Exploration of the medical periodic table: towards new targets†

Metallodrugs offer potential for unique mechanisms of drug action based on the choice of the metal, its oxidation state, the types and number of coordinated ligands and the coordination geometry. We discuss recent progress in identifying new target sites and elucidating the mechanisms of action of anticancer, anti-bacterial, anti-viral, anti-parasitic, anti-inflammatory, and anti-neurodegenerative agents, as well as in the design of metal-based diagnostic agents. Progress in identifying and defining target sites has been accelerated recently by advances in proteomics, genomics and metal speciation analysis. Examples of metal compounds and chelating agents (enzyme inhibitors) currently in clinical use, clinical trials or preclinical development are highlighted.

Introduction A significant number of clinical trials now involve metal compounds, metal-chelating agents (metalloenzyme inhibitors) or other agents which interfere with metabolic pathways for metals, both for therapy and for diagnosis. Some of the areas of current clinical interest are listed in Table 1.
These areas have been stimulated by recent successes with platinum anticancer drugs (used as a component of nearly 50% of all cancer treatments), with gadolinium complexes as MRI contrast agents (about 20 million doses administered per year) and 99m-technetium radiopharmaceuticals for g-ray imaging (used in about 20 million radiodiagnostic procedures each year). However, many other diseases and conditions are of current interest including neurodegeneration, microbial and parasitic (and other neglected tropical diseases) infections and inflammation. [1][2][3][4] About 24 elements are essential for man, Fig. 1. In principle, we should consider the roles of all essential elements in therapy (control of homeostasis), but this is currently hampered in many cases (e.g. Si, V, Ni, Sn) because their natural biochemistry is poorly understood and we do not know how to diagnose conditions associated with their metabolism. For some elements it is already possible to use genomics to understand medical conditions related to them (e.g. Fe, Cu, Zn), but in other cases it is not clear how, or if, genomes code for these elements. Genomes do not code for the elements themselves, but for particular chemical forms (e.g. vitamin B 12 for cobalt). Genomic codes for metals are usually codes for proteins, proteins which are highly, but not totally, selective for particular metal ions. As is evident from the above examples of Pt, Gd and Tc, inorganic medicinal chemistry can make use of non-essential as well as essential elements for the design of drugs and diagnostic agents, Fig. 1. 4 For many centuries the use of metallodrugs has been driven by empiricism. Whilst random screening is still a useful weapon in drug discovery, these days it should be guided largely by rational design. This implies that target sites (usually gene products -proteins) should be identified and verified early in the design process. Immediately this presents challenges for medicinal metal coordination chemistry: either the synthesised complexes need to be inert to redox and ligand substitution reactions, or these reactions need to be controlled under biological conditions en route to the target. In turn, metal-and/or ligand-centred redox reactions provide a unique platform for drug design with concepts quite distinct from those for purely organic drugs.
In this article we highlight some areas of current interest in metallodrugs. We focus especially on their interactions with (proposed) target sites, including DNA and proteins. Despite the fact that successful clinical platinum anticancer drugs have DNA as their major target, DNA is nowadays not considered to be a favoured target site. This is partly because DNA is also likely to be attacked in normal healthy cells as well as in cancer cells. However, downstream processing of platinated DNA can differ in normal and cancer cells so differential cytotoxicity can still be achieved. Moreover, it is becoming evident that anticancer drugs that target a single protein or enzyme are not always successful, since cells readily become resistant to such drugs and utilise alternative metabolic pathways. Metallo-anticancer drugs have the potential advantage of attacking several sites (multi-targeting), which can be an highly effective strategy. It is important to identify these target sites and elucidate the molecular mechanisms of action. As we shall see, a range of new target sites are being proposed for metallodrugs but there is still much progress to be made on target-site validation. Tailoring the design of metallodrugs to treat specific diseases and conditions is likely to be a major part of future personalised medicine, which will include genomic and proteomic profiling of individuals.  [5][6][7][8][9][10][11][12][13][14][15] More than 4000 Pt compounds have now been tested as potential anticancer drugs, with the worldwide approval for clinical use of carboplatin (2,1989) and oxaliplatin (3,2002), and of three others -nedaplatin (4), lobaplatin (5) and heptaplatin (6) (Chart 1) for clinical use only in Japan, China and South Korea, respectively. The ability to form DNA adducts is important for the activity of all of these drugs. Table 2 lists the X-ray structures of cis-diam(m)ine Pt-DNA adducts registered in the Protein Data Bank by January 2013. Dicycloplatin (7) is a new complex developed at the University of Beijing by Yang et al. that has passed through Chinese clinical phase I trials. This compound is composed of one molecule of carboplatin and one molecule of cyclobutane-1,1-dicarboxylic acid interacting via hydrogen bonds (see Chart 1). 16 The mechanism by which the presence of free ligand affects the activity of this platinum drug is not clear.
Compounds based on cis-diam(m)ine Pt II structures tend to produce very similar types of adduct on target DNA -as exemplified in Table 2. It is therefore not surprising that they induce similar biological consequences, although protein recognition of the DNA adducts is dependent on the ligands. This consideration led to the hypothesis that development of Pt compounds structurally different from cisplatin which form different types of DNA-Pt adducts, may lead to a different spectrum of biological activity, perhaps complementary to cisplatin. 36 In the late 1980s, Farrell began a fruitful investigation of linear multi-platinum complexes, where the platinum centres are separated by aliphatic chains, and the chlorido ligands are cis or trans at the extremities of the complex. [37][38][39][40] These poly-platinum complexes are generally highly positivelycharged. This leads to a stronger initial electrostatic recognition of DNA in comparison with monomeric species. Another critical feature of these multinuclear complexes is flexibility. The potent multi-platinum complex [(trans-PtCl(NH 3 ) 2 ) 2 -{m-trans-Pt(NH 3 ) 2 (NH 2 (CH 2 ) 6 NH 2 ) 2 }] 4+ , BBR3464 (8, Chart 2), 41 completed a phase I trial, 42 but failed phase II due to instability in blood. [43][44][45][46][47] Fig. 1 A medical periodic table: essential elements for man (symbols in white font); medical radioisotopes (green fill); elements currently used in therapy (blue fill) or diagnosis (orange fill). The entries (limited to 2 fill colours, illustrative and not comprehensive) are mainly restricted to elements/compounds which are clinically approved or on current clinical trials (e.g. as listed on http://www.clinicaltrials.gov/). Some entries for implants are included (e.g. Ti, Ta). The basis for some of the entries is given in Table S1 (ESI †).
Chart 1 Molecular structures of some platinum anticancer drugs that are approved or undergoing clinical trials.
It seems to be the only Pt compound not based on the cisplatin chemotype to have entered human clinical trials. 48 Hydrolysis of BBR3464 and DNA adduct formation are well documented. 49,50 In general polynuclear platinum compounds bind rapidly to DNA (t 1/2 ca. 40 min). 47 The most relevant feature of the special BBR3464-DNA binding is the lack of severe DNA distortions such as a kink, or significant unwinding of the helix, which are characteristics of DNA adducts of mononuclear platinum complexes. One of the direct consequences of this mild Pt-induced DNA conformational change is that these adducts are poor substrates for recognition by proteins, such as those containing the HMG domain, which binds to rigidly-bent DNA, as induced by cisplatin. 51 Overviews of multi-nuclear platinum drugs 52,53 and DNA binding to polynuclear platinum complexes 54 have been published.
Low-spin octahedral 5d 6 Pt IV complexes have potential advantages as anticancer drugs because they are more inert to substitution reactions than square-planar Pt II complexes and so are expected to undergo fewer side reactions en route to the tumour. Also they often have higher aqueous solubility, a feature exploited long ago by Tobe et al. who synthesised iproplatin (CHIP, JM9, cis,trans,cis-[PtCl 2 (OH) 2 (isopropylamine) 2 ]). 55 Iproplatin entered phase I and II clinical trials, 56,57 and even phase III, 58 but ultimately was found to be less active than cisplatin and so not registered for clinical use. 59 Tetraplatin (ormaplatin) [PtCl 4 (D,L-cyclohexane-1,2-diamine)], also showed promise in preclinical studies but caused severe neurotoxicity in treated patients, and trials were subsequently abandoned at phase I. 60 The unpredictability of the rate, extent and localisation of in vivo reduction of these Pt IV complexes to the active Pt II species may be a complication in clinical use.
