MDPI and ACS Style
Carraher, C.E.; Roner, M.R.; Frank, J.; Moric-Johnson, A.; Miller, L.C.; Black, K.; Slawek, P.; Mosca, F.; Einkauf, J.D.; Russell, F. Synthesis of Water-Soluble Group 4 Metallocene and Organotin Polyethers and Their Ability to Inhibit Cancer. Processes2017, 5, 50.
Carraher CE, Roner MR, Frank J, Moric-Johnson A, Miller LC, Black K, Slawek P, Mosca F, Einkauf JD, Russell F. Synthesis of Water-Soluble Group 4 Metallocene and Organotin Polyethers and Their Ability to Inhibit Cancer. Processes. 2017; 5(3):50.Chicago/Turabian Style
Organometallic Anticancer Compounds
Gilles Gasser,*†§Ingo Ott,*‡ and Nils Metzler-Nolte*§
†Institute of Inorganic Chemistry, University of Zurich, Winterthurerstrasse 190, CH-8057 Zurich, Switzerland
‡Institute of Pharmaceutical Chemistry, Technische Universität Braunschweig, Beethovenstrasse 55, 38106 Braunschweig, Germany
§Chair of Inorganic Chemistry I, Bioinorganic Chemistry, Faculty of Chemistry and Biochemistry, Ruhr University Bochum, Universitätsstrasse 150, 44801 Bochum, Germany
*To whom correspondence should be addressed. For G.G.: phone, +41(0)44 635-4611; fax, +41(0)44 635-6803; e-mail, email@example.com; Web page, www.gassergroup.com. For I.O.: phone, +49 (0)531-391 2743; fax, +49 (0)531-391 8456; e-mail, firstname.lastname@example.org; Web page, www.pharmchem.tu-bs.de/forschung/ott/. For N.M.-N.: phone, +49 (0)234-32 28152; fax, +49 (0)234-32 14378; e-mail, email@example.com; Web page, www.chemie.rub.de/ac1.
Author information ►Article notes ►Copyright and License information ►
Received 2010 Jan 8
Copyright © 2010 American Chemical Society
This is an open-access article distributed under the ACS AuthorChoice Terms & Conditions. Any use of this article, must conform to the terms of that license which are available at http://pubs.acs.org.
J Med Chem. 2011 Jan 13; 54(1): 3–25.
Published online 2010 Nov 15. doi: 10.1021/jm100020w
This article has been cited by other articles in PMC.
The quest for alternative drugs to the well-known cisplatin and its derivatives, which are still used in more than 50% of the treatment regimes for patients suffering from cancer, is highly needed.1,2 Despite their tremendous success, these platinum compounds suffer from two main disadvantages: they are inefficient against platinum-resistant tumors, and they have severe side effects such as nephrotoxicity. The latter drawback is the consequence of the fact that the ultimate target of these drugs is ubiquitous: It is generally accepted that Pt anticancer drugs target DNA, which is present in all cells.3,4 Furthermore, as a consequence of its particular chemical structure, cisplatin in particular offers little possibility for rational improvements to increase its tumor specificity and thereby reduce undesired side effects.
In this context, organometallic compounds, which are defined as metal complexes containing at least one direct, covalent metal−carbon bond, have recently been found to be promising anticancer drug candidates. Organometallics have a great structural variety (ranging from linear to octahedral and even beyond), have far more diverse stereochemistry than organic compounds (for an octahedral complex with six different ligands, 30 stereoisomers exist!), and by rational ligand design, provide control over key kinetic properties (such as hydrolysis rate of ligands). Furthermore, they are kinetically stable, usually uncharged, and relatively lipophilic and their metal atom is in a low oxidation state. Because of these fundamental differences compared to “classical coordination metal complexes”, organometallics offer ample opportunities in the design of novel classes of medicinal compounds, potentially with new metal-specific modes of action. Interestingly, all the typical classes of organometallics such as metallocenes, half-sandwich, carbene-, CO-, or π-ligands, which have been widely used for catalysis or biosensing purposes, have now also found application in medicinal chemistry (see Figure 1 for an overview of these typical classes of organometallics).
Summary of the typical classes of organometallic compounds used in medicinal chemistry.
In this Perspective, we report on the recent advances in the discovery of organometallics with proven antiproliferative activity. We are emphasizing those compounds where efforts have been made to identify their molecular target and mode of action by biochemical or cell biology studies. This Perspective covers more classes of compounds and in more detail than a recent tutorial review by Hartinger and Dyson.(5) Furthermore, whereas recent reviews and book contributions attest to the rapid development of bioorganometallic chemistry in general,6,7 this Perspective focuses on their potential application as anticancer chemotherapeutics. Another very recent review article categorizes inorganic anticancer drug candidates by their modes of action.(8) It should be mentioned that a full description of all currently investigated types of compounds is hardly possible anymore in a concise review. For example, a particularly promising class of organometallic anticancer compounds, namely, radiolabeled organometallics, has been omitted for space limitations. Recent developments of such compounds have been reviewed in detail by Alberto.(9)
Metallocenes is the name for compounds with two π-bonded cyclopentadienyl (Cpa) ligands on a metal atom. Research into this class of compounds started in 1952 with the discovery of ferrocene (bis-cyclopentadienyl iron, Cp2Fe) and the elucidation of its C5-symmetric structure with two equivalent, π-bonded Cp rings. Because of their symmetrical structure, such compounds are also frequently referred to as “sandwich complexes”. Today, other metal complexes with cyclic π-perimeters are also sometimes named metallocenes. Compounds with only one π-perimeter are classified as “half-sandwich metallocenes”, such as the Ru(arene) complexes discussed below in some of the following sections of this Perspective. Structurally, the bis-cyclopentadienyl complexes can be classified into two classes, namely, the “classical” ones with parallel Cp rings and the “bent” metallocenes, which have other ligands bonded to the metal in addition to the Cp rings (Scheme 1). The sufficiently robust metallocenes used for medicinal applications contain metals from the iron and cobalt triad, with Fe, Ru, and Co being relevant to this article. The bent metallocenes typically comprise metals from the earlier transition metals, most importantly Ti, Zr, V, Nb, Mo in a medicinal context. Interestingly, all medicinally important bent metallocenes have a cis-dihalide motif as depicted in Scheme 1, which is similar to the cis-dichloro motif of the well-established anticancer drug cisplatin. This resemblance has spurred interest in metallocenes in the early days of medicinal inorganic chemistry, particularly through the work of Köpf and Köpf-Maier.10−12
