Induction of caspase 8 and reactive oxygen species by ruthenium-derived anticancer compounds with improved water solubility and cytotoxicity.

Biochemical Pharmacology 84 (2012) 1428–1436 Contents lists available at SciVerse ScienceDirect Biochemical Pharmacology journal homepage: www.elsevier.com/locate/biochempharm Induction of caspase 8 and reactive oxygen species by ruthenium-derived anticancer compounds with improved water solubility and cytotoxicity Vania Vidimar a,d,g,1, Xiangjun Meng e,g,1, Marcelina Klajner c,g,1, Cynthia Licona a,g, Ludivine Fetzer b,g, Sébastien Harlepp c,g, Pascal Hébraud c,g, Marjorie Sidhoum f,g, Claude Sirlin b,g, Jean-Philippe Loefﬂer f,g, Georg Mellitzer a,g, Gianni Sava d, Michel Pfeffer b,g, Christian Gaiddon a,g,* a UMRS682 INSERM, Strasbourg, France UMR 7177 CNRS, Institut de Chimie, Strasbourg, France c UMR 7504 CNRS, IPCMS, Strasbourg, France d University of Trieste, Department of Life Sciences, and Callerio Foundation Onlus, Trieste, Italy e Department of Gastroenterology, Shanghai First People’s Hospital, Shanghai Jiao Tong University, China f UMRS692 INSERM, Strasbourg, France g Université de Strasbourg, Strasbourg, France b A R T I C L E I N F O A B S T R A C T Article history: Received 5 June 2012 Accepted 27 August 2012 Available online 1 September 2012 Organometallic compounds which contain metals, such as ruthenium or gold, have been investigated as a replacement for platinum-derived anticancer drugs. They often show good antitumor effects, but the identiﬁcation of their precise mode of action or their pharmacological optimization is still challenging. We have previously described a class of ruthenium(II) compounds with interesting anticancer properties. In comparison to cisplatin, these molecules have lower side effects, a reduced ability to interact with DNA, and they induce cell death in absence of p53 through CHOP/DDIT3. We have now optimized these molecules by improving their cytotoxicity and their water solubility. In this article, we demonstrate that by changing the ligands around the ruthenium we modify the ability of the compounds to interact with DNA. We show that these optimized molecules reduce tumor growth in different mouse models and retain their ability to induce CHOP/DDIT3. However, they are more potent inducers of cancer cell death and trigger the production of reactive oxygen species and the activation of caspase 8. More importantly, we show that blocking reactive oxygen species production or caspase 8 activity reduces signiﬁcantly the activity of the compounds. Altogether our data suggest that water-soluble ruthenium(II)-derived compounds represent an interesting class of molecules that, depending on their structures, can target several pro-apoptotic signaling pathways leading to reactive oxygen species production and caspase 8 activation. ß 2012 Elsevier Inc. All rights reserved. Keywords: Organometallic Anticancer Caspase 8 ROS Ruthenium 1. Introduction Cancer remains one of the ﬁrst causes of death in industrialized countries. Nevertheless, few examples of successful chemotherapy are still driving the search for more potent, more selective, less prone to resistance and better tolerated drugs. One of these examples is cisplatin, a drug showing a signiﬁcant role in the management of a number of tumors [1–3]. However, cisplatin Abbreviations: CHOP, C/EBP homologous protein; DDIT3, DNA damage inducible transcript 3; FACS, ﬂuorescence activated cell sorter; FRET, Förster resonance energy transfer; NAC, n-acetylcysteine. * Corresponding author at: U682 INSERM, 3 Avenue Molière, Strasbourg 67200, France. Tel.: +33 03 88 27 77 27; fax: +33 03 88 26 35 38. E-mail address: gaiddon@unistra.fr (C. Gaiddon). 1 Authors contributed equally to this work. 0006-2952/$ – see front matter ß 2012 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.bcp.2012.08.022 likewise platinum-derived drugs display an often-severe toxicity and a relatively frequent inactivity due to natural- or inducedresistance [4–6]. More recently, a signiﬁcant interest was addressed to ruthenium-based drugs, because of some favorable properties that make them a suitable basis for the development of antitumor drugs, such as their metal oxidation state (Ru(II) or Ru(III)), the fairly facile ligand exchange and the binding to biologically relevant proteins. Compared to many organic compounds, ruthenium-based drugs also offer the advantage of a relatively low cost for their synthesis and puriﬁcation. Various ruthenium complexes were shown to present cytotoxicity against cancer cells, ligand-exchange abilities similar to those of platinum complexes, no cross-resistance with cisplatin and a reduced toxicity against healthy tissues at least in part explained by the selective transportation to cancer cells by the iron transport system [1–3,7]. Following the pioneering work with ruthenium red V. Vidimar et al. / Biochemical Pharmacology 84 (2012) 1428–1436 [8] a number of ruthenium-based drugs were