The need to overcome platinum resistance, to reduce toxic side-effects and broaden the spectrum of activity has also focussed attention on the potential of anticancer drugs containing other metals. Among them, Ru-based-drugs are promising. 61 The Ru III anticancer drug imidazolium-trans-dimethylsulfoxideimidazole-tetrachlororuthenate (NAMI-A) (9) 61 (11) and [(Z 6 -p-cymene)Ru(PTA)Cl 2 ] 64 (p-cymene = para-cymene and PTA = 1,3,5-triaza-7-phosphaadamantane) (12) (see Chart 3) have been widely studied and several others also show promise. [65][66][67][68][69][70][71][72][73][74][75][76] The antimetastatic activity of NAMI-A is thought to be due to combined effects on the control of angiogenesis (possibly because it interferes with NO metabolism) 77,78 and antiinvasive properties towards tumour cells and blood vessels, and not to interaction with nucleic acids, although it can interact with DNA in vitro. 79 DNA has been proposed as one of the biological targets of KP1019, and it also triggers apoptosis. 80 However, the cellular mechanism of the activation of apoptosis is not understood. 81 These Ru III complexes may undergo activation by reduction to Ru II in vivo. 82 Different mechanisms of action have been investigated for explaining the anticancer activity of arene ruthenium complexes. Interactions between complexes containing reactive Ru-Cl and nuclear DNA can occur, through the formation of intermediary aqua complexes able to ruthenate DNA specifically at guanine residues. 83,84 If the arene is extended (e.g. biphenyl, dihydroanthracene), DNA binding can involve not only direct coordination to G bases but also arene intercalation (see Section 1.1.1.3.). Ligand oxidation can provide a way to activate thiolato complexes of Ru II arenes, which may be of importance when the intracellular thiol glutathione binds to them. Mono-(to give the sulfenate) and bis-(sulfinate) oxygenation appear to be facile, but surprisingly do not weaken the Ru-S bond 85   this provides a route to nucleobase binding. 86 Protonation of the sulfenate oxygen on the other hand does labilise this bond. 87 Osmium, the heavier congener of ruthenium and a third row transition metal, commonly exhibits slower kinetics than ruthenium, and is often considered to be relatively inert. However, it is possible to tune the biochemical reactivity of the arene Os II complexes through understanding their aqueous solution chemistry. The rates of hydrolysis of [(Z 6 -arene)Os(XY)Cl] n+ complexes can be controlled by the choice of the chelating ligand XY (faster for XY = O,O, followed by O,N and slowest for N-donor chelating ligands, especially when N,N is a strong p -acceptor such as azopyridine). The pK a values of the aqua adducts also follow this order, being highest for O,O ligands, with coordinated water tending to be more reactive than coordinated hydroxide. 88 Ligand substitution rates on arene Os II are often ca. 100Â slower than for Ru II and coordinated aqua ligands ca. 1.5 pK a units more acidic. Thus, active arene Os II complexes have been designed. [89][90][91][92][93][94][95][96][97][98][99] In particular chlorido Os II picolinate complexes can bind to DNA, but interestingly some inert iodido Os II azopyridine complexes exhibit nanomolar potency to a wide range of cancer cells and are also active in vivo. 100 The azopyridine complexes appear to have redox mechanisms of action. 101 Interestingly injected OsO 4 has a long history of use in therapy in Scandinavia for chemical synovectomyin the treatment of chronic synovitis. 102 In addition to binding to DNA, 103 Ir III Cp* anticancer complexes can also have redox mechanisms of action, readily forming hydride complexes on reaction with coenzyme NADH, and even producing H 2 catalytically. 104 These findings highlight the existence of new potential targets for such complexes.

even though
1.1.1.2. Non-covalent DNA binders. Metal complexes of an appropriate shape and polarity can intercalate between DNA base pairs. 105 Intercalators can be potent mutagens, due to their ability to induce structural and therefore functional changes in duplex DNA, leading to inhibition of transcription, replication and DNA repair processes. Intercalators are potential antibiotics, antibacterials, trypanocides, schistosomicides, and antitumour agents. 106,107 Several metallo-intercalator complexes have been reported recently, Table 3.
Besides intercalation and groove binding, highly charged multinuclear complexes can bind strongly to the phosphate backbone of DNA, forming a ''phosphate clamp. The trinuclear complex triplatinNC [{trans-Pt(NH 3 ) 2 (NH 2 (CH 2 ) 6 (NH 3 )} 2 -m-{trans-Pt(NH 3 ) 2 (NH 2 (CH 2 ) 6 NH 2 ) 2 }](NO 3 ) 8 (13, see Chart 4) exclusively utilises backbone functional groups to interact with DNA, and can associate with DNA even in the absence of direct coordination to DNA bases. 129 1.1.1.3. Coordination plus intercalation: dual mode DNA binders. Metal complexes with s-bonded aromatic side arms  118,119,128 or organometallic complexes with extended p-bonded arenes can act as dual-function complexes, binding to DNA both by direct metal coordination to a DNA base and intercalation between DNA bases through an attached aromatic ligand. 105 The utilisation of s-bonded side arm-containing platinum complexes acting as intercalators was reported in the late 1980s. 130,131 For example, Pt complexes incorporating planar aromatic ligands such as acridine orange, 9-aminoacridine, ethidium bromide, and acridinylthiourea (ACRAMTU) bind to DNA by both coordination and intercalation. [131][132][133] The dual mode interaction causes a strong structural modification by increasing the length and unwinding the DNA duplex. On the other hand, helical bending, C3 0 -endo deoxyribose puckering and rolling are not observed, which distinguishes the dual modeinduced damage from intrastrand cross-link damage. 35 The adducts formed by the potent complex [PtCl(en)(ACRAMTU-S)](NO 3 ) 2 (14) (Fig. 2) inhibit transcription of the damaged DNA by RNA polymerase II. 134 Additionally, biphenyl Ru II complex 11 (Chart 3) not only binds strongly and preferentially to G bases in DNA but also intercalates via the phenyl substituent on the Z 6 -arene. 135,136 In monofunctional adducts of {(Z 6 -biphenyl)Ru(en)} 2+ with the 14-mer (ATACATGGTACATA)Á(TATG*TACCATGTAT) ruthenation occurs at N7 of each guanine residue. At one site (G*), not only was intercalation of the arene between G* and adjacent T observed, but also stacking of a non-intercalated arene on a tilted adjacent thymine on the surface of the major groove. 137 Other dual-function coordination complexes have also been investigated, 138,139 such as the di-Rh II complex cis- A series of low-spin 5d 6 cyclopentadienyl Ir III organometallic half-sandwich complexes has been shown to form adducts with 9-ethylguanine and/or 9-ethyladenine readily, depending on the chelating ligands. Moreover, in the case of complexes incorporating extended Cp* ligands, the ability to intercalate into DNA appears to contribute to the anticancer potency of these Ir III complexes. 123 Combining Pt and other metal ions also provides a strategy for designing complexes able to bind DNA both by coordination and by intercalation. For example, in heterobimetallic complexes containing Pt and Ru centres joined by a linker, 140 Pt II offers coordinative binding through ligand substitution, and octahedral Ru II chelated by terpyridine or extended pyridyl ligands can act as an intercalator as well as possessing useful optical properties. 140 Alternatively, the Pt unit is capable of interacting with DNA by p-p stacking or coordination through Cl substitution, and the Ru unit can bind to DNA by electrostatic or surface binding, or partial intercalation. 141 1.1.1.4. G-quadruplex binders. Telomeres in human genes contain many repeats of the sequence d(GGTTAG) (G-quartets) and are responsible for maintaining cell division. In these G-rich strands, four-stranded G-quadruplexes can be formed consisting of planar G-quartets stabilised by K + or Na + . These inhibit the enzyme telomerase. Targeting either telomerase or stabilising G-quartets is an effective strategy for the design of anticancer agents.