(Top Row) Classical Metallocenes with Parallel Cp Rings (Left) and Bent Metallocenes (Right) with the Medicinally Relevant Metals Indicated and (Bottom Row) Reversible Oxidation Chemistry of Ferrocene
Historically, however, the medicinal properties of ferrocene were previously investigated because it was the first organometallic compound for which antiproliferative properties were reported.(13) This report sparked the development of organometallic anticancer compounds.14,15 Ferrocene by itself is not a particularly toxic compound. It can be injected, inhaled, or taken orally without causing major health problems. Like most xenobiotics, it is degraded in the liver by cytochromes. Because of its aromatic character, a metabolism related to benzene had been expected and was indeed found experimentally. As shown in experiments with rats that were orally given a single dose of ferrocene in sesame oil, ferrocene is enzymatically hydroxylated in the liver and urinally excreted in the form of conjugates to sulfate (minor product) and glucuronic acid as the main products.(16) In vitro, intact liver microsomes, NADPH, and molecular oxygen were found to be necessary for the hydroxylation of ferrocene. This process was inhibited in vitro by CO but significantly stimulated in vivo by pretreatment of the rats with phenobarbital. These findings give conclusive evidence that the hydroxylation of ferrocene is carried out by cytochrome P450 enzymes, similarly to benzene and many other hydrocarbons. Although such studies have hardly been carried out on other organometallic complexes, it is not unreasonable to assume that a similar fate is experienced by several of the metal-based drugs discussed in this article, at least by those who possess π-bonded (quasi-aromatic) ligands. On the other hand, hydroxyferrocene is rather unstable and decomposes in aqueous solution, finally releasing solvated iron atoms. Indeed, ferrocene derivatives have been proposed as antianemics, and one such compound, ferrocerone, was clinically approved in the USSR. To the best of our knowledge, this compound was the first marketed transition organometal drug. In this context, it is worth mentioning that another ferrocene-containing compound, which is a close derivative of chloroquine, has successfully passed clinical phase II trials as an antimalarial drug candidate (ferroquine, 1, Figure 2).17,18 This compound is now undergoing field testing and may reach approval as a new antimalarial drug in the near future. Ferroquine has an activity similar to chloroquine on the malaria parasite P. falciparum but most notably is similarly active against chloroquine-resistant P. falciparum strains. It has been discussed that changes in lipophilicity, but possibly also some redox activation, could be responsible for the unexpected activity of this ferrocene antimalarial.(18) Following the success of ferroquine, many other organometallic antimalarials were synthesized and tested but as yet with lesser success.
Ferroquine (1) is presently the most advanced organometallic drug candidate and about to enter phase III clinical trials as an antimalarial drug.
The toxicity of ferrocene was also tested in beagle dogs that were fed up to 300 mg kg−1 per day for 6 months or even 1 g kg−1 for up to 3 months.(19) While no acute toxicity or even deaths were observed, massive Fe overload was diagnosed. However, all the dogs recovered afterward. The ferrocene-induced hepatic Fe overload could be reduced after the removal of large quantities of Fe by repeated venesection.(19)
Ferrocene can undergo a one-electron oxidation, yielding the ferrocenium cation (see Scheme 1, bottom). This cation is rather stable and the redox reaction is reversible for most ferrocene derivatives. Simple ferrocenium salts were the first iron compounds for which an antiproliferative effect on certain types of cancer cells was demonstrated.(13) The mechanism of action is still uncertain. Nuclear DNA, cell membrane, and the enzyme topoisomerase II20,21 were proposed as possible targets. More precisely, Osella et al. showed that ferrocenium salts may generate hydroxyl radicals in physiological solutions.(22) An earlier report suggests that these radicals damage the DNA in a Fenton-type reaction.(23) The cytotoxic effect of decamethylferrocenium tetrafluoroborate (Cp*2FeBF4, Cp* = pentamethylcyclopentadienyl) was correlated to the production of 8-oxoguanine, the initial product of DNA oxidation. Direct evidence for hydroxyl and superoxide radicals stems from ESR and spin-trapping experiments. In one of the few studies of this kind, a synergistic effect between Cp*2FeBF4 and the iron-dependent antitumor drug bleomycin was observed.(24) Finally, the cell membrane may be the cellular target similar to the peroxidation of membrane lipids which is the consequence of excess hepatic iron. A more detailed review on the physiological chemistry of ferrocene and the antiproliferative properties of ferrocene or ferrocenium alone has recently been given elsewhere.(25)
Neuse et al. used an interesting approach to enhance the cytotoxicity of ferrocene. They bound ferrocene to polymeric supports such as poly aspartamide.26−30 The underlying idea is that enhanced water solubility may be a crucial factor for the activity of ferrocene. This assumption is confirmed by the fact that the cytotoxicity of ferricenium salts depends greatly on the nature of the counterion. Indeed, the poorly soluble heptamolybdate is inactive while ferricenium salts with good aqueous solubility such as the picrate and trichloroacetate display high antitumor activity.(13) It must be noted that smaller ferrocenyl polyamines were also tested by Brynes and co-workers at almost the same time but with limited success.(31) The work on polymer-bound ferrocene as anticancer drugs has recently been reviewed by Neuse.(32)
Numerous ferrocene derivatives have been tested for antiproliferative purposes.33−38 Among those, a ferrocene−acridine conjugate was found to be highly cytotoxic. The acridine moiety served to bring the ferrocene close to DNA by intercalation.(36) Wagner and co-workers investigated the activity of borylated ferrocenes in boron neutron capture therapy (BNCT)(34) and found that their compounds exhibit an interesting organ distribution. One derivative in particular was found to penetrate the blood−brain barrier (BBB), which is of high importance for the treatment of brain tumors. Schmalz and co-workers synthesized several nucleoside analogues of ferrocene(39) (e.g., 2 in Figure 3) with IC50 values in the low micromolar range, although not as low as the iron tricarbonyl nucleoside analogues from the same group (see also the section Metal Carbonyl Complexes below).40,41 No molecular target has been proposed so far for 2, but the structure makes protein targets, maybe in RNA/DNA synthesis or repair pathways, likely candidates.