shown to endow antitumor potential. Indeed, several teams have synthesized and characterized compounds containing ruthenium in oxidative state (II) or (III), showing their anticancer activity [9–16]. Two, namely NAMI-A and KP1019, have successfully passed some initial phases of clinical trials [17,18]. Besides these partial successes, the emergence of new ruthenium-based therapies have been slowed down by limitations, such as a relatively poor level of water solubility, and/or stability in aqueous solutions, a not impressive cytotoxicity (IC50 between 20 and 100 mM), and an uncertainty on the molecular mechanisms of action responsible for the antitumor effect. The mechanism of action and the direct targets of rutheniumbased drugs are still a matter of debate. Indeed, depending on the drug, several modes of action have been proposed, such as interaction with DNA and activation of DNA damage pathways [19–24], inhibition of kinases [25], or other enzymatic activities [26,27], including extracellular metallo-proteases [28]. The differences observed may be due to variations in their structure. Even if most of the ruthenium-containing compounds have ligands that are relatively weakly bound to the metal via a heteroatom (N, O, S), there are differences in the types of ligands attached. In order to improve the stability of ruthenium complexes and possibly to enhance their cytotoxicity and their pharmacokinetics, we have previously generated several ruthenium-based complexes in which the ligand is bound to the metal via strong covalent bonds such as a C-M s bond [22,29]. Beside the stability, these compounds present differences in their Ru(II)/ Ru(III) redox potential and a new variety of ligands. We have called these molecules RDCs (Ruthenium-Derived Compounds) and we have previously shown that some RDCs are cytotoxic for several cancer cell lines resistant to cisplatin [22]. One of them, RDC11, showed a good antitumor activity both in vitro and in vivo [30] with an IC50 often between 2 and 5 mM, and anticancer properties on models of ovarian cancer, melanomas and gliomas. Importantly, they showed, compared to cisplatin a reduced neurotoxicity. As we previously showed that cisplatin exerts its neurotoxic effect in part through induction of the p53 pathway [31,32], we analyzed the ability of RDC11 to interact with DNA and induce p53 proteins. We demonstrated that RDC11 exerts its antitumor effect via DNA-dependent and DNA-independent modes of action. We also identiﬁed one of the DNA-independent signaling pathways by showing an activation of the endoplasmic reticulum stress pathway, and in particular the transcription factor CHOP/DDIT3. However, the silencing of CHOP/DDIT3 by siRNA was not able to completely abolish RDC11 cytotoxicity, strongly suggesting that other signaling pathways are also involved [30,33,34]. In the present study, we developed RDC11 variants designed for improving cytotoxicity and water solubility. The biological properties of these novel ruthenium-based organometallics are studied in vitro and in vivo. 2. Materials and methods 1429 2.2. Statistical analysis In MTT (n = 8), RT-qPCR (n = 3), ROS (n = 8), caspase 8 activity (n = 4), FACS analysis (n = 3), bars are mean SD and asterisks indicate a statistically signiﬁcant difference (p < 0.01) compared to control, as calculated by a one-way ANOVA test followed by a Tukey post-test. 2.3. Western blot Cells were treated in triplicates and Western blots were performed as previously described [37]. Immunoprobing was performed with anti-phospho 137-H2AX antibody (1/3000, Millipore, Molsheim, France), anti-CHOP (1/1000, Santa Cruz Biotechnology, CA), anti-human p53 (421, supernatant 1/3, C. Prives, Columbia University NYC), or actin (Dr. Aunis, Strasbourg) antibodies. Membranes were probed with a secondary horseradish-peroxidase-conjugated antibody (anti-rabbit, -goat or -mouse, 1/2000, Thermo Fisher, IL), and revealed with ECL (Thermo Fisher, IL). 2.4. Quantitative real-time PCR RNA was extracted using RNAII Nucleospin (Macherey-Nagel, Strasbourg, France). Reverse transcription was performed with 1 mg RNA (iScript kit, Bio-Rad), followed by qPCR in Bio-Rad iCycler thermal cycler using iQ SYBR Green supermix (Bio-Rad Laboratories, Hercules, CA). Starting quantities of genes of interest were reported to those of housekeeping genes (TBP, 18s). Speciﬁcity of the ampliﬁcation was controlled by a melting curve [31]. Probes: for noxa, 50 -ggagatgcctgggaagaag-30 ; 50 -cctgagttgagtagcacactcg-30 ; for fas, 50 -atggccaattctgccataag-30 , 50 -tgactgtgcagtccctagctt-30 ; tbp, 50 -cggctgtttaacttcgcttc-30 , 50 -cacacgccaagaaacagtga-30 . 2.5. Evaluation of mitochondrial membrane potential The changes in DCm were assessed using the membrane potential sensitive dye JC-1 (5,50 ,6,60 -tetrachloro-1,10 ,3,30 -tetraethylbenzimidazolylcarbocyanine iodide, Life technology, France). Cells were grown on cover glasses and the staining procedure was the following: DMEM was removed and 10 mg/ml JC-1 solution (in PBS) was added for 15 min [38]. The staining solution was removed and cultures were rinsed with PBS. Subsequently, cells were ﬁxed in 4% PFA for 15 min and then coved with Vectashield mounting medium (Vector labs, CA) on glass slides. Fluorescence was measured at 488 nm excitation/510 nm emission. 