The first, and only example to date, of an X-ray crystal structure of human telomeric G-quadruplex DNA bound to a metal complex was reported in 2012. 142 Nickel II and copper II salphen metal complexes were co-crystallised in the presence of human telomeric DNA and the effective binding of these metalsalphen complexes to human telomeric quadruplexes by direct end-stacking was demonstrated (see Fig. 3). Other methods have also been used to demonstrate the binding of metal complexes to quadruplex DNA. Table 4 lists examples of coordinative and non-coordinative metal-based G-quadruplex DNA binders reported between 2009 and 2012.   polyamides with multiple platinum groups leading to high DNA-sequence-selectivity, water-solubility and potentially highly active complexes. 165,166 Specific recognition of DNA sequences can be also used to overcome side-effects and resistance to platinum drugs. Similarly incorporation of an oligonucleotide into a Pt complex allows sequence recognition within double stranded DNA by the formation of triple helices. 167 Various pathways to metal-containing oligonucleotides have been designed, such as conjugation of an end-functionalised oligonucleotide, chelation of an oligonucleotide to the metal or incorporation of metalcontaining phosphoramidites or hydrogenophosphonates. Oligodeoxynucleotide-tethered bifunctional cis-dichlorido Pt complexes 168 can retain cross-linking ability.
1.1.2.2. Protein-mediated DNA recognition. The design of metallodrugs incorporating groups capable of interacting specifically with DNA sequences is a challenge. An elegant strategy is to use a presenter protein and a small molecule that can bind to both protein and target. 169 The formation of such ternary complexes has been described for organic drugs, such as the immunosuppressive antibiotic rapamycin, 170 and has been illustrated as a concept for metallodrugs by an arene-Rubiotin derivative 171 embedded in the tetrameric streptavidin protein (Fig. 4). 172 The supramolecular structure binds G-quadruplex DNA with selectivity towards DNA telomeres, even in the presence of competing targets (such as glutathione). However, because of a poor cellular uptake, the in vivo activity of the system was not investigated.

Protein targeting
A number of proteins (e.g. kinases, bacterial Zn enzymes) have been proposed as potential biological targets for metal complexes or chelating agents in the case of metalloprotein targets. [173][174][175][176][177][178][179][180][181][182][183][184][185][186][187] Modern bioanalytical techniques with high sensitivity and selectivity have facilitated the discovery of potential protein targets for metal complexes. 188 However, as for DNA, the ability of a metallodrug to interact with amino acids, peptides or proteins in vitro does not necessarily imply a direct involvement in the cellular mechanism of action of this metallodrug. The involvement of particular targets has to be validated (e.g. by knockdown experiments, silencing gene expression by RNAi techniques). This involves blocking production of the protein in the cell and observing the effects on the activity of the drug candidate. So far, this has been rarely done for metallodrugs. Correlations of activity with protein binding (including inverse) can be useful for structure-activity relationships. 189 We now illustrate recent work on interactions of metallodrugs with potential protein targets. The examples are illustrative and not comprehensive.
1.2.1. Hormone receptors. Prostate cancer is the most common cancer for men and the second killing-cancer after lung cancer. The majority of prostate cancers are initially androgen-dependent and might be treatable by hormone therapy. This therapy aims to block androgen receptors (AR) by binding antagonists (antiandrogens) to these receptors instead of the endogenous testosterone. 190 Three are currently in clinical use for treatment of prostate cancer: flutamide, 191 nilutamide, 192 and bicalutamide, 193 (Chart 6). Despite promising results, resistance acquired by cancer cells (possibly due to mutations of the androgen receptor) is one of the major limitations of the use of hormone therapy. 194 The discovery of new antiandrogen compounds effective for both hormone-dependent  and hormone independent prostate cancers is therefore of major importance. Among these potential antiandrogens, are steroid hormones progestogens (and synthetic derivatives progestins). Ferrocenyl organometallic complexes incorporating steroidal androgens testosterone and dihydrotestosterone (DHT) exhibit a strong antiproliferative effect on the hormone-independent prostate cancer cells PC-3 with IC 50 values in the low micromolar range. 195 Tamoxifen ((Z)-2-[4-(1,2-diphenyl-1-butenyl)phenoxy]-N,Ndimethylethanamine, Chart 6) is a bioavailable selective estrogen receptor modulator commonly prescribed for preventing and treating hormone-dependent breast cancers. Its metabolite, hydroxytamoxifen, is the active form which competitively binds to estrogen receptors. Despite this clinical success, tamoxifen suffers from critical limitations: ineffectiveness towards ER-tumours (a third of hormone-dependent tumours), development of resistance mechanisms (another third), and increases of uterine and endometrial cancer risks. 196 Ferrocenyl tamoxifen derivatives, obtained by substitution of the b-phenyl ring of hydroxytamoxifen by a ferrocenyl fragment (Chart 6), show promise for the treatment of both hormone-dependent and independent breast cancer cells. 197 The replacement of the phenyl group by ferrocene reduces receptor affinity by about 40%, whilst the increase in length of the dimethylaminoalkyl chain has an adverse effect on receptor binding. In addition to tamoxifen-like binding to the estrogen receptors, ferrocifens are likely to be activated by oxidation to quinone methides in cells, suggesting a dual biological mode of action. The mechanism of formation of these quinone methides from ferrocenyl phenols is a two-step pathway involving two successive single-electron oxidations, accompanied by deprotonation which stabilises the quinone radical, resulting in the quinone methide structure. 88 The proposed quinone methides have been characterised, and indeed appear to be the active metabolites of ferrocifens. 198 Ruthenocenyl tamoxifen analogues behave essentially as antiestrogens. 199 Electrochemical studies of such complexes showed that the oxidation of the ruthenocenyl fragment is  irreversible and leads to rapid decomposition of the organometallic entity. 197 Similarly, cyclopentadienyl Re analogues do not exhibit antiproliferative activity, ({CpRe(CO) 3 } acting as a spectator group). 200 Therefore, the redox activity of the ferrocenyl group is of central importance, along with the antiestrogenic properties of the hydroxytamoxifen derivative, for providing a unique dual mechanism of action of these ferrocenyl tamoxifen derivatives.
The anticancer activity of titanocene dichloride has been known for decades, but its efficacy in Phase II clinical trials in patients with metastatic renal cell carcinoma 201 or metastatic breast cancer 202 was too low to be pursued. The titanocenylderivative of tamoxifen 203 has estrogenic effects on hormonedependent breast cancer cells in the nanomolar range. Unfortunately, hydrolysis of the titanocene fragment leads to the generation of Ti IV species that behave in a similar way to estradiol towards the estrogen receptor. This explains the estrogenic effect observed for Cp 2 TiCl 2 and precludes the use of Cp 2 TiCl 2 derivatives of tamoxifen as drug candidates for the treatment of breast cancer. It also poses the problem of the possible role of titanium salts as endocrine disruptors. 197 Nonetheless, the recent development of highly active water-soluble ring-substituted cationic titanocene dichloride derivatives has reactivated interest in such organometallic complexes. [204][205][206] Another promising way of targeting hormonal receptors is utilisation of carboranes as estrogen receptor agonists and antagonists. Estrogen receptors are over-expressed in estrogenreceptor-positive tumours, such as in some breast cancers. 207 Carboranes are versatile pharmacophores and possess unique properties that make them useful in inorganic chemistry. 208,209 Compounds incorporating carborane cages structurally close to estradiol, 210 can act as efficient estrogen receptor agonists. 211 Closo carborane tamoxifen is more stable than tamoxifen itself, and the Z carborane tamoxifen isomer exhibits similar inhibition properties as tamoxifen.
1.2.2. Mitochondrial protein targeting. Mitochondria in cells regulate energy production and modulate redox potentials as well as generating reactive oxygen species (ROS e.g. O 2 À and H 2 O 2 ). In cancer cells, mitochondria have hyperpolarised membranes and over-produce ROS. Their defective function in cancer cells makes them potential targets for anticancer drugs.
Mitochondrial proteins are potential targets for arsenic compounds which can bind strongly to vicinal (Cys) thiol groups, increase ROS production and induce apoptotic signalling pathways. Several studies have highlighted the potential of arsenic metallodrugs in cancer therapy. Arsenic(III) trioxide (As 2 O 3 ) has been used as a therapeutic agent for over 2000 years. [212][213][214] More recently Ehrlich screened hundreds of organoarsenic compounds for biological activity and in 1910 introduced Salvarsan (arsphenamine) for the treatment of syphilis. Currently, As 2 O 3 , (ATO, 'Trisenox', Table 1) is the most effective single agent for the treatment of acute promyelocytic leukaemia. 215 In aqueous solution ATO exists as the trihydroxide As(OH) 3 and is taken up into cells via the aquaporins (especially aquaglyceroporins 216 )membrane transport proteins.