Nucleoside analogue of ferrocene.
In more general terms, redox activity is a property that is not unique to metal compounds but frequently encountered with them. It is thus interesting to correlate the redox properties of metal compounds with electron transfer, oxidative stress, the formation of reactive oxygen species, and generally the redox status of cells.33,42 While it is difficult to determine the exact “redox potential” or even “redox status” of a whole cell, the correlation between the redox activity of metal complexes and their antiproliferative properties has been only tentatively investigated. However, a mechanism whereby redox activation induces anticancer activity in ferrocene derivatives has recently been suggested by Jaouen and co-workers.(42) They substituted phenyl rings in established drugs and natural products with ferrocene groups.(43) Most significantly, this work has uncovered a group of derivatives of the anticancer drug tamoxifen, called ferrocifens by Jaouen’s group.42,44 Tamoxifen (3 in Figure 4) is the front-line chemotherapeutic agent for patients with hormone-dependent breast cancer. Its active metabolite is hydroxytamoxifen (4 in Figure 4). In general, breast tumors can be divided into two groups depending on the presence (ER(+)) or absence (ER(−)) of the estrogen receptor. About two-thirds of all cases belong to the ER(+) type, rendering them susceptible to hormone therapy by selective estrogen receptor modulators (SERMs) such as tamoxifen and giving the patients significantly improved chances for successful treatment compared to the ER(−) group of patients.(42) The antiproliferative action of tamoxifen arises from the competitive binding to ERα subtype, thus repressing estradiol-mediated DNA transcription in the tumor tissue. Unfortunately, expression of the ERα may become down-regulated under tamoxifen treatment, whereby the drug becomes ineffective.
Tamoxifens and ferrocifens. The most active derivative of 5 with n = 4 is referred to as the ferrocifen.
It is believed that the same principle explains the activity of ferrocifens (5) against ER(+) cancer cell lines. Some SARs were derived from a group of ferrocene derivatives of tamoxifen as shown in Figure 4. Replacement of the phenyl group by ferrocene reduces receptor affinity by about 40%. Increasing the length of the dimethylaminoalkyl chain has an adverse effect on receptor binding. In addition, it also changes the bioavailability and determines whether estrogenic or antiestrogenic activity is observed in animal experiments. It appears that an optimum value is around n = 4. While the Z isomers bind more strongly to the ERα than the E isomers in in vitro tests, there is rapid isomerization under physiological conditions. Finally, ferrocifens were shown to be effective anti-estrogens in MCF-7 breast cancer cell lines (ER(+)) and against estrogen-dependent tumor xenografts in nude mice.
Surprisingly, however, compound 5 with n = 4 was shown to be active against the ER(−) MDA-MB231 tumor cell line, which lacks the ERα and is hence not susceptible to treatment with tamoxifens. This indicates a new and different mode of action for 5. Interestingly, the ruthenocene analogue of 5 also acts as an antiestrogen in ER(+) breast cancer cells but lacks the antiproliferative effect of ferrocifen against ER(−) cell lines.(45) Other organometallic fragments in place of the ferrocenyl group were also tested but were found to be inactive.44,46 This suggests a dual mode of action for ferrocifen. In addition to tamoxifen-like binding to the ERα receptor, the second pathway must critically depend on the properties of ferrocene. In an elegant study, redox activation has been proposed as the second mode of action.(47) The active metabolite hydroxyferrocifen is readily oxidized, yielding a quinone methide intermediate. This intermediate is activated for nucleophilic attack by nucleophiles. Quinone methides of the metal-free 4-hydroxytamoxifen are known to be stable for hours under physiological conditions. Adducts of such tamoxifen metabolites with glutathione and nucleobases are thought to be responsible for its general toxicity and mutagenic potential. It is now proposed that related chemistry applies to the activated ferrocifens. Extensive SAR studies48−53 in correlation with electrochemical properties52,54−56 support this hypothesis. Moreover, production of reactive oxygen species has been demonstrated in cell lines treated with ferrocifen and derivatives.(57) In this mode of action, which is summarized in Scheme 2, the metallocenes serve as a “redox antenna”. It is particularly noteworthy that redox activity of the metallocene is the key for additional biological activity that exceeds that of a purely organic analogue. Once this redox-activation mode of action was established, it is clearly not dependent on the tamoxifen-related substructure. Recently, the same group has presented work on ferrocenyl diphenols and unconjugated phenol derivatives that also have good antiproliferative activity, presumably via a related mechanism of activation and formation of similar intermediates.58,59
Redox Activation of Ferrocifens as Proposed by Jaouen and Co-Workers
In order to advance the use of ferrocifens toward clinical studies, several formulation studies were performed using nanoparticles,(60) lipid nanocapsules,61,62 and cyclodextrins.(63) Further work from the same group includes, beyond synthesis, at least a preliminary testing of the antiproliferative activity of ferrocene derivatives of several classes of compounds, i.e., curcuminoids,(64) androgen derivatives,(65) and anti-androgens derived from the nilutamide lead structure,(66) indolones,(67) and ferrocenophane polyphenols.(68)
For the bent metallocene dihalides, SARs were established for the halides, and substitution of the Cp rings11,14,15,69 and model studies of such compounds with amino acids, nucleic acids, proteins, and blood plasma were performed.(70) Titanium compounds were most active, and titanocene dichloride has even entered clinical trials.(71) Although very promising in animal models, the clinical response was not significant enough to justify continuing trials; i.e., titanocene dichloride clinical trials were recently abandoned. Furthermore, because of its decomposition and low solubility in water, there were also problems with the formulation of the drug. Earlier work investigated DNA interaction, induction of apoptosis, and topoisomerase inhibition as possible modes of action.72−74
Despite the resemblance of titanocene dichloride with cisplatin, there has never been clear evidence of a similar mode of action, i.e., binding to DNA and eventually apoptosis of the cancer cell.(4) Instead, binding of the Ti4+ cation to transferrin following complete hydrolysis of Cp2TiCl2 was proposed,(75) and even a stimulatory effect of aqueous Ti species on hormone-dependent breast cancer cells was observed.(46) From an inorganic point of view, however, the existence of simple hydrated Ti4+ cations is highly unlikely in aqueous solution at pH 7, as oligomeric species and eventually insoluble titanium dioxide will form. A recent computational study on a benzyl-substituted titanocene (titanocene Y (6), Figure 5) suggests that the Cp(R)2Ti2+ dication binds to a DNA phosphate group, with additional interactions stabilizing the binding to DNA.(76) Although this result is in accord with chemical intuition, i.e., the hard Ti cation binding to anionic oxygen atoms, it is a single point computation assuming DNA as the molecular target. No protein targets were so far considered for the bent metallocenes.