2.6. Assay for caspase 8 activity Caspase 8 activity was assayed by measuring the light intensity using a kit (Caspase-Glo18 Assay, Promega) and a luminometer (Perkin Elmer HTS 7000, France). Brieﬂy, cells were cultured in 96well plates in a ﬁnal volume of 200 ml. Then 50 ml caspase-8 reaction buffer was added and incubated at room temperature for 1 h before measurement. 2.1. Cell culture, MTT test, ﬂow cytometry analysis 2.7. FRET B16F10, U87, A172, 3LL, A2780, HCT116, SW480 cells were obtained from ATCC (Manassas, VA). Cells were maintained in DMEM with 10% FBS and incubated in presence of 5% CO2/95% air at 37 8C. MTT tests were performed with cells cultured in 96-well culture dishes as previously described [35]. Hypodiploid DNA was measured as described [36] using propidium iodide (Sigma, MO). The ﬂuorescence of 10,000 cells was analyzed using a FACScan ﬂow cytometer and CellQuest software (Becton Dickinson, San José, CA). FRET measurements are performed on a home build setup as previously described [34]. 2.8. Intracellular reactive oxygen species (ROS) measurement 5-(and-6)-carboxy-20 ,70 -dichlorodihydroﬂuorescein diacetate (carboxy-H2DCFDA) (Life Technology, France) was used to detect intracellular ROS according to the manufacturer’s instructions. For 1430 V. Vidimar et al. / Biochemical Pharmacology 84 (2012) 1428–1436 ROS quantiﬁcation, cells were seeded in 96-well black plates (Greiner Bio-One, France) and treated with RDCs at the indicated concentrations and times. Afterwards, cells were washed with PBS and incubated with 10 mM carboxy-H2DCFDA in DPBS for an hour. Cells were then washed and ﬂuorescence was measured by a plate reader (Perkin Elmer, France) with an excitation wavelength of 485 nm and an emission wavelength of 535 nm. 2.9. Chemical synthesis RDC11 [39], RDC34,37,40,41,44 [40] were synthesized according to published procedures. All complexes were puriﬁed over chromatography columns carried out on Merk aluminum oxide 90 standardized. Complexes were used with a purity >98%, as demonstrated by several protocols [40]. ES-MS spectra and elemental analyses were carried out by the corresponding facilities at the Institut de chimie, Université de Strasbourg and at the Service Central d’Analyse du CNRS, Vernaison. 3. Results 3.1. Generating ruthenium compounds with an IC50 in the nanomolar range On the basis of previously structure–function studies showing an improved cytotoxicity of RDCs having a phenantroline ligand [22], we substituted the two acetonitrile ligands of RDC11 by a second phenanthroline, naming the new molecule RDC34 (Fig. 1). An equivalent of RDC34 with another counter-ion (PF6) was also synthesized (RDC37). Previous works have established the ability of RDCs to modulate the activity of cellular enzymes through their redox potential [39]. We therefore modiﬁed the redox potential of RDC34 by adding either a NO2 (electron withdrawing) or a NH2 unit (electron releasing) on the phenylpyridine ligand, leading to RDC40 and RDC41 respectively. Finally, in order to increase the water solubility of RDC34, we added a spermine unit to the phenylpyridine ligand, leading to RDC44 (water solution of up to 25 mM was obtained) [40]. As colon cancers are one of the indications for platinum-derived treatments, the cytotoxicity of the RDC was ﬁrst tested on a human colon cancer cells (HCT116). Interestingly, RDC34, RDC37, and RDC40, characterized by the presence of the NO2 group, showed an increased cytotoxicity (IC50 < 2 mM) (Fig. 1) compared to RDC11 and cisplatin. However, the addition of the NH2 group decreased the cytotoxic activity (IC50 = 2–4 mM, RDC41) and that of the spermine moiety had an even worse effect with rising the IC50 to over 16 mM. Additional experiments gave IC50s for RDC34, RDC40 and RDC41 of respectively 0.25 mM, 0.75 mM and 2–4 mM (supplementary data 1). 3.2. Effects of the hydrophilic spermine substituent chain on cytotoxicity, cell uptake and DNA binding The addition of the spermine chain (leading to RDC44) signiﬁcantly decreased the cytotoxicity (Fig. 1). In order to understand the reason for this behavior, we tested RDCs’ cellular uptake. The two phenanthrolines confer to RDC34 and RDC44 luminescent properties allowing us to follow their entry into living cells (Fig. 2A and B). The maximum of accumulation of RDC34 inside the cells was reached 1 h after the compound was added. Cell uptake of RDC44 was lower. In PBS, the emitted intensity of RDC44 is 20% lower than the emission of RDC34, whereas in DMEM, RDC44 cell uptake is strongly diminished (Fig. 2C). However, the increased cell uptake of RDC44 in PBS did not lead to an increased cytotoxicity of this compound