GSAO (4-(N-(S-glutathionylacetyl)amino) phenylarsonous acid) is a promising new compound, known to inhibit adenine nucleotide translocase (ANT) in the inner membrane of mitochondria. A phase I clinical study is in progress in patients with solid tumours refractory to standard therapy (Table 1). 217 Metabolism of GSAO is required for biological activity. This metabolism involves a two-step mechanism, first cleavage by g-glutamyltranspeptidase at the cell surface to give GCAO (4-(N-(S-cysteinylglycylacetyl)amino) phenylarsonous acid). Then GCAO enters cells via an organic ion transporter and is metabolised by dipeptidases to CAO (4-(N-(S-cysteinylacetyl)amino) phenylarsonous acid) in the cytosol (Fig. 5). Finally, CAO enters the mitochondrial matrix and As III cross-links cysteine residues 57 and 257 of human ANT1, 218 so inhibiting this enzyme.
The analogue of CAO, PENAO (4-(N-(S-penicillaminylacetyl)amino) phenylarsonous acid) (Fig. 5), is accumulated in cells ca. 85-fold faster than GSAO, has a 44Â higher antiproliferative activity and 20Â higher antitumour efficiency in mice. In 2012, patients with solid tumours refractory to standard therapy were recruited for Phase I/IIa dose escalation studies. 218 In general, many positively-charged lipophilic complexes are taken up by mitochondria, for example cartionic gold(I) phosphines and carbenes. 219,220 Another example is inert polypyridyl Ru II complexes which target mitochondrial function and induce apoptosis. A strategy for combinatorial parallel coordination chemistry has been recently used, giving access to a library of more than 500 monocationic polypyridyl ruthenium complexes. 221 These complexes were screened for cytotoxicity towards cancer cells, and structure-activity relationships led to the discovery of a lead complex [Ru( t Bu 2 bpy) 2 (phox)]PF 6 ( t Bu 2 bpy = 4,4 0 -di-tert-buty-2,2 0 -bipyridine and Hphox = 2-(2 0 -hydroxyphenyl)oxazoline) (15, Chart 7), which is active at submicromolar concentrations in clinically relevant Burkitt-like lymphoma cells. This complex strongly reduces the mitochondrial membrane potential, suggesting involvement of the intrinsic pathway of programmed cell death. These complexes are chiral, but all compounds in this study were formed as mixtures of enantiomers or diastereoisomers. It would be interesting to test the separated enantiomers for activity.
Another example of mitochondrial targeting is the complex [Ru(dppz) 2 (CppH)] 2+ (CppH = 2-(2 0 -pyridyl)pyrimidine-4-carboxylic acid; dppz = dipyrido[3,2-a:2 0 ,3 0 -c]phenazine, 16, Chart 7), which impairs the mitochondrial membrane potential in HeLa cells as early as 2 h after treatment and induces apoptosis. 222 1.2.3. Kinases, TNF-a and thioredoxin. The selective targeting of enzyme kinases is a particularly complex task, since more than 500 proteins belong to this family, although they all possess highly conserved ATP binding sites. 223 Inert octahedral metal complexes are emerging as promising scaffolds for targeting kinase active sites, 224 thanks to their rigid and globular shapes, and high synthetic versatility. 225 Meggers and co-workers have demonstrated that specific Ru II , 226 Os II , 227 Rh III , 228 and Ir III (ref. 229) complexes can serve as highly potent (micromolar to nanomolar range) and selective inhibitors of kinases. In most cases, the racemic complexes were resolved into the enantiomers by chiral HPLC (D/LÀ, or R/S-enantiomers), and the individual isomers were tested against different kinase enzymes. The affinities of the isomers for kinase ATP-binding sites are markedly different (for instance, the L-enantiomer of a ruthenium complex (FL-172) 27-fold more potent for kinase PAK1 compared to the D-enantiomer 226 ). The Ru centre is not involved in any direct interactions and has solely a structural role. 230 A striking example of this work is the crystal structure of PIM2 (a serine/threonine kinase over-expressed in human leukaemia and lymphomas) with a bound R-configuration organoruthenium inhibitor (17, see Fig. 6 for molecular structure), which shows the good shape complementarity between the Ru complex and the ATP site of kinase PIM2, Fig. 6. An interesting strategy is the combination of kinase inhibition and photoactivity demonstrated for an angiogenic octahedral organoiridium complex. This can undergo substitution of a selenocyanate ligand on irradiation with visible light and induces apoptosis in cancer cells. 231 There are good prospects for further improving the selectivity of photoactivated metallodrugs by conjugation to targeting vectors such as peptides. 232 Aberrant activity of tumour necrosis factor-a (TNF-a, a proinflammatory cytokine involved in the regulation of many key biological processes, e.g. haematopoiesis, immunity, and inflammation) 234 is associated with a number of diseases, such diabetes, tumourigenesis, and autoinflammatory diseases. Clinical trials in ovarian cancer suggest that synthetic therapeutic antibodies (e.g. infliximab), which bind directly to TNF-a, may be effective in blocking its interaction with the tumour necrosis factor receptor (TNFR). [235][236][237] Despite this success, synthetic antibodies suffer from limitations, such as development of anti-antibody response, which has led to the search of alternative small-molecule-based therapies as inhibitors of TNF-a. Based on the approach developed for kinase inhibition, inert octahedral metal complexes, such as the cyclometalated biquinoline iridium(III) complex 18 (see Chart 8), hold potential as direct TNF-a inhibitors. 238 Interest in targeting the thioredoxin system -including thioredoxin (Trx), thioredoxin reductase (TrxR) and NADPHled in 2006, to the submission of a New Drug Application to the FDA for Motexafin gadolinium, an inhibitor of thioredoxin reductase and ribonucleotide reductase, for the treatment of lung cancers with brain metastases. This compound is currently in clinical development as a radiation and chemotherapy sensitiser (Table 1). Thioredoxin reductase remains an attractive target for metallodrugs (Table 5), particularly for gold complexes, [239][240][241][242][243] although there is a need to validate this as a target for these metal complexes in vivo. Thioredoxin reductase (TrxR) contains FAD and NADPH binding domains and a redox-active disulphide (Cys-Cys) bond in its active site. It transfers electrons to thioredoxin which, in turn, reduces disulphide bonds and other substrates. Mammalian TrxRs contain a second redox-active site, a C-terminal -Cys-SeCys-(where SeCys is selenocysteine). 244 1.3. Metallodrug delivery and activation 1.3.1. Light activation of Pt IV prodrugs. Spatially-targeted activation by light is a possible way of increasing the efficiency of Pt IV prodrugs, avoiding unnecessary damage to normal tissues, and delivering the active drug mainly to the tumour itself to enhance activity or activate the drug specifically in the cancer cells. 260 The metal complexes used are generally stable to thermal activation so that they reach the target site intact. However the excited singlet and triplet states (which are often reached very quickly, in pico-nano seconds) have different electron distributions to the ground state and hence different geometries and different reactivities. 261 Excited state drugs offer the prospect of novel mechanisms of action. 262 The wavelength of activation is important because longer wavelengths (e.g. red light) penetrate tissues more deeply than shorter wavelengths (e.g. blue light).
Platinum(IV) complexes of the type [Pt(OH) 2 (N 3 ) 2 (amine1)-(amine2)] have strong azide-to-Pt IV charge-transfer bands, are stable in the dark, and towards the intracellular reducing agent glutathione. 263 After short treatment and short irradiation times (e.g. 1 h), they react rapidly with DNA bases such as guanine and are potently cytotoxic. Interestingly, the trans diam(m)ine diazido complexes appear to be more effective as photoactivatable anticancer agents than the cis isomers (Fig. 7). 264 These complexes are also more effective than cisplatin when used under conditions appropriate for clinical phototherapeutic drugs (short treatment times, short irradiation times).  The DNA lesions are unusual and include interstrand crosslinks. Trans,trans,trans-[Pt(OH) 2 (N 3 ) 2 (NH 3 )(py)] is active in vivo on activation with blue light in an oesophageal cancer model. 265 Trans,trans,trans-[Pt(OH) 2 (N 3 ) 2 (py) 2 ] (py = pyridine) undergoes photoreduction when irradiated by UVA, blue or green light, 266 and also produces azidyl radicals which can be quenched by L-Trp. 267 Therefore these diazido complexes may have a dual mechanism of action involving the production of reactive Pt II and radicals from the Pt IV prodrug. Unlike conventional photosensitizers (which convert 3 O 2 to excited 1 O 2 : photodynamic therapy), these complexes would not rely of O 2 for activity, a potential advantage since tumours are often relatively hypoxic.