Titanocene Y (6) and the ansa-bridged derivatives titanocenes X (7) and Z (8).
The two main problems of the titanocene dihalides, i.e., poor aqueous solubility and hydrolytic instability, were both addressed in recent years by chemical synthesis. To increase aqueous solubility, amino-substituted bent metallocenes were successfully prepared.(77) The two Cp rings are covalently linked together in ansa-titanocenes, and indeed, such compounds exhibit improved hydrolytic stability.78,79 Both groups of compounds show promising biological activity. The group of Tacke has developed a versatile synthetic access to Cp-substituted bent metallocenes via the fulvene route. This approach yields unbridged (via hydrido lithiation) as well as ansa-bridged metallocenes (via carbo lithiation).(80) In vitro cytotoxicity tests were performed for several derivatives. A screen against 36 human tumor cell lines of 14 different tumor types on the derivatives 6−8 (termed titanocenes Y, X, and Z, Figure 5) revealed the p-methoxybenzyl substituted titanocene Y (6) as the most active derivative. Even more interesting, this compound showed very good activity against renal cell cancer and pleura mesothelioma cell lines, for which no effective chemotherapeutic agents are currently available.(81) Further testing of this compound was performed, including tests against freshly explanted tumors82,83 and in vivo tests against xenografted renal cancer (Caki-1),(84) prostate cancer (PC-3),(85) and breast adenocarcinoma (MCF-7) in mice.(82) Mechanistic studies particularly, but not exclusively, on titanocene Y revealed antiangiogenic effects but no myelosuppression,(86) activation of the immune system, and induction of apoptosis via caspases 3 and 7 but not caspase 8.87,88 This is a desirable combination of properties for anticancer drugs. In an obvious extension of their work and inspired by second-generation platinum drugs, the Tacke group has recently replaced the two chloride ligands on titanocene Y with carboxylate groups to yield equally active compounds with possibly even more favorable pharmacokinetics.89,90
While Cp-substituted vanadocenes, zirconocenes, and even stannocenes were recently evaluated,91−94 more in-depth research has concentrated on molybdocene derivatives as the most promising alternative to Cp2TiCl2. Again with relation to cisplatin, DNA was envisaged as the target, and in early work, several X-ray structures with the Cp2Mo fragment coordinated to nucleobases were obtained.95−97 Also early on, comparative hydrolysis studies of several bent metallocene dihalides were performed that identified Cp2MoCl2 as one of the most stable simple metallocenes. Unlike Cp2TiCl2, the Cp rings in the Mo derivative are less prone to hydrolysis. Furthermore, extensive spectroscopic studies, mainly by 1H and 31P NMR, were carried out in solution to assess the binding mode of molybdocene dichloride with DNA.96,98−100 In more recent work, Harding and co-workers investigated cellular uptake and intracellular localization of several bent metallocenes dihalides by X-ray fluorescence.101,102 Only low levels of Ti and V were detected inside cells, and only Mo seemed to accumulate in significant amounts in the cellular nuclei (Figure 6). Together, these findings agree well with the notion that all metallocenes have a different biological profile.
Intracellular distribution of Mo (from Cp2MoCl2), Ti (from Cp2TiCl2), and K as shown by X-ray fluorescence microscopy. Reproduced from ref 102, with kind permission of Springer Science + Business Media.
Organometallic Ruthenium Half-Sandwich Complexes
The idea of using ruthenium-containing organometallics as anticancer agents was first developed by Tochter et al.(103) before being intensively investigated in the Sadler and Dyson research groups. It was initially anticipated that the binding of all ruthenium compounds to DNA was the main reason for their anticancer effect, similar to the platinum derivatives; i.e., the coordination of the metal center to DNA causes structural modifications, which would ultimately lead to the induction of apoptosis. Indeed, the ability of ruthenium complexes to bind to DNA or model compounds has been amply demonstrated,104−114 although it was found that the actual DNA binding of certain ruthenium compounds was weaker than or/and different from the one observed for platinum derivatives.106,115,116 But recent studies for a series of ruthenium anticancer compounds revealed that DNA is not always the primary target and that these species were actually binding more strongly to proteins than to DNA.117−119 These findings clearly indicated the occurrence of significantly different modes of action, depending on the type of ruthenium complexes. However, the exact mechanism by which these metallodrugs exert their effects has not (yet) been fully understood. Nonetheless, in this section, we will highlight recent developments on the elucidation of the mechanism of action of anticancer ruthenium half-sandwich organometallic compounds, as well as the exact role of the metal center. A nonexhaustive catalogue of ruthenium organometallic antitumor agents can be found in recent reviews or book chapters.5,120−123 We will use structure comparisons to explicit the mechanism differences/analogies of these compounds.
