a 'Goldilocks' strategy, if you will), and that hexadentate molecules may correspondingly be too tightly binding and strip iron from important molecules that need it. What is particularly important, as well as a good plasma half-life, is the ability to cross cell membranes, as this is necessary both for oral administration and for ensuring that the chelator in question actually accesses the intracellular 'free' iron pools of interest. Which carriers are used for this in humans in vivo is presently uncertain 181703.

The high investment of time, money and intellectual activity necessary to get a drug approved clinically has led to a number of strategies to exploit those that already have been approved and are thus considered 'safe'. One such strategy is the combination therapy of approved drugs that can yet serve for novel indications (e.g. 17041705170617071708). Another strategy is to look for antioxidant or iron-binding chemical motifs in drugs that have already been approved for other purposes 1709 (or to measure such properties directly).

Clioquinol (CQ) 1062167417101711 (Fig 7) is one existing (anti-parasitic) drug that has been proposed for use as an iron chelator, as it contains the known iron-chelating 8-hydroxyquinoline moiety. It has indeed enjoyed some success in this role. However, clioquinol toxicity has been reported if it is used over an extended period 1712 and this may be due to the formation of a Zn-clioquinol chelate 1713.

A particular attraction of such existing drugs is that they are likely to have favourable pharmacokinetics and pharmacodynamics, and in particular are likely to be cell-permeable. Note that despite a widespread assumption that lipophilicity or log P is sufficient to account for drug distribution this is not in fact the case, as there are hundreds of natural transporters that drugs can use (e.g. 181714). For instance, the iron-chelating 8-hydroxy quinoline motif contained in molecules such as clioquinol is also present in the tryptophan catabolite xanthurenic acid (Fig 7), and it is likely that transmembrane transport of the synthetic drug molecule occurs via natural carriers whose 'normal' role is to transport endogenous but structurally related molecules 181703.

Given the importance of the field, many academic investigators have sought to develop their own iron chelators that might exhibit the desirable properties listed above. One class of molecule includes isonicotinylhydrazones. Thus, pyridoxal isonicotinyl hydrazone (PIH) 1672171517161717171817191720 is a promising molecule (also proposed in anti-cancer therapy), although it is hydrolysed both in vivo and in vitro 1721. Other analogues include salicylaldehyde (SIH) 1722 and 2-hydroxy-1-napthylaldehyde (NIH) isonicotinyl hydrazones. PIH was disclosed before being patented, and is thus seen as having no pharmaceutical (company) interest. Various other derivatives are therefore being considered 10351462, including pyrazinylketone isonicotinoyl hydrazones 1723.

A variety of 8-hydroxyquinolines (8HQs) 1724 have been considered, although as with other bidentate and tridentate ligands that cannot necessarily effect complete liganding of iron there is always a danger that inadequate concentrations might be pro-oxidant (e.g. 17251726). One molecule, VK-28, combines various pertinent moieties and has shown some promise in the treatment of neurological disorders 1727172817291730. This strategy of combining drug elements that can hit multiple targets ('polypharmacology' 8071731173217331734) has much to commend it, including on theoretical grounds, and we discuss these in the section on systems biology below. Another 8HQ that has elicited interest is O-trensox 143117351736173717381739174017411742.

Other ligands or motifs that might be considered include di-2-pyridylketone-4,4,-dimethyl-3-thiosemicarbazone (Dp44mT), that has been shown to be effective against tumours 1743, 2,2'-dipyridyl, 1,10-phenanthroline 17441745, 2-benzoylpyridine thiosemicarbazones 1746 and thiohydrazones 1747. HBED (Fig 7) (N,N'-bis-(2-hydroxybenzyl)ethylenediamine-N,N'-diacetic acid) forms a 1:1 complex with Fe(III) but is probably only tetradentate 1748. It seems not to be very orally active 1749 but may be more effective than is DFO 175017511752. Poly-hydroxylated 1,4-naphthoquinones occur as sea urchin pigments and have shown protective effects 1753.

Continuing the theme of polypharmacology, R-(α)-lipoic acid 1754175517561757 is also an antioxidant, that may in addition act by stimulating other anti-oxidant pathways 1758. Finally, one interesting area is that of prochelators (e.g. 1759) in which the oxidant itself triggers the formation of a chelator able to inhibit the Fenton reaction.