Other metal complexes are potent light activated anticancer agents. For instance, strained octahedral tris-bipyridyl ruthenium(II) complexes have been shown to be inert until triggered by visible light. 268 Under irradiation, a light-activated ligand release mechanism occurs, leading to DNA binding, and to an increase in cytotoxicity of 2 orders of magnitude in cancer cells (with potencies superior to cisplatin against 3D tumour spheroids). The use of intramolecular strain is a promising strategy for developing light-activated Ru complexes for PDT applications.
1.3.2. Photo-release of biologically-active small molecules. Photoactivation is also a promising strategy for the release of biologically-active small molecules such as NO, CO, and H 2 S. NO is an important signalling molecule with a wide range of functions in the cardiovascular, nervous, and immune systems. 269 The interactions of NO with metal complexes in vivo -heme and non-heme Fe in particular -are of prime importance to its physiological role. Metal complexes show promise for both controlled release and scavenging of NO. 270 Examples include [Ru(terpy)(bdqi)NO] 3+ (terpy = terpyridine, bdqi = 1,2-benzoquinone diimine) which can release NO on irradiation with visible light. This complex can be delivered in lipid nanoparticles via topical administration and provides a possible treatment for skin cancer. 271 CO is also a natural signalling molecule which can be released from a metal centre by light activation, 272 and metal carbonyls have been extensively investigated as potential CO-donating pharmaceuticals. 273,274 In vivo, CO appears to have a role as a messenger, has anti-inflammatory properties and an ability to suppress organ graft rejection. The design of metal complexes that can release CO at a predicable rate is therefore valuable as a relatively non-toxic source of CO. Cell viability studies of HT29 colon cancer cells treated with the CO-releasing compound [Mn(CO) 3 (tpm)]PF 6 (tpm = tris-(pyrazolyl)methane) have revealed a significant photoinduced cytotoxicity, comparable to that of the established anticancer agent 5-fluorouracil. 275 1.3.3. Anticancer platinum drug delivery. The limitations of Pt IV drug candidates might also be overcome by the utilisation of drug cargos. In principle, these transporters can carry a large number of platinum centres, shield the drug from premature activation, and deliver platinum specifically to cancer cells. Moreover, cancer cells divide rapidly and have a higher demand for nutrients than normal cells. Because of this need for nutrients, the vascular endothelial growth factor (VEGF) is expressed and induces angiogenesis (formation of new blood vessels for the tumour cells) which increases the permeability of the cell membrane. The increased permeability leads to a higher uptake of large molecules and proteins. However the major difference from normal cells resides in the lymphatic drainage, which is severely impaired for cancer cells, thus leading to the retention of large molecules and lipids inside the cell. 276 This enhanced permeability and retention (EPR) effect of cancer cells is an excellent target for the development of anti-cancer agents: 277 Large molecules might be selectively taken up by cancer cells and retained whereas excretion by normal cells through their healthy lymphatic system might occur.
A large number of carrier systems have been developed in recent years, spanning functionalised carbon nanotubes, nanorods, metal-organic frameworks, metalla-cages, nanoparticles, liposomes, nanogels, proteins, and polymers. This wide diversity of systems offers an interesting pool of carriers, and each system possesses unique advantages. 260 The delivery system can also influence the activity of metallodrugs. For example, a Pt IV complex has been recently tethered via amide linkages to gold nanoparticles (AuNPs) functionalized with thiolated oligonucleotides. The resulting systems exhibit 12-fold higher activity than free cisplatin towards A549 lung cancer cells. 278 The delivery of a lethal dose of cisplatin to prostate cancer cells by the encapsulation of Pt IV prodrug in prostate-specific membrane nanoparticles (NPs) has also been recently reported. 279 The utilisation of prostate-specific membrane antigen to target aptamers overexpressed in tumour cells, allows an specific delivery of the active drug into cancer cells. The release of the Pt IV complex from the nanoparticles is followed by reduction to Pt II (cisplatin), which can subsequently form 1,2-d(GpG) intrastrand cross-links on nuclear DNA.
Ruthenium metalla-cages also show promise as drug delivery systems. Supramolecular metalla-prisms based on arene ruthenium complexes can encapsulate square-planar acetylacetonato Pd II and Pt II complexes in their cavities and deliver them into cells, (e.g. [19] 6+ , see Chart 9). 280,281 The activity of these carceplexes towards human ovarian cancer cell lines is more than an order of magnitude higher than the empty metalla-cage. 281 Metalla-prisms can be synthesised with larger portal sizes (e.g. [20] 6+ , [21] 6+ , and [22] 6+ , see Chart 9). 282,283 The guest can be released from such metalla-cages without rupture of the cage, [284][285][286][287][288] and the extent of drug release correlates with the portal size of the cage. 289 The host-guest capability of these systems has been used for encapsulating hydrophobic pyrenyl-cycloplatinate complexes, 290 pyrenylcontaining dendrimers of different generations, for targeting cancer cells via the EPR effect, [291][292][293] and for incorporating guests having affinity for G-quadruplex DNA. 294 They can also be used as vehicles for intracellular delivery of photosensitisers, with possible use in photodynamic therapy. 295 These systems being robust, highly water-soluble, and versatile are promising for drug delivery.

Anti-viral, anti-microbial and anti-diabetic metallodrugs
2.1. Anti-viral metallodrugs 2.1.1. Hepatitis. According to The World Health Organization, hepatitis C viral infection (HCV) affects 150 million people with consequences spanning from a mild illness lasting a few weeks to a serious, lifelong condition that can lead to cirrhosis of the liver or liver cancer. Every year, more than 350 000 people die from hepatitis. 296 This disease can be treated by combination therapy involving pegylated recombinant interferon and ribavirin. However, interferon is not always well tolerated, some HCV genotypes respond better to interferon than others, and both interferon and ribavirin are not selective for HCV or viral disease in general. 297 Moreover, the absence of an effective vaccine is stimulating the search for new drugs. A possible strategy for such development is to target specifically the critical RNA sequences present in hepatitis C virus but rare in the genome of infected host cells, For instance, the synthesis of a catalytic metallodrug that targets stem-loop IIb of the internal ribosomal entry site (IRES) RNA of hepatitis C virus has been recently reported. 298 This metal complex incorporates an amino-terminal copper and nickel binding motif (ATCUN) that is found naturally at the N-terminus of many albumins 299 as well as natural peptides such as histatin 5 300 and neuromedin, 301 and which binds copper and nickel with very high affinity. 302 The ATCUN ligand provides a stabilisation of the redox states Cu(III)/Cu(II) and prevents the formation of the labile Cu(I) state. A C-terminal tetrapeptide (YrFK-amide) targeting domain is also coordinated to the metal centre in order to provide a selective recognition of the HCV IRES stem-loop IIb domain (Chart 10). Interestingly, neither the targeting peptide alone YrFK-amide, lacking the metal binding ATCUN motif, nor the metal binding ATCUN domain alone show any cellular efficacy. However, the complex reacts in vitro with the HCV IRES stem-loop IIb domain and inactivates catalytically and irreversibly the replication of hepatitis C virus with a turnover number of about 32. 298 Despite this low turnover number, this preliminary study represents a promising new approach to the design of anti-hepatitis drugs.

Human immunodeficiency virus.