At a first glance, the structural similarity of the half-sandwich “piano stool” type organometallics presented in Figure 7 might suggest an analoguous mechanism of cytotoxic action. However, to the best of our current knowledge, they appear to be much different.
(η6-Arene)ruthenium anticancer complexes.
Salder et al. established that the mechanism of action of their compounds [(η6-arene)Ru(en)(Cl)]+ (en = ethylenediamine) (Figure 7, left) has many of analogies to that of cisplatin. It first involves hydrolysis of the Ru−Cl bond of the prodrug to generate an active [(η6-arene)Ru(en)(H2O)]2+ species. Detailed kinetic studies showed that the Ru−Cl bond hydrolysis can be strongly influenced by the nature of the coligands as well as the nature of the metal ion (see also the section Organometallic Osmium Half-Sandwich Complexes below).121,122 Importantly, this step is suppressed in the blood because of the high chloride concentrations enabling [(η6-arene)Ru(en)(Cl)]+ to cross the cell and nuclear membranes. Once inside the cell, the hydrolysis of the chloro anion takes place because of the much lower chloride concentration (∼25 times lower). It is then assumed that the aqua complex [(η6-arene)Ru(en)(H2O)]2+ binds to nuclear DNA with a high affinity for the N7 position of guanine bases as shown by NMR and X-ray crystallographic studies and transcription mapping experiments.111−113 It must be pointed out that the analogy in the mode of action between [(η6-arene)Ru(en)(Cl)]+ and cisplatin stops at this point. Indeed, the Ru arene compounds can only form monofunctional adducts compared to cisplatin which is known to form bifunctional adducts and DNA cross-links. Importantly also, [(η6-arene)Ru(en)(Cl)]+ derivatives were found to be active against cisplatin-resistant cell lines, indicating that the detoxification mechanism is different from the one of cisplatin.(124) However, in silico calculations undertaken by Deubel et al. to compare the difference in selectivity of cisplatin to organometallic ruthenium complexes toward biological targets show that organometallic ruthenium anticancer complexes are more similar to cisplatin than to inorganic Ru(II) complexes.(125)
Ru-RAPTA derivatives were originally designed to improve the aqueous solubility (pta = 1,3,4-triaza-7-phosphatricyclo[18.104.22.168]decane, Figure 7). As for Ru(II) arene ethylenediamine compounds, RAPTA derivatives(126) containing two chloride ligands were also found to be susceptible to hydrolysis and it was first anticipated that DNA was a primary target.(127) Dyson et al. recently prepared RAPTA carboxylato derivatives (oxalo-RAPTA-C and carbo-RAPTA-C, Figure 8).(128) This work was evidently inspired by the structures of carboplatin and oxaliplatin. In analogy to the Pt compounds, it was assumed that the carboxylato ligands would hydrolyze more slowly and in a more controllable way than the chloride ligands in the original RAPTA-C compound. These RAPTA derivatives have an in vitro activity similar to that of RAPTA-C. All evidence taken together, RAPTA compounds seem to operate by a different mode of action compared to cisplatin, Ru(II) arene ethylenediamine compounds, and most of the known anticancer compounds in general. In vitro cytotoxicity studies showed that these compounds were much less cytotoxic than cisplatin. Indeed, many of the RAPTA compounds could not even be classified as cytotoxic and were also nontoxic to healthy cells. The extent of this nontoxicity was proven in an in vivo study when healthy mice were treated at quite high doses with RAPTA compounds without triggering toxic side effects.(129) But the most striking result observed was that both RAPTA-C and RAPTA-T inhibited lung metastasis in CBA mice bearing the MCa mammary carcinoma (the number and weight of the metastases were reduced) while having only mild effects on the primary tumor.(2) The only other metallo drug candidate displaying this outstanding behavior is imidazolium trans-[tetrachloro(dimethylsulfoxide)(1H-imidazole)ruthenate(III) (NAMI-A).2,118 This discovery is of high practical interest, as the removal of the primary tumor by surgery is frequently an efficient procedure while the treatment options for metastases are quite limited.(2)
Nonetheless, these very exciting findings engendered naturally a new and obvious question: If DNA is not the target for these RAPTA derivatives, then what is the target? The final answer has not yet been determined, but at this stage of the research, enzyme binding is the most probable explanation. It was shown by mass spectrometry that RAPTA compounds form adducts with proteins(130) and that the reactivity of RAPTA-C and cisplatin in the presence of proteins was much different.(131) To get more insight, Messori et al. studied the inhibition activity of a series of RAPTA compounds to two proteins, i.e., cathepsin B (Cat B) and thioredoxin reductase (TrxR), which are possible targets for anticancer metallodrugs.(132) They found that all tested Ru compounds were inhibitors of Cat B while none of them, with the exception of RAPTA-C, was inhibiting TrxR. Computer docking experiments validated this finding. Assuming that one of the two chloride ligands of the RAPTA derivatives was first replaced by a water molecule, it was then found that the Ru(II) center coordinates to the active site cysteine residue. Furthermore, other atoms of RAPTA (chloride, nitrogen of pta, etc.) bind other amino acids of Cat B, thereby stabilizing the metallodrug−enzyme complex.(132) Interestingly, a good correlation was observed between the inhibiting potency of the RAPTA derivatives and the calculated stability of the corresponding Cat B/RAPTA adducts.(132)
Other proteins have been proposed as the target for Ru organometallics. P-Glycoprotein (Pgp) is a plasma membrane protein that is responsible for drug efflux from cells and is involved in multidrug resistance (MDR).(120) Inhibitors of Pgp, namely, phenoxazine and anthracene derivatives, were synthetically modified and coordinated to Ru organometallics.(133) The aim was to obtain a synergistic effect by combining the selectivity of ruthenium complexes toward cancer cells and the ability of the phenoxazine and anthracene derivatives for Pgp inhibition. These newly formed complexes were found to be, in general, more cytotoxic and inhibited to a lesser extent the Pgp protein than the original Pgp inhibitor derivatives used as ligands. Interestingly, for one of these ruthenium derivatives (13, Figure 9), it was shown that the ruthenium coordination to the Pgp inhibitor derivative induced an even stronger protein inhibition. Furthermore, because of the presence of the fluorescent anthracene group, it was observed that 13 was accumulating in cell nuclei, suggesting a DNA synthesis inhibition as the mechanism of cytotoxic action. Nonetheless, because of the strong increase in cytotoxicity upon ruthenium coordination, Dyson et al. believe that their organometallic compound not only inhibits the enzyme but also induces cell death via a second mechanism, implying a bifunctionality of this compound.(133)