as the IC50s were closed to those observed when cells were treated in culture medium (Fig. 2D). As DNA is one of the direct targets of RDCs, we used the FRET approach based on a double-stranded oligonucleotide labeled at each end by two ﬂuorophores to test RDC/DNA binding. RDC44 showed a higher afﬁnity for DNA compared to RDC34, as demonstrated by the drop in FRET transfer energy that occurred at the lowest concentration of RDC44 (Fig. 2E). The afﬁnity constants are obtained from the analysis of these data according to the McGhee and van Hippel model [41]: Ka = 2.2 103 M1 for RDC34 and 7.8 105 M1 for RDC44. The number of base pairs occupied by RDC44 along the DNA chain is equal to 3.9, larger than the occupation site of RDC34 (2.3). It thus seems that the lack of signiﬁcant cytotoxicity of RDC44 for HCT116 cells can be attributed neither to a default in cell entry and nor to the lack of interaction with DNA. 3.3. Modiﬁcations of RDC ligands modulate the selectivity between cancer cell lines To further characterize the cellular effect of RDC34, we performed FACS analysis on HCT116 cells (Fig. 3A). RDC34, better than cisplatin, induced the accumulation of an elevated subG1 Fig. 1. Optimized RDC for water solubility and cytotoxicity reduce cell growth of HCT116 cells. HCT116 human colon cancer cells were treated for 48 h with cisplatin or RDC (mM). Viability was evaluated using a MTT test (n + 8). Insets: RDC structures. Data are representatives of 3 independent experiments. Statistical analyses are described in Section 2. V. Vidimar et al. / Biochemical Pharmacology 84 (2012) 1428–1436 1431 Kidney-derived cancer cell lines were mostly resistant toward RDC11 and RDC41. Globally, RDC34, unlike RDC11 and RDC41, was endowed with a broader range of sensitive cell isotypes. These tests also showed the negative effect of NH2 function (RDC41), leading to a signiﬁcant decrease of cytotoxicity in almost all the cell lines. However, RDC41 showed an activity better than RDC11 on EKVX cells, MDA-MB-435, UACC-257 and HS57BT (supplementary data #3). It is also interesting to stress the signiﬁcant activity of RDC11 on lung-derived cancer cells, on which RDC41 is only marginally effective. These data indicate the correlation between the modiﬁcations of the ligands around the ruthenium atom and the antitumor activity, and suggest a possibility to modulate the intensity of the cytotoxicity and the speciﬁcity for selected tumor types. 3.4. Modiﬁcations of RDC ligands change the mode of action: role of caspase 8 and ROS Fig. 2. Cell permeability and DNA interaction of optimized RDC. HCT116 cells incubated with RDC44 and visualized with white light (A) and ﬂuorescence (at 750 nm, B) and quantiﬁcation (C). Cells were incubated for 1 h with RDCs (5 mM) in either DMEM or PBS. (D) HCT116 cells were treated for 2 h either in DMEM or PBS, medium was replaced by DMEM with serum for 48 h and MTT was performed. Data are representative of three independent experiments. (E) Efﬁciency of the Förster Resonance Energy Transfer (FRET) between the extremities of a 14 base pairs-long double-stranded DNA labeled with Alexa-488 and Alexa-568. The measurements were performed at the equilibrium of complexation of DNA with the metal complex. Solid line is a guide to the eye. population at 24 h and 48 h, indicative of cell death. RDC34 induced also a G2/M arrest at 24 h, while cisplatin blocked HCT116 cells in G1. The cytotoxic potential of RDC11, RDC34 and RDC41 was submitted to the US National Cancer Institute (NCI) test on 60 cancer cell lines (Fig. 3B–D). RDC34 showed a strong cytotoxic effect (indicated by the % of cell survival compared to the initial number) on almost all cell lines (Fig. 3B–D, supplementary data #3). These results were similar to those we obtained on additional cell lines (N2A, U87, F10B16, SW480, supplementary data #2). In order to understand how each modiﬁcation of the ligands affects RDCs cytotoxicity and speciﬁcity, we compared their DNA interactions. RDCs were incubated with double stranded supercoiled DNA and then samples were submitted to migration on an agarose gel (Fig. 4A). RDC34 showed an afﬁnity for DNA slightly greater than that of RDC11, but inferior to RDC40 and RDC41. In particular, RDC40 formed a stable complex with DNA that remained in the loading pocket. These data stress the lack of correlation between the DNA binding activity and the cytotoxicity. We therefore looked for another possible explanation for the higher cytotoxicity of RDC34. Previous observations indicated that redox potential allowed RDCs to modulate the oxido-reductase enzyme activity. We therefore hypothesized that such regulation might lead to the production of reactive oxygen species (ROS). Incubation of HCT116 cells with RDCs and a ﬂuorescent ROS probe (Fig. 4B–G) showed that RDC34 was the most potent ROS inducer (Fig. 4G and H), suggesting a correlation between the induction of ROS and cytotoxicity, which was conﬁrmed by the reduction of RDC34 