Drugs for treatment of the Human Immunodeficiency Virus (HIV) are targeted to various pathways and enzymes: nucleoside reverse transcriptase inhibitors, non-nucleoside reverse transcriptase inhibitors, protease inhibitors, and integrase inhibitors. 303 The enzyme integrase (IN) catalyses the integration of viral DNA into the host cell DNA and is a particularly interesting target for HIV inhibitors. The active centre of the enzyme possesses two Mg II ions held in place by a triad of protein carboxylate side chains, as well as an HHCC zinc finger site. Ligands which can chelate to the Mg II ions might be effective enzyme inhibitors and drugs. Such a drug is raltegravir, approved by the FDA in 2007 (see Fig. 8). Despite its clinical success it suffers from an overall dose burden, and a lack of potency. To help to overcome rising raltegravir resistance, Cohen et al. have synthesised a series of raltegravir-chelator derivatives (RCD) as HIV integrase inhibitors in order to assess the role of the metal-binding group. 304 The binding mode for this series of RCD molecules was elucidated by docking simulations, based on a crystal structure of prototype foamy virus (PFV) integrase in complex with raltegravir (PDB ID 3OYA). The docking of raltegravir and of RCD-1 (structure in Fig. 8) into PFV IN gives identical binding modes (Fig. 8) 306 An effective anti-HIV strategy is to target the membrane protein CXCR-4, a seven-helix transmembrane G-proteincoupled receptor and one of the several chemokine receptors that HIV uses to infect CD4+ T cells. The bis-macrocycle xylylbicyclam has potent anti-HIV activity but its use in the treatment of HIV was hindered by its lack of oral availability and cardiac disturbances. 307 During clinical trials, its effectiveness in mobilising stem cells from the bone marrow was discovered, which led to clinical approval by the FDA in 2008 for this purpose (e.g. in transplant therapy) as the drug Mozobil. 308 Metal ions such as Zn II and Cu II bind to cyclam strongly 309 and relatively rapidly, 310 and it seems likely that metal complexation by xylyl-bicyclam is involved in the mechanism of action of the drug in vivo. The affinity of xylyl-bicyclam for the CXCR4 receptor is enhanced by factors of 7, 36, and 50 by incorporation of Cu II , Zn II , or Ni II , respectively, into the cyclam rings. Dizinc xylybicyclam tetraacetate forms the unusual folded cis-V configuration with an acetate carboxylate bound to Zn II on one side of the cyclam ring and acetate forming a double H-bond to cyclam NH groups on the other side of the ring. 311 In a model of the Zn drug docked onto human CXCR4, a similar coordination can be achieved involving the carboxylates of Asp171, Asp162 and Glu288. Hydrophobic interactions involving Trp side chains and the periphery of the cyclam rings together with metal-carboxylate binding and H-bonding are observed in the X-ray crystal structure of Cu II -cyclam and Cu II 2 -bicyclam adducts of the protein lysozyme (Fig. 9). 312 Configurationally constrained metal-cyclam complexes have potential as even more potent antiHIV agents. For example, Zn II hexyl-dimethyl-cyclam complexes are active, but more active are    biological mechanisms are needed, and although metal complexes might provide efficient alternatives, very few examples of such complexes have been reported, and in most cases, their mechanisms of action have rarely been thoroughly studied.
Nonetheless, some recent studies have demonstrated the potential of metal-based drugs for the development of antituberculosis agents. Among them, the redox activation of isoniazid iron(II) complexes is of interest. 316 Indeed, isoniazid has been used as a front-line drug in the treatment of TB, although resistant TB strains have limited its use. Isoniazid is a prodrug that needs to be activated by an electron transfer reaction. This reaction is catalysed by the catalase-peroxidase KatG, and leads to the formation of an intermediate isonicotinic acyl radical that promptly reacts with NADH, generating a NAD-isoniazid adduct. This adduct is an efficient inhibitor of the enoyl reductase enzyme InhA, a major target for anti-Mycobacterium tuberculosis agents. 317 About 50% of isolated isoniazid-resistant strains have either a deletion or mutations in the katG gene and this KatG enzyme disruption (blocking electron transfer reactions) has been shown to be the major cause of isoniazid resistance. 318 For overcoming such resistance, the coordination of isoniazid to metal complexes (e.g. cyanoferrates) might allow a rapid oxidation of the metal centre, triggering isoniazid activation intramolecularly, independently of the activation by KatG enzymes (see Fig. 10 for a possible mechanism of action). This hypothetical mechanism still needs to be supported by experimental evidence, but preliminary studies on an isoniazid-containing pentacyanoferrate complex show that an inner-sphere electron transfer reaction occurs between the metal and isoniazid, thereby activating this prodrug and overcoming the necessity for KatG. This metal complex inhibits both wild-type InhA and its isoniazid-resistant mutant InhA I21V, even in the absence of KatG and NADH, bypassing the enzymatic activation.
Interestingly, the ruthenium analogue (isoniazid)pentacyanoruthenate(II) complex does not inhibit InhA, probably because its very high electrochemical potential is outside the biological range.
2.2.2. Other anti-microbial agents. Many metal compounds show appreciable antimicrobial activity, some established examples are based on silver, bismuth, mercury, arsenic, and antimony. For instance, the arsenic-based antimicrobial agent Salvarsan ((AsR) n , R = 3-amino-4-hydroxyphenyl, where n = 3 and 5 are the most abundant species in solution) was used historically to treat syphilis and trypanosomiasis, although in recent times it has been replaced. It is suggested that oxidation in vivo generates the active form of the drug, such that Salvarsan serves as a slow-release source of RAs(OH) 2 (oxidation from As I to As III ). 319 The importance and historical role of inorganic and organometallic complexes as antimicrobial agents have been recently reviewed. 320,321 A wide range of inorganic compounds from organometallic complexes to metal organic frameworks, nanoparticles or metallic surfaces have also been recently investigated for killing or inhibiting microbial growth, including mercury, 322 silver, 323 gold wires 324 and nanoparticles, 325 copper, 326 cobalt, 327 platinum, 326 palladium, 328 ruthenium, 329 iron (chelation therapy), 330,331 manganese complexes, 332 titanium dioxide nanoparticles, 333 noble metal nanoparticles, 334,335 and nanostructures as antibacterial drug delivery systems. 336

Anti-diabetic metallodrugs
Diabetes mellitus (DM) is a group of chronic metabolic diseases characterised by persistent hyperglycaemia associated with absolute or relative deficiency in insulin secretion from the beta cells of pancreas. The dysfunction of insulin receptors might also be associated with diabetes. The current (285 million patients worldwide) and dramatically-increasing incidence of this disease is stimulating the search for new anti-diabetic agents. 337 Potential insulin mimetic metallodrugs have received much interest in the last two decades. The aim of developing new oral insulin mimetic drugs that act independently of insulin, for both insulin-dependent and insulin-independent diabetes, is to overcome side effects such as hyperinsulinemic hypoglycaemia due to current insulin preparations and synthetic drugs used in clinic.
Zinc and vanadium complexes have been widely studied as potential metallopharmaceutics for treating diabetes mellitus. Zinc plays an essential role in the physiology, structure and function of insulin, 338 and vanadium is also an essential element. The complex bis(maltolato)oxovanadium(IV) (BMOV, Chart 11), underwent phase II clinical trials but failed due to effects on the kidneys of patients. To reduce toxicity, Wei et al. have synthesised a vanadium complex which incorporates a modified pyranone derivative (BBOV, Chart 11). 339 BBOV has low in vivo toxicity and oral administration of this complex to STZ-induced diabetic rats leads to a dramatic reduction of hyperglycaemia, along with an increase of the impaired glucose tolerance activity.
A third metal of interest for development of anti-diabetic drugs is chromium, an element for which there is now no good  evidence that it is essential. 340,341 Table 6 summarises recent examples of anti-diabetic Cr III complexes and their activity. Tetrahedral vanadate, molybdate, and tungstate complexes can compete with phosphate substrates for binding to phosphatases, 352 and are potent inhibitors of muscle glycogen phosphorylases by competing with glucose-1-phosphate. 353 Sodium tungstate Na 2 WO 4 has been particularly investigated, due to the restoration of hepatic glucose metabolism by the stable oxoanion [WO 4 ] 2À . 354 Moreover, [WO 4 ] 2À appears to mimic most of the metabolic effects of insulin and stimulate insulin output. 355 The Keggin anion [PW 12 O 40 ] 3À has been also investigated for its ability to mimic insulin. [356][357][358] Sodium molybdate Na 2 MoO 4 is similarly effective for preventing or treating of diabetic mellitus in the early stages of the disease. 359 However none of these compounds has yet received clinical approval as a new antidiabetic drug.