In a similar line of thought, namely, a dual cytotoxic mode of action, ethacrynic acid (EA) was coupled to two RAPTA derivatives (15 and 16, Figure 9)(134) as well as to other Ru arene organometallics.(135) EA is an effective glutathione transferase (GST) inhibitor, which has been investigated as a potential anticancer drug. EA is known to bind competitively to the hydrophobic cosubstrate (H-site) of GST, while the RAPTA compounds are recognized to react with soft nucleophilic centers such as thiol groups (see above).(134)15 and 16 compounds were therefore thought to be able to bind not only to the enzyme at the H-site but also to interact with the reactive cysteine residues of GST P1-1 (this GST protein possesses two solvent-accessible cysteine residues that affect catalytic activity when modified). As assumed, these two new compounds were found to bind the catalytic H-site in a similar fashion as EA. Furthermore, the inhibition constants Ki of the complexes on GST P1-1 were 3 or 4 times lower than EA. The authors therefore concluded that the ruthenium centers were also involved in the inhibition of GST P1-1. Interestingly, it was demonstrated by X-ray crystallography and by ESI-MS that 16 decomposed, over a period of time, into a ruthenium derivative and EA. It is anticipated that the cleavage occurs, by virtue of a possible allosteric effect or simply over time, when the EA moiety of 16 is bound to the H-site. Importantly, this (selective?) release of the ruthenium moiety should enhance the toxic effect of the compound on cancer cells, which had already been sensitized by the EA moiety that inactivated GST.(134) This feature could be used to specifically deliver a cytotoxic payload for targeted chemotherapy.(134)
Still in a metal−drug synergism context, the group of Sánchez-Delgado coupled different Ru arene complexes to chloroquine (CQ), which is known to be an effective antimalarial compound as well as having anticancer properties (see the section Metallocenes above).(136) Contrary to the ferroquine mentioned above,17,137 in which the ferrocenyl moiety is nontoxic (or at least commonly assumed to be), these compounds are made of two toxic moieties. The compounds had a consistently higher potency against CQ-resistant parasites than the standard drug chloroquine diphosphate (CQDP). In addition, two of their compounds (17 and 18, Figure 10) inhibit the growth of two HCT-116 colon cancer cell lines with IC50 values between 20 and 35 μM. They also observed that liposarcoma cell lines were especially sensitive to 17 with an IC50 value of 8 μM. This is of clinical interest, as this type of tumor does not respond to currently employed chemotherapies.(136)
Ru arene chloroquinone antimalarial and antitumor agents.(136)
Other proteins have been shown to be the target of cytotoxic ruthenium organometallics. For example, Sheldrick et al. have used an automated multidimensional protein identification technology (MudPIT), which combined biphasic liquid chromatography with electrospray ionization tandem mass spectrometry (MS/MS) to analyze tryptic peptides from Escherichia coli cells, which were first treated with [(η6-p-cymene)RuCl2(DMSO)] (14, Figure 9).(138) They showed that five proteins, namely, the cold-shock protein CspC, the three stress-response proteins ppiD, osmY, and SucC, and the DNA damage-inducible helicase dinG were the target of their Ru arene compounds.(138) Using electrophoretic mobility shift assays, Brabec, Sadler and co-workers also examined the binding properties of the mismatch repair (MMR) protein MutS in Escherichia coli with various DNA duplexes (homoduplexes or mismatched duplexes) containing a single centrally located adduct of Ru(II) arene compounds.(139) They showed that the presence of the Ru(II) arene adducts decreased the affinity of MutS for ruthenated DNA duplexes, which either had a regular sequence or contained a mismatch, and that intercalation of the arene contributed considerably to this inhibitory effect.(139)
Interestingly, it was recently demonstrated that iodo-containing Ru(II) arene organometallic derivatives 19 and 20 (Figure 11) were highly cytotoxic to human ovarian A2780 and human lung A549 cancer cells. Although these complexes are remarkably inert toward ligand substitution. No hydrolysis was observed by NMR and ESI-MS.(140) Fluorescence-trapping experiments in A549 cells suggested that this potency arose from an increase in reactive oxygen species (ROS). Surprisingly, these Ru complexes act as catalysts for the oxidation of the tripeptide glutathione (GSH), which is a strong reducing agent present in millimolar concentrations in cells. Indeed, millimolar amounts of GSH were oxidized to glutathione disulfide in the presence of micromolar ruthenium concentrations! The same group showed that the anticancer complex [η6-bip)Ru(en)Cl]+21 (bip = biphenyl) (Figure 11) readily reacts with GSH, at pH 7, in a typical cytoplasmic concentration of chloride and at Ru concentrations relevant to cytotoxicity to give a thiolato complex [η6-bip)Ru(en)(GS-S)] as the major product.(113) Unexpectedly, this complex is very sensitive to air and is oxidized to the sulfenato complex [η6-bip)Ru(en)(GS(O)-S)]. However, under physiologically relevant conditions, competitive reaction of complex 21 with GSH and guanosine 3′,5′-cyclic monophosphate (cGMP) gave rise to a cGMP ruthenium adduct, accounting for 62% of total Ru, even in the presence of a 250-fold molar excess of GSH.(113) This suggests that oxidation of coordinated glutathione in the thiolato complex to the sulfenate provides a facile route for displacement of S-bound glutathione by the guanine N7 atom, a route for RNA and DNA ruthenation even in the presence of a large excess of GSH.(113)
Structures of catalytically active organometallic anticancer complexes.