cytotoxicity by NAC, a ROS inhibitor (Fig. 5I). We therefore investigated in more detail the mode of action of RDC34, comparing how this ruthenium compound was regulating the expression of p53 and CHOP, two transcription factors previously shown to be regulated by RDC11 and ROS. We also followed the phosphorylation of H2AX at serine 137, which is a marker for DNA damages [30]. RDC34 was as efﬁcient as RDC11 to induce CHOP, but was more potent to induce p53 protein levels (Fig. 5A, supplementary data 4). In these experimental conditions, cisplatin was unable to regulate CHOP, whereas both RDCs were less able to induce H2AX phosphorylation. Besides the signiﬁcant induction of p53 protein levels, RDC34 increased the expression of pro-apoptotic p53 target genes involved in either the mitochondrial-dependent apoptotic pathway (noxa, bax, siva) or the mitochondria-independent pathway (fas, trail) (Fig. 5B and C). A ﬂuorescent probe (JC-1) was used to follow changes in the mitochondrial membrane potential, indicative of the involvement of the mitochondria in the apoptotic processes induced by RDC34 (Fig. 5E and G). The involvement of a mitochondria-independent pathway was conversely tested by following caspase 8 activity. Compared to RDC11, RDC34 was more potent in inducing caspase 8 (Fig. 5H). The concomitant use of ROS and caspase 8 inhibitors reduced the cytotoxicity or RDC34, suggesting their importance for the antitumor activity of this drug (Fig. 5I). 3.5. In vivo antitumor activity of optimized RDCs To further examine the anticancer potential of the optimized compounds, we tested their activity on tumor growth. A 1432 V. Vidimar et al. / Biochemical Pharmacology 84 (2012) 1428–1436 Fig. 3. Optimized RDC induced cytotoxicity on cancer cells of various origins. (A) Cell cycle proﬁle analysis (FACS) of HCT116 cells treated with cisplatin or RDC34 (2.5 mM) for 24 h or 48 h. (B–D) Graph indicating the cytostatic (above 0) and cytotoxic (bellow 0) response of 60 cancer cells lines of the speciﬁed origins (detailed in supplementary data #3). The test was performed by the NCI with the indicated RDC at a concentration of 5 mM. preliminary experiment to evaluate host toxicity after single dose and after chronic treatment (3 weeks, 2 injections a week) showed that repeated doses of RDC34 higher than 4 mmol/kg were lethal (supplementary data #1). Similar results were obtained with RDC44, a compound with signiﬁcant lower cytotoxicity in vitro (Fig. 1). RDC41 was the most tolerated compound and showed a DL50 of 30 mmol/kg. We ﬁrst tested the compounds on a lung cancer as these cancers are commonly treated by cisplatin and we observed a good cytotoxicity of RDC34 on lung cancer cells, such as on 3LL cells (Figs. 3C and 6A). When tested at their maximum tolerated doses on the syngeneic model of 3LL cells implanted subcutaneously in B6 mice, RDC34 (4 mmol/kg, 2/week), RDC44 (4 mmol/kg, 2/ week) and RDC41 (13 mmol/kg, 2/week) reduced tumor growth to approximately 40% (Fig. 6A). The anticancer activity of RDC34 on human ovarian cancer cells (A2780) implanted into nude mice (Fig. 6B) showed a signiﬁcant reduction of tumor growth to approximately 36% of the controls. In both models the activity of RDC34 is slightly higher compared to cisplatin. 4. Discussion The development of novel anticancer compounds is a constant challenge. In the recent years, metal-based compounds have been the focus of a particular interest based on the proven relative efﬁciency of platinum-derived compounds and their intrinsic physico-chemical properties of transition metals, such as ruthenium. Especially, the vast possibilities of combination with organic ligands and small ions make ruthenium-based molecules particularly suitable to be modeled and adapted to a speciﬁc need. In particular, the wide range of redox states and the hexahedral structure of this metal allow endless metal/ligand combinations with variations in redox state, lipophilic/hydrophilic status, ligand exchange properties, stability, and geometry. V. Vidimar et al. / Biochemical Pharmacology 84 (2012) 1428–1436 1433 Fig. 4. Interaction of the optimized compounds with DNA and production of reactive oxygen species. (A) Circular double-stranded DNA was incubated with RDCs at the indicated ratio (DNA base pairs/molecule of drugs). Complexes were run on a 1% agarose gel, and then stained with ethidium bromide to observe DNA relaxation. Data are representative of 3 independent experiments. (B–G) HCT116 cells were treated with RDC (2.5 mM) for 16 h and labeled with carboxy-H2DCFDA for 1 h. (B) Control cells, (C) cells treated with the positive control, menadione, (D) cells treated with RDC11, (E) cells treated with RDC40, (F) cells treated with RDC41, (G) cells treated with RDC34. (H) HCT116 cells were treated with RDC as described in B–G, then ﬂuorescence was quantiﬁed. Recently, a ruthenium (II)-based organometallic compound, called RDC11, was shown endowed with anticancer properties. In order to