There is also increasing interest in the potential use of vanadium compounds for treating leishmaniasis, Chagas' disease and amoebiasis, and viral infections. 360 3. Anti-parasitic, anti-inflammatory and anti-neurodegenerative metallodrugs 3.1. Anti-parasitic metallodrugs 3.1.1. Malaria. The absence of a vaccine for the prevention and/or treatment of malaria is one of the major factors stimulating the search for new anti-malarial therapeutics. Malaria is treated mainly with a combination of chloroquine and antifolate drugs, or by artemisinin derivatives. Innovative drugs having new biological mechanisms of action are needed, due to the emergence of a high number of drug-resistant Plasmodium strains. In this context, metallodrugs are a promising class of compounds that might lead to a breakthrough in antimalarial therapy.
Plasmodium falciparum is one of the five species of Plasmodium parasite that causes the most lethal form of malaria. Plasmodium falciparum is particularly sensitive to oxidative stress, 361 and targeting of thioredoxin reductase is an approach that is being widely studied. 362 Auranofin (Fig. 11) and a few related gold complexes strongly inhibit Plasmodium falciparum growth, probably due to a direct inhibition of Plasmodium falciparum thioredoxin reductase. In 2012, based on studies demonstrating that auranofin is a pro-drug giving rise to the active fragment {Au(PEt 3 )} + (while the tetraacetylthioglucose ligand is excreted in vivo), 363 molecular docking experiments 364 were carried out and suggest that {Au(PEt 3 )} + can bind to a N atom of an histidine in the active site of Plasmodium falciparum thioredoxin reductase, Fig. 11. This mode of binding to a protein His sidechain (despite the presence of Cys thiol sulfurs) was observed previously for {Au(PEt 3 )} + in the X-ray structure of cyclophilin-3. 365 Gold(I) also has a high affinity for selenocysteine which provides selectivity for inhibition of thioredoxin reductase over glutathione reductase. 251 Another approach for designing metallodrugs with potency against malaria is to coordinate biologically-active quinolone derivatives to metal centres. The ferrocene-quinoline conjugate ferroquine has been on clinical trials (Chart 12, Table 1). 366 Chloroquine is thought to interfere with the digestion of haemoglobin in the blood stages of the malaria life cycle. Even though some similarities between the biological mode of action Chart 11 Molecular structures of vanadyl anti-diabetic drug candidates BMOV and BBOV.  of chloroquine and ferroquine have been observed, 367 ferroquine exhibits additional mechanisms. 368 In particular, ferrocene can undergo a one-electron oxidation, yielding the ferrocenium cation, which may generate hydroxyl radicals under physiological solutions, 369,370 leading to potential DNA 371 and cell membrane damage. 372 Moreover, the weaker base properties and higher lipophilicity at physiological pH of ferroquine compared to chloroquine, as well as intramolecular H-bonding with the lateral side chain of ferroquine (in non polar conditions) leads to the improved ability of ferroquine to cross membranes and a higher accumulation in the digestive vacuole. 373 Ferroquine and ruthenoquine analogues possess slightly different mechanisms of action, Ruthenoquine does not produce reactive oxygen species (ROS) under the oxidative conditions of the parasitic digestive vacuole of Plasmodium falciparum. 374 However, both compounds are active against drug-susceptible and drug-resistant strains of Plasmodium falciparum. 375 Ruthenocenic derivatives of chloroquine with and without an intramolecular hydrogen bond have been used to study the localisation and quantification of ruthenoquine in Plasmodium falciparum-infected erythrocytes. 374 This study suggested that the presence of intramolecular H-bonding (Chart 12) substantially improves the membrane permeability and transport of the drug to its target.
3.1.2. Amoebiasis. Amoebiasis is an infection of the intestine caused by the parasite Entamoeba histolytica, responsible for about 70 000 deaths a year. 376 The World Health Organisation estimates that amoebiasis is the fourth leading cause of death due to protozoan infections (after malaria, Chagas disease, and trichomoniasis). Current medications for the treatment of amoebiasis are based on nitroimidazole derivatives, such metronidazole, and tinidazole. However, the appearance of resistant E. histolytica strains 377 and side effects associated to these drugs, are strong limitations for this family of compounds. New drugs are therefore needed with new mechanisms of action.
Metal complexes of active antiamoebiasis drugs have been investigated including Au I , Ru II and Cu II complexes of metronidazole ([1-(2-hydroxyethyl)-2-methyl-5-nitro-1H-imidazole]) which have higher activity than uncomplexed metronidazole. 378 Cyclooctadiene Ru II complexes of thiosemicarbazone derivatives are also more active than metronidazole. 379 Recently, an automated, high-throughput screen has been developed to facilitate drug screening for Entamoeba histolytica. 380 Interestingly, this screening identified that auranofin is 10Â more potent against E. histolytica than metronidazole. The capability of this drug to inhibit E. histolytica thioredoxin reductase prevents the reduction of thioredoxin and enhances the sensitivity of trophozoites to reactive oxygen-mediated killing. This new use of auranofin represents a promising therapy for amoebiasis, and the drug has been granted orphan-drug status by the FDA (Table 1).

Anti-inflammatory metallodrugs
3.2.1. Arthritis. Chrysotherapy (utilisation of metallic gold and its complexes in medicine) has been used for the treatment of various diseases for many centuries. For instance, Au I cyanide was used to treat tuberculosis in the early 20th century and was replaced by Au I thiolates such as the polymer aurothiomalate. French physicians first used injectable Au I compounds to treat rheumatoid arthritis in 1929. 381 In 1985 the Au I triethylphosphine complex auranofin (Fig. 11) was approved by the FDA as an oral antiarthritic agent. The mechanism of action of this compound is still not understood, 382 although recent studies have shown that cathepsin B is inhibited by Au I complexes. 383 Auranofin is also highly cytotoxic to cancer cells and clinical trials for certain cancers are now in progress (Table 1).

Ulcers.
Similarly to gold, utilisation of bismuth is traditional, both in Chinese and occidental medicines. Dyspepsia, syphilis, colitis, wound infections and quartan malaria are examples of diseases in which bismuth is effective. Yang and Sun comprehensively reviewed the biological chemistry of bismuth in 2011. 384 However, the use of bismuth in therapy has found its main application for the inhibition of Helicobacter pylori, 385,386 a bacterium that can prevent ulcers from healing. 387 Due to the development of antibiotic drug resistance by H. pylori, triple 388 and quadruple 389 regimens based on two antibiotics, plus an acid-suppressing agent, plus bismuth salts are nowadays investigated for the treatment of this bacterium. Time-resolved ICP-MS studies of the uptake of bismuth-based drugs suggest a competitive Fe III /Bi III transport pathway into H. pylori. 390 Homoleptic tri-substituted bismuth(III) sulfonate complexes possess interesting bactericidal activity towards three laboratory strains of H. pylori (B128, 26 695 and 251), with minimum inhibitory concentration (MIC) values in the nanomolar range.
In the blood, the Fe III binding site of transferrin is also a strong site for Bi III . 391 The recent 2.4 Å X-ray crystal structure of human transferrin with Bi in the N-lobe and Fe in the C-lobe shows Bi bound in a partially-opened cleft via only one of the two binding cleft Tyr side-chains together with nitrilotriacetate, carbonate and water. 392

Neurodegenerative diseases
Neurodegenerative diseases (NDs), the progressive loss of structure or function of neurons -are attracting much Chart 12 Molecular structures of chloroquine, ferroquine, and ruthenoquine. The intramolecular hydrogen bond facilitating membrane permeability of ruthenoquine is shown. attention due to the global increase of life expectancy in modern societies; 35 million people currently live with dementia; Alzheimer's disease (AD), Parkinson's disease (PD), and prion diseases (PrDs) are the most prevalent NDs. 393 However the lack of knowledge of the origins, mechanisms, and development factors for these diseases results in poorly efficient drugs, which treat symptoms at best (for ADs and PDs, but no medication available for PrDs), 394 and do not reverse or slow down disease progression.
Senile plaques, neurofibrillary tangles, neutrophil threads, amyloid-b peptide (Ab) deposition, selective loss of neurons and decreased synaptic density in post-mortem brains are the pathognomonic indicators of Alzheimer's disease. Aberrant metal biochemistry in the pathogenesis of AD has been demonstrated. 395,396 Related to the synaptic activity, Cu and Zn are of importance, since millimolar concentrations of Zn are released upon neuronal activation, whilst Cu is involved in the regulation of synaptic functions. Amyloid-b peptide (Ab) is released upon neuronal activation, and Ab deposition and oxidative stress may be attributed to interactions between Ab and metal ions. Bush et al. have recently reported the post-hoc analysis of a Phase IIa double-blind, randomised, placebo-controlled clinical trial for a copper/zinc ionophore, PBT2, 397 an hydroxyquinoline derivative that facilitates the clearance of Ab aggregates in the cortex by targeting the zinc and copper ions that mediate the assembly of these aggregates in amyloid and diffuse deposits, effectively detoxifying the Ab. 398 The output of this study shows clear improvement for treated patients compared to placebo group. 397 In 2012, the bioinorganic chemistry of Alzheimer's disease was surveyed by Kepp,399 and the role that metal-Ab association species play in AD was comprehensively reviewed by Pithadia and Lim. 400 Inhibition of the interactions between Ab and metal ions is a promising therapeutic approach for developing new and effective anti-AD agents. This inhibition may be achieved by utilisation of competitive agents (for occupying the metal binding site on Ab), or by using chelating agents. Chelation therapy is a powerful tool for metal depletion and excretion and has been extensively studied in recent years for the treatment of Wilson's disease and neurodegenerative diseases. 401,402 Metal chelators, especially inhibitors of histone deacetylases (HDACs), are also of much interest in anticancer therapy, as exemplified by suberoylanilide hydroxamic acid (SAHA, vorinostat, Zolinza) which inhibits Zn(II)-dependent class I and class II histone deacetylases (HDACs). Histone acetylation plays a key role in controlling the affinity of histones for DNA and gene expression. Deacetylation of Lys on histones produces a positive charge on its side chain and increases DNA affinity.
In 2013, Telpoukhovskaia and Orvig, surveyed how coordination chemistry might play a role in anti-neurodegenerative drug development, by understanding the binding preferences of metal ions for key proteins involved in the propagation of NDs, and by using the tools of inorganic chemistry for investigating the competition of synthetic ligands with proteins for metal ions. 403

Diagnostic and therapeutic radiopharmaceuticals
There is an increasing interest in the development of both diagnostic and therapeutic radiopharmaceuticals (Fig. 1), as illustrated by the large number of ongoing clinical trials worldwide ( Table 1). The development of radiopharmaceuticals is aided by their rapid passage from the laboratory into the clinic, since very small doses are usually administered, posing a negligible toxicity hazard.
Therapeutic radiopharmaceuticals are useful for delivering locally cytotoxic doses of ionising radiation. The radionuclides used emit b À -particles (electrons) or a-particles (used in targeted a-therapy (TAT)). Most radiotherapeutic nuclides in the clinic are b À emitters, such 32 P, 47 Sc, 64 Cu, 67 Cu, 89 Sr, 90 Y, 105 Rh, 111 Ag, 117m Sn, 131 I, 149 Pm, 153 Sm, 166 Ho, 177 Lu, 186 Re, 188 Re. 404 Recent advances in this field include the first clinical trial using an a-particle emitting 225 Ac complex, labelled with a humanised antibody, lintuzumab, which targets the CD33 antigen expressed on the blast cells of most cases of acute myeloid leukaemia. The phase I clinical trial was initiated by the Scheinberg group at Memorial Sloan-Kettering Cancer Center with a primary goal to define both safety and the maximum tolerated dose of 225 Ac TAT in patients with advanced myeloid leukaemia (AML) through a dose escalation series. 405 Promising results have led to the ongoing phase I/II clinical trial, sponsored by Actinium Pharmaceuticals (Table 1). This complex is also undergoing a phase I clinical trial for the treatment of leukaemia myelodysplastic syndrome. The Scheinberg group reported the first proof-of-concept 213 Bi TAT clinical trial, again targeting CD33 with antibody lintuzumab to treat 18 patients with advanced myeloid leukaemia in a phase I trial. 406 This compound is now undergoing a phase II clinical trials (Table 1). Jurcic et al. also conducted a follow up study with 213 Bi TAT, wherein 13 newly diagnosed patients and 18 patients with relapsed/refractory acute myeloid leukaemia were first treated with continuous cytarabine (cytosine arabinoside) infusion for 5 days. 407 They observed marrow blast reductions at all dose levels.
Two types of radioimaging are used in clinic: single-photon emission computed tomography (SPECT), and positron emission tomography (PET). SPECT is based on the utilisation of pharmaceuticals labeled with a g-emitting radionuclide, while PET requires a radiopharmaceutical labeled with a positron b + -emitting radionuclide. Useful g-emitting nuclides include 99m Tc, 67 Ga, 111 In, and 201 Tl; useful b + -emitting nuclides include 55 Co, 64 Cu, 66 Ga, 68 Ga, 82 Rb, 86 Y. 404 Recent clinical developments of such radioimaging agents include for instance 68 Ga-DOTA-TATE 408 and 64 Cu-ATSM for PET/CT scans 409 (Table 1, and Chart 13 for the molecular structure of 68 Ga-DOTA-TATE).
Multimodality imaging (positron emission tomographycomputed tomography (PET-CT)) has been used in clinic for more than a decade, and is nowadays one of the main cancer imaging techniques. 410 Following this clinical success, combined PET/magnetic resonance (MR) systems for clinical use have been recently developed. Ga-DOTATOC PET/MR imaging is feasible in patients, with a good diagnostic image quality (average visual rating PET/CT, 2.83; PET/MR, 2.08). Moreover, detectability of focal PET lesions is equivalent to PET/CT on a patient basis and organ-system basis. 411 The clinical value of Ga-DOTATOC PET/MR with additional diagnostic MR protocols needs to be evaluated against PET/CT with multiphase contrastenhanced CT protocols in future studies.

Conclusions
The exploration of the medical periodic table presents exciting challenges. Inorganic compounds, and metal complexes in particular, offer mechanisms of drug action that can be quite distinct from those of organic drugs. About 13 metal ions are essential for mammalian life, which can all be used in therapy. However, not only can essential metals be used, but also nonessential metals, a strategy that might be particularly important for fighting bacteria which are becoming increasingly resistant to organic drugs. Also radionuclides with suitable ligands offer targeted agents both for diagnosis and therapy. The ligands can play critical roles in the activity of all metallodrugs and diagnostic agents. It is important to identify the parts of metallodrugs which are critical for activity (the pharmacophores).
We have tried to focus here particularly on the discovery of new targets for metallodrugs. These days an understanding of targets and mechanisms of action is essential if a drug is to receive approval for clinical use. Such understanding will eventually become important when patients can be screened on a personal-medicine basis for the optimum drug to treat their particular conditions. There are signs that rapid progress is currently being made, aided by advances in metal analysis and especially speciation techniques, and by the methods of modern molecular biology (proteomics and genomics), which can be fruitfully applied to the identification of target sites and understanding of metabolic pathways.
Although many new targets are currently being proposed for metallodrugs, and provide promising new leads for development, few have been validated in vivo. Advances in controlling metallodrug activation are also needed so that metallodrugs reach target sites and do not undergo unwanted side-reactions. From a general point of view, the field will benefit greatly from the application of the drug design principles used for the development of organic drugs. In this sense, we have tried to highlight some recent and promising strategies. Specific G-quadruplex binders, metallodrugs with DNA sequence specificity, inert polypyridyl metal complexes specifically able to target kinases, catalytic metallodrugs, and multi-labeled nanoparticle drug delivery systems are striking examples of the new strategies that can be used for designing metal-based anticancer drug candidates. The potential of inorganic and organometallic complexes as antibiotics, and for the treatment of other diseases, such as viral and parasitic diseases is also apparent.
We also have highlighted some organic drugs which are targeted to metal ions, often to the metal in a metalloenzyme. Organic drugs are often designed so they contain H-bond acceptors (lone pair donors). These acceptor atoms are also potential metal binding sites but perhaps not always recognised as such.
There is an urgent need to find new and effective therapies for a wide range of diseases and conditions. As highlighted in Table 1, there is current clinical and industrial interest in such advances.