It must be pointed out that Sadler et al. recently unambiguously identified diruthenium and tetraruthenium clusters when complex 21 was reacted with GSH, using nanoscale liquid chromatography Fourier transform ion cyclotron mass spectrometry combined with 18O-labeling.(141) The same group also recently observed the binding of Ru arene compounds with human serum albumin by means of mass spectrometry combined with trypsin digestion, specific side chain modifications, and computational modeling.(142) Finally, a loss of cytotoxic activity of Ru arene organometallics upon oxidation of their amine ligand such as o-phenylenediamine (o-pda) to the corresponding imine ligand o-benzoquinonediimine (o-bqdi) (Figure 12) was observed.(143) For example, the IC50 values against A2780 ovarian cancer cells of 22 is 11 μM while compound 23 displays a value above 100 μM. Interestingly, the o-bqdi complexes can be reduced by GSH but readily undergo reoxidation in air.
Ru arene anticancer complexes with redox-active diamine/diimmine ligands.
Dyson et al. recently described the preparation of a series of RAPTA-type complexes with fluoro-substituted η6-arene ligands.(144) Electron-withdrawing fluoro or CF3 units were added to the arene to modulate the pKa values of the complexes. The activity of these organometallics was found to be strongly influenced by the presence of the substituents. IC50 values of the fluoro compounds were in general much lower than those of the nonfluorinated analogues.
Organometallic Osmium Half-Sandwich Complexes
Probably because of its reputation of being highly toxic (see OsO4) and relatively inert toward substitution, osmium organometallics have been neglected as therapeutic agents in comparison to its lighter congener ruthenium.121,145 Recently however, several research groups146−154 investigated the anticancer activity of osmium(II) arene half-sandwich complexes. Their results indicate that Os(II) organometallics might be promising candidates as antitumor drugs.
The kinetic and thermodynamic properties found for the first Os complexes were unsatisfactory despite the fact that they were isostructural to the active Ru complexes.(121) Concomitantly, the biological activity was low. SARs were then established and then used to improve their activity. The ligand exchange rate (chloride against water, see discussion above in the section Organometallic Ruthenium Half-Sandwich Complexes) was too slow (40−100 times slower than the Ru analogues, depending on the pH value). The water bound to the osmium is indeed more acidic by 1−2 pKa units than when bound to analogous ruthenium complexes. In order to restore activity, a new series of Os complexes were prepared in which the neutral N,N-chelate ethylenediamine was replaced by O,O and N,O anionic chelating ligands with a stronger trans effect. Following this thought, picolinate was identified as an ideal ligand candidate and the respective complexes [(η6-arene)Os(pico)Cl] (24, Figure 13) had faster hydrolysis rates and potent anticancer activity comparable to that of carboplatin.(148) Furthermore, their mechanism of action is thought to be similar to that of their Ru organometallics, nuclear DNA being the biological target as shown by studies demonstrating the binding of such complexes to DNA.(150)
Arion, Keppler, and co-workers investigated the binding of ruthenium and osmium to paullone derivatives, which are known to be potent inhibitors of cyclin-dependent kinases (CDKs) (25, Figure 13).(152) This “metalation” was thought to increase the solubility and bioavailability of the paullone ligands. They showed that complexes such as 25 had respectable antiproliferative activity in submicromolar to very low micromolar concentrations in three cell lines, with no significant differences between the Os and Ru complexes. No CDK inhibition was published so far on those compounds, and their binding to 5′-GMP was found to be significantly different depending on the complexes. This indicates that they exert their anticancer activity either by binding to crucial proteins or by noncovalent DNA interactions.(152)
Dyson et al. have also evaluated the activity of Os(II) and Rh(III) analogues of RAPTA-C (26, 27, Figure 13).153,154 Depending on the cell lines, significantly different IC50 values were determined for 26, 27, and RAPTA-C, with 27 being more cytotoxic than the two other compounds and 26 exhibiting essentially similar cytotoxicity as RAPTA-C.(154) Furthermore, using a combined experimental and theoretical approach, it was reported that the binding of RAPTA-C and 26 to a 14-mer oligonucleotide was nonspecific contrary to cisplatin, indicating a different mechanism of action and/or a different biological target.(153)
Telomerase is a ribonucleoprotein with DNA polymerase activity that maintains the length of telomeric DNA by adding hexameric units to the 3′ single strand terminus. It is therefore a crucial enzyme for cancer progression. Rosenberg, Osella, and co-workers investigated telomerase inhibition by a series of water-soluble cyclometalated benzoheterocycle triosmium clusters (28−31, Figure 14).(155) Their motivation was that quinoline derivatives had shown interesting biological properties, especially in inhibiting enzymes.(156) Among all compounds, only the negatively charged clusters (by virtue of the sulfonated phosphines) exhibited good activity as telomerase inhibitors when tested on semipurified enzymes in a cell-free assay. However, they were ineffective in vitro on Taq, a different DNA polymerase. Furthermore, none of the osmium clusters decreased the telomerase activity in the MCF-7 breast cancer cell line, as observed by the telomeric repeat amplification protocol (TRAP assay). This may well be due to the low aptitude of these organometallics to cross the cell membrane. However, all compounds were acutely cytotoxic, probably because of their accumulation on cell membranes, as shown for compound 29a by inductively coupled plasma mass spectrometry (ICP-MS). It was hypothesized that 29a interfered with the normal trafficking and functions of the membrane. Gobetto, Rosenberg, and co-workers also investigated the interaction of other positively and negatively charged triosmium carbonyl clusters with albumin, using the transverse and longitudinal relaxation times of the hydride resonances as 1H NMR probes of binding to the protein.(157) Evidence of binding was observed for both the positively and negatively charged clusters. However, they exhibit distinctly different rotational correlation times.(157) It was anticipated that the negatively charged clusters bind more tightly than their positive analogues, as albumin is rich in positively charged amino acids.(157) The same researchers also established guanines as the binding sites for another positively charged water-soluble benzoheterocycle triosmium cluster to single- and double-stranded DNA by using a range of different biochemical methods.(158)
Triosmium clusters as potential inhibitors of telomerase enzyme. Ligand sites on Os denoted by (−) indicate CO ligands.