optimize this compound, we have modiﬁed the ligands to gain in water solubility and in cytotoxicity. As we hypothesized, the addition of a second phenanthroline signiﬁcantly increased the cytotoxicity of the new compound RDC34 in comparison to RDC11, with an IC50 often in the nanomolar range, conﬁrming the important role of the phenanthroline residue in the cytotoxic properties of RDCs. Compared to RDC11, RDC34 is also more toxic when given to mice [30]. However, at the maximum doses free of host toxicity, RDC34 signiﬁcantly reduces tumor growth in two models. The increased toxicity of RDC34 could be related to its high lipophilicity [40], a property that might favor RDC34 cell uptake and distribution in the body after dosing. Although the addition of a second phenanthroline signiﬁcantly increased the cytotoxicity, the modiﬁcation of the phenylpyridine by addition of a NO2, NH2 or spermine moiety modify also signiﬁcantly the cytotoxicity of the RDCs, highlighting 1434 V. Vidimar et al. / Biochemical Pharmacology 84 (2012) 1428–1436 Fig. 5. Optimized RDCs induce multiple signaling pathways, including ROS and caspase 8. (A) Western blot analysis of cells treated for 24 h. Immunoblotting were performed with anti-p53, anti-actin, anti-phospho-H2AX, and anti-CHOP antibodies. Data are representative of 3 independent experiments. (B and C) HCT116 cells were treated with cisplatin, or RDC34 (in mM) for 24 h. RT-qPCR were performed using primers for noxa, and fas. Data are represented as fold inductions relative to untreated cells (Ct) and were normalized with both 18s and TBP levels. (D–G) Cells were incubated with the membrane potential sensitive dye JC-1. We used FCCP as positive control (E). Intracellular distribution of the dye was assessed by confocal microscopy at 488 nm excitation/510 nm emission. (H) Cells were incubated for the indicated time and concentration of drugs (in mM). (I) HCT116 were treated with the combination of drugs (NAC, 10 mM; caspase 8 inhibitor, C8inh 10 mM) added to RDC34 at the indicated concentrations. Survival was tested 48 h after using MTT tests. the functional interaction occurring between the ligands and their interdependence to produce a speciﬁc biological effect. By adding the spermine chain (RDC44), the NH2 function (RDC41), and in a lesser extend the NO2 (RDC40) function, we purposely improve the water solubility of RDC34 as indicated by their LogP, which is 2.35 for RDC34, 2.05 for RDC40, 0.9 for RDC41 and inferior to 0 for RDC44. RDC44 is freely soluble in water up to a concentration of 25 mM, simplifying its in vivo dosing of the experimental animals. However, as observed with RDC44 and in a lesser extent with RDC41 and RDC40, increasing the water solubility decreased the cytotoxicity. Similar observation has been made previously for the ruthenium compound KP1019 and its water-soluble version KP1339 [42]. Although the low cytotoxicity of RDC44 was disappointing, the in vivo studies revealed that RDC44 displayed similar anticancer properties compared to RDC34. A possible explanation could be that the spermine moiety confers a favorable pharmacokinetics of distribution, or alternatively that the spermine is somehow removed after its injection in mice, suggesting that RDC44 is a water-soluble pro-drug of RDC34. The lower cytotoxicity of RDC44 could be due to a diminished cellular uptake through the lipophilic membrane barrier. Indeed, in standard medium, RDC44 showed a diminished cellular accumulation. However, under PBS conditions both compounds (RDC34, RDC44) can enter the cells and the water-soluble compound RDC44 is still barely able to affect cell survival compared to the lipophilic RDC34. Moreover, RDC44 is still able to interact with DNA with an afﬁnity similar to RDC34. This lack of correlation between the cytotoxicity and the ability to interact with DNA is also supported by the results obtained with RDC40 and RDC41, which display a better afﬁnity to DNA than RDC34, although they show a weaker cytotoxicity. Therefore, our data suggest that the improvement of RDC water solubility does not affect signiﬁcantly RDC uptake and RDC-DNA interaction. We can therefore hypothesize that besides DNA, RDCs recruit additional direct intracellular targets that can account for a variation in their biological function depending on RDCs lipophilicity. In favor of this hypothesis is the change of selectivity towards cancer cells of different origins observed in RDCs with variation in their ligands. Indeed, by modifying the ligand around the ruthenium center, we have observed that RDC34 displayed a strong cytotoxicity toward a broad range of cancer cell lines. RDC34 is particularly efﬁcient on cancer cell lines of kidney origin that are poorly affected by RDC11. The introduction of the NH2 function diminished the cytotoxicity of RDC41 at all levels comparable to that of RDC11. However, few exceptions exist in which RDC41 is more active than RDC11, such as EKVX cells, MDAMB-435, UACC-257 and