There is also a considerable and positive role for nutrients in terms of their chelation of iron. Indeed, polyphenolic compounds, many of which have known health benefits 1804180518061807180818091810181118121813, are widely used as food antioxidants 18141815. There is of course considerable epidemiological evidence for the benefits of consuming fruit and vegetables that are likely to contain such antioxidants (e.g. 1816181718181819), and – although possibly a minimum – this has been popularised as the 'five a day' message (e.g. http://www.fruitsandveggiesmatter.gov/ and http://www.5aday.nhs.uk/). Even though elements of the 'Mediterranean' diet that are considered to be beneficial are usually assumed to be so on the basis of their antioxidant capabilities (but cf. 1820), many of the polyphenolic compounds (e.g. flavones, isoflavones, stilbenes, flavanones, catechins (flavan-3-ols), chalcones, tannins and anthocyanidins) 18211822182318241825182618271828 so implicated may also act to chelate iron as well 1073182918301831183218331834183518361837183818391840184118421843. This is reasonable given that many of these polyphenols and flavonoid compounds 18211844184518461847184818491850185118521853 have groups such as the catechol moiety that are part of the known iron-binding elements of microbial siderophores. Examples include flavones such as quercetin 9141813182918541855185618571858185918601861186218631864, rutin 18291857185818651866, baicalin 18601867, curcumin 181318681869187018711872, kolaviron 1873, flavonol 1874, floranol 1875, xanthones such as mangiferin 1876187718781879, morin 1876, catechins 107318071838185418801881 and theaflavins 1882, as well as procyanidins 18351883 and melatonin 16281884188518861887. However, the celebrated (trans-)-resveratrol molecule 188818891890189118921893189418951896189718981899190019011902 may act mainly via other pathways.

A considerable number of studies with

non-purified

We have above adduced considerable evidence that decreasing the amount of hydroxyl radical by any means is valuable, whether by removing initially generated ROSs such as superoxide and peroxide or by chelating poorly liganded iron in a way that stops these ROSs forming the hydroxyl radical. While pro-inflammatory cytokines can themselves increase ROS production and modulate the activities of signaling pathways such as NF-κB and p38, there are also anti-inflammatory cytokines. A particularly interesting example is that of erythropoietin (also discussed above as being protective in a number of iron-mediated diseases).

Erythropoietin was originally recognized via its role in erythropoiesis 214521462147 (hence its name, of course), but it has become evident that it has many other roles, and in particular it is observed phenomenologically that erythropoietin (and non-erthyropoetic derivatives) is protective in a number of inflammatory conditions that accompany many diseases such as those listed above 94094221482149215021512152215321542155. These included cardiovascular disease 72172221562157215821592160216121622163216421652166216721682169217021712172217321742175, stroke and other related neurological diseases 924925926928929933937938215521762177217821792180218121822183218421852186218721882189219021912192219321942195219621972198219922002201220222032204, diabetic neuropathy 2205, kidney injury 21732206220722082209221022112212, intestinal injury 2213 and shock (both septic and non-septic) 159615972214.

The question then arises as to how it is doing this mechanistically, and the proximate answer is that it (and other anti-inflammatory agents, e.g. 180822152216) seem to act via many of the same signalling pathways as do inflammatory agents 94321502217221822192220222122222223222422252226. There is evidence that it can help maintain superoxide dismutase activity 22142227, invoke haem oxygenase 2228, and in particular – from the perspective of this review – that it may remove poorly liganded iron 2229 and interact with hydroxyl radical directly 2230223122322233.

It is notable that appropriate levels of erythropoietin appear not only to be efficacious but to be safe, even in pregnancy 22342235223622372238223922402241. Erythropoietin may itself be a marker of hypoxia and oxidative stress in pregnancy 5312242224322442245, consistent with a view that the body is attempting to deal with these problems by creating anti-inflammatory cytokines.

What has emerged in recent years is a recognition that the structure (i.e. topology) of the modules of metabolic and signalling networks, somewhat independent of the individual activities of their components, can have a profound controlling influence on their behaviour (e.g. 21072134213522612262). Classically, negative feedback structures are considered to confer stability, while positive feedbacks tend to have the opposite effect. However, negative feedbacks with delay loops can cause oscillations 20222109 while some kinds of positive feedback loops can confer stability 22622263. However, there is no doubt that structures in which a damaging agent causes the production of a second damaging agent that itself catalyses the production of the first or a separate damaging agent can exhibit a runaway kind of damage. This is

exactly

autocatalytic cycles