It is worth mentioning that Os3(CO)9 type clusters and dicobalt carbonyl fragments were also reacted with derivatives of tamoxifen, a widely used drug in the treatment of hormone-dependent breast cancer (see also the section Metallocenes above).(159) The organometallic moiety was found to increase the lipophilicity and reduced the affinity, via steric hindrance, for the estrogen receptor, but no cytotoxicity studies were carried out on the compounds.
Organometallic Iridium and Rhodium Complexes
In contrast to their Ru(II) congeners, the isoelectronic Rh(III) and Ir(III) half-sandwich compounds have attracted much less attention as potential anticancer agents.154,160−162 But interestingly, among the few examples reported in the literature, different biomolecules were reported as (potential) targets. Nevertheless, even though it was shown that these compounds were indeed targeting the desired biomolecules, their exact mode of cytotoxic action is still unknown. Hence, Sheldrick et al. showed that Ir(III) and Rh(III) complexes such as 32 (Figure 15) bind DNA through intercalation of their polypyridyl ligands.160,161 Polypyridyl-containing half-sandwich complexes with Ru(II) and Rh(III) central atoms and hexamethylbenzene or pentamethylcyclopentadienyl ligands showed stable intercalative binding with DNA and exhibited excellent cytotoxic activities.114,161 Cellular uptake studies by atomic absorption spectroscopy (AAS) revealed that the antiproliferative effects of the complexes were mainly correlated to the size of the polypyridyl ligands, thereby highlighting the special role of ligand lipophilicity on the bioactivity of this class of organometallic antitumor drug candidates. Whereas an interaction with the DNA might significantly contribute to the cytotoxic activity of the agents, the presence of additional cellular targets or alternative modes of action is very likely to contribute to this activity and is therefore the subject of ongoing research projects.
Examples of Rh(III), Ir(III), Rh(I), and Ir(I) cytotoxic organometallic compounds.
Another example of anticancer Ir(III) compounds has been reported by Lo and co-workers. In order to design new biological probes for bovine serum albumin (BSA), Lo et al. prepared a series of luminescent Ir(III) complexes (33, Figure 15) containing an indole derivative (indole is known to bind to BSA) which were found to be highly cytotoxic toward HeLa cells.(162) It is also interesting that other cationic Ir(III) complexes have been recently reported for phosphorescence staining in the cytoplasm of living cells and were shown to be noncytotoxic.(163)
Interestingly, the heterobiorganometallic ferrocene-containing Rh(I) derivative 34 (Figure 15) had a similar cytotoxicity in prostate cancer cell lines to cisplatin but a significantly different pathway for activation of cell death.(164) While cisplatin predominantly induces apoptosis, 34 induces late necrosis and abnormal nuclear morphology.(164) This latter finding is of interest because apoptosis-resistant cells might be better killed with drugs inducing the necrotic pathway.(164) Furthermore, the same complex 34 and other related Rh(I) and Ir(I) analogues (35 and 36, Figure 15) were also found to be cytotoxic toward Chinese hamster ovary (CHO) cells with the Rh(I) complexes having IC50 values close to that of cisplatin, while the Ir(I) complex had a slightly higher IC50 value.(165) These compounds were also tested for their capacity to sensitize hypoxic CHO cells against irradiation. Indeed, tumors are notoriously hypoxic and radioresistant. Both factors limit the success of radiotherapy. Modulation of the radiosensitivity by drugs such as cisplatin is in routine clinical application.(165) Indeed, the Rh(I) complex 34 proved to be an excellent radiosensitizer with properties similar to those of cisplatin.(165)
Re organometallics are another very new class of promising antiproliferative compounds. Until recently, only few examples of cytotoxic Re complexes were described in the literature.166−168 However, over the past years, several compounds with interesting cytotoxicity were reported and their possible mode of action was explored.169−174 A nonexhaustive list of toxic Re compounds is presented in Figure 16. It is still premature to draw any definite conclusions on a molecular basis for the activity of the Re organometallics presented in this figure. However, a few targets are now envisaged. Hor and co-workers assumed that complexes such as 40 and 41 (Figure 16) were likely to bind to DNA bases or side chains of amino acid residues in peptides and proteins after displacement of the labile ligands.166,167 Other related Re compounds such as [Re2(μ-OH)3(CO)6]−, [Re2(μ-OH)(μ-OPh)2(CO)6]−, [Re2(μ-OMe)2(μ-dppf)2(CO)6]−, and [Re2(μ-OPh)2(μ-dppf)2(CO)6]− (dppf = 1,1′-bis(diphenylphosphino)ferrocene) have been shown to interfere with nucleic acid metabolism at multiple enzyme sites in L1210 lymphoid leukemia cells, causing DNA strand scission after 60 min of incubation.(168) Ma et al. observed by spectroscopic titrations and viscosity experiments that complex 39 (Figure 16) had a modest DNA binding constant and was interacting with DNA via groove binding.(169) Modeling studies suggested that the minor groove was the favored binding site.(169) Lo et al. prepared and carefully characterized a series of luminescent Re complexes such as 37, 38, and 44, which were generally highly cytotoxic.172−174 However, the exact target or mechanism of action of these compounds is unknown at this stage of the research. Recently, Doyle, Zubieta, and co-workers prepared two new fluorescent Re tricarbonyl bioconjugates, namely, a folate (43)(170) and a vitamin B12 (45)(171) conjugate (Figure 15). 43