HS57BT. It is likely that the ability of the various RDCs to interact with cellular DNA does not explain their selectivity for different cell lines but rather their ability to interact with different intracellular targets and to induce different stress signaling. As we have published for RDC11, the new generation of improved RDCs triggers at least two mechanisms: an interaction with DNA and an induction of CHOP expression. The interaction V. Vidimar et al. / Biochemical Pharmacology 84 (2012) 1428–1436 1435 Fig. 6. Anticancer activity of optimized RDC. 3LL cells (A) or A2780 cells (C) were treated for 48 h with cisplatin or RDC (in mM). Viability was evaluated using a MTT test. (B) C57BL/6 mice (8-weeks old) were injected subcutaneously with 5 105 3LL cells. Injections of RDC (RDC34 and RDC44 at 4 mmol/kg; RDC41, RDC11 at 13.3 mmol/kg) or cisplatin (13.3 mmol/kg) started when tumors were palpable (80 mm3) and were performed twice a week. Solutions were prepared in PBS/5% Cremophore. (D) Nude mice (Swiss nu/nu, 8-weeks old) were injected subcutaneously with 5 106 A2780 cells. Injections of RDC34 (4 mmol/kg) or cisplatin (13.3 mmol/kg) were performed twice a week when tumors were palpable. Graph represents tumor volumes. Data are representative of two independent experiments (n = 7). Asterisks indicate statistically signiﬁcant difference (p < 0.01) compared to control, as calculated by a one-way ANOVA test followed by a Tukey test. with DNA does not seem to play a major role in their cytotoxicity, as there is not a good correlation between the cytotoxicity and their ability to interact with DNA. Interestingly RDC34 is a more potent inducer of the p53 protein levels and the p53 target genes compared to RDC11 even though its interaction with DNA is similar to RDC11. This stronger effect might involve the higher ability of RDC34 to induce ROS production, as ROS can induce p53. The role of ROS in the cytotoxicity of ruthenium compounds has been previously suggested [27,43] and is demonstrated by our present study as NAC signiﬁcantly reduces RDC34 cytotoxicity. However, the correlation between production of ROS and RDC cytotoxicity is only partial, as there is no difference in ROS production between RDC40 and RDC11 even though RDC40 is more cytotoxic than RDC11. Previous studies indicated that RDCs modulate the activity of oxido-reductase enzymes [44], suggesting that the production of ROS might be triggered by alterations of the activity of enzymes committed to produce or remove ROS, although these targeted enzymes remain to be identiﬁed yet. Anyhow, RDC34 is a strong inducer of p53 and p53 target genes connected to the extrinsic pro-apoptotic pathways involving caspase 8. The role of caspase 8 seems crucial to the pathway leading to cell death after treatment with RDC34 since its inhibition reduced these cytotoxic effects. Overall this study demonstrates the crucial role of modiﬁed ligand around the ruthenium center to optimize the molecules in order to improve the cytotoxicity and target cancers subtypes. The variation in cytotoxicity between the compounds tested in the present study might involve several factors: (1) the redox potential, (2) the lipophilic status, and (3) the geometry. These factors might inﬂuence the ability of the compound to enter the cells, interact physically with intracellular targets and modify their functions. These factors also certainly affect the pharmacokinetic and the tissues distribution of the compounds in vivo and might explain the differences in toxicity. Moreover, this study shows the possibility of producing a RDC pro-drugs that present fair water solubility without affecting signiﬁcantly the in vivo anticancer activity of the pharmacophore. Acknowledgements This work was supported by CNRS (CG), Association pour la Recherche Contre le Cancer Grant 3288, La Ligue Contre le Cancer, 1436 V. Vidimar et al. / Biochemical Pharmacology 84 (2012) 1428–1436 ANR (PhotoBioMet project, AAPADC project), Institut National du Cancer. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.bcp.2012.08.022. References [1] Bruijnincx PC, Sadler PJ. New trends for metal complexes with anticancer activity. Curr Opin Chem Biol 2008;12(2):197–206. [2] Jakupec MA, Galanski M, Arion VB, Hartinger CG, Keppler BK. Antitumour metal compounds: more than theme and variations. Dalton Trans 2008;(2):183–94. [3] Dyson PJ, Sava G. Metal-based antitumour drugs in the post genomic era. Dalton Trans 2006;(16):1929–33. [4] Kannarkat G, Lasher EE, Schiff D. Neurologic complications of chemotherapy agents. Curr Opin Neurol 2007;20(6):719–25. [5] Markman M. Toxicities of the platinum antineoplastic agents. Expert Opin Drug Safe 2003;2(6):597–607. [6] Kelland L. The resurgence of platinum-based cancer chemotherapy. Nat Rev Cancer 2007;7(8):573–84. [7] Allardyce CS, Dyson PJ. Ruthenium in medicine: current clinical uses and future prospects. Platinum Metals Rev 2001;45:62. [8] Anghileri LJ. The in vivo inhibition of tumor growth by ruthenium red: its relationship with the metabolism of calcium in the tumor. Z Krebsforsch Klin Onkol Cancer Res Clin Oncol 1975;83(3):213–7. [9] Giraldi T, Sava G, Bertoli G, Mestroni G, Zassinovich G. Antitumor action of two rhodium and ruthenium complexes in comparison with cis-diamminedichloroplatinum(II). Cancer Res 1977;37(8 Pt 1):2662–6. [10] Sava G, Giraldi T, Mestroni G, Zassinovich G. Antitumor effects of rhodium(I): iridium(I) and ruthenium(II) complexes in comparison with cis-dichlorodiammino platinum(II) in mice bearing Lewis lung carcinoma. Chem Biol Interact 1983;45(1):1–6. [11] Keppler BK, Balzer W, Seifried V, Synthesis. antitumor activity of triazoliumbis(triazole)-tetrachlororuthenate (III) and bistriazolium-triazolepentachlororuthenate (III). Two representatives of a new class of inorganic antitumor agents. Arzneimittelforschung 1987;37(7):770–1. [12] Sava G, Pacor S, Zorzet S, Alessio E, Mestroni G. Antitumour properties of dimethylsulphoxide ruthenium (II) complexes in the Lewis lung carcinoma system. Pharmacol Res 1989;21(5):617–28. [13] Fruhauf S, Zeller WJ. New platinum: titanium, and ruthenium complexes with different patterns of DNA damage in rat ovarian tumor cells. Cancer Res 1991;51(11):2943–8. [14] Novakova O, Kasparkova J, Vrana O, van Vliet PM, Reedijk J, Brabec V. Correlation between cytotoxicity and DNA binding of polypyridyl ruthenium complexes. Biochemistry 1995;34(38):12369–78. [15] Morris RE, Aird RE, Murdoch Pdel S, Chen H, Cummings J, Hughes ND, et al. Inhibition of cancer cell growth by ruthenium(II) arene complexes. J Med Chem 2001;44(22):3616–21. [16] Scolaro C, Bergamo A, Brescacin L, Delﬁno R, Cocchietto M, Laurenczy G, et al. In vitro and in vivo evaluation of ruthenium(II)–arene PTA complexes. J Med Chem 2005;48(12):4161–71. [17] Hartinger CG, Zorbas-Seifried S, Jakupec MA, Kynast B, Zorbas H, Keppler BK. From bench to bedside—preclinical and early clinical development of the anticancer agent indazolium trans-[tetrachlorobis(1H-indazole)ruthenate(III)] (KP1019 or FFC14A). J Inorg Biochem 2006;100(5–6):891–904. [18] Rademaker-Lakhai JM, van den Bongard D, Pluim D, Beijnen JH, Schellens JH. A Phase I and pharmacological study with imidazolium-trans-DMSO-imidazoletetrachlororuthenate: a novel ruthenium anticancer agent. Clin Cancer Res 2004;10(11):3717–27. [19] Mei HY, Barton JK. Tris(tetramethylphenanthroline)ruthenium(II): a chiral probe that cleaves A-DNA conformations. Proc Natl Acad Sci U S A 1988;85(5):1339–43. [20] Brabec V. DNA modiﬁcations by antitumor platinum and ruthenium compounds: their recognition and repair. Prog Nucleic Acid Res Mol Biol 2002;71:1–68. [21] Zeglis BM, Pierre VC, Barton JK. Metallo-intercalators and metallo-insertors. Chem Commun (Camb) 2007;(44):4565–79. [22] Gaiddon C, Jeannequin P, Bischoff P, Pfeffer M, Sirlin C, Loefﬂer JP. Ruthenium (II)-derived organometallic compounds induce cytostatic and cytotoxic effects on mammalian cancer cell lines through p53-dependent and p53-independent mechanisms. J Pharmacol Exp Ther 2005;315(3):1403–11. [23] Hayward RL, Schornagel QC, Tente R, Macpherson JS, Aird RE, Guichard S, et al. Investigation of the role of Bax: p21/Waf1 and p53 as determinants of cellular responses in HCT116 colorectal cancer cells exposed to the novel cytotoxic ruthenium(II) organometallic agent, RM175. Cancer Chemother Pharmacol 2005;55(6):577–83. [24] Chatterjee S, Kundu S, Bhattacharyya A, Hartinger CG, Dyson PJ. The ruthenium(II)-arene compound RAPTA-C induces apoptosis in EAC cells through mitochondrial and p53-JNK pathways. J Biol Inorg Chem 2008;13(7):1149–55. [25] Smalley KS, Contractor R, Haass NK, Kulp AN, Atilla-Gokcumen GE, Williams DS, et al. An organometallic protein kinase inhibitor pharmacologically activates p53 and induces apoptosis in human melanoma cells. Cancer Res 2007;67(1):209–17. [26] Ang WHet al. Organometallic ruthenium inhibitors of glutathione-S-transferase P 1-1 as anticancer drugs. ChemMedChem 2007. [27] Dougan SJ, Habtemariam A, McHale SE, Parsons S, Sadler PJ. Catalytic organometallic anticancer complexes. Proc Natl Acad Sci U S A 2008;105(33):11628–33. [28] Gava B, Zorzet S, Spessotto P, Cocchietto M, Sava G. Inhibition of B16 melanoma metastases with the ruthenium complex imidazolium trans-imidazoledimethylsulfoxide-tetrachlororuthenate and down-regulation of tumor cell invasion. J Pharmacol Exp Ther 2006;317(1):284–91. [29] Leyva L, Sirlin C, Rubio L, Franco C, Le Lagadec R, Spencer J, et al. Synthesis of cycloruthenated compounds as potential anticancer agents. Eur J Inorg Chem 2007;(19):3055–66. [30] Meng Xet al. A ruthenium-containing organometallic compound reduces tumor growth through induction of the endoplasmic reticulum stress gene CHOP. Cancer Res 2009. [31] Benosman S, Gross I, Clarke N, Jochemsen AG, Okamoto K, Loefﬂer JP, et al. Multiple neurotoxic stresses converge on MDMX proteolysis to cause neuronal apoptosis. Cell Death Differ 2007. [32] Benosman S, Meng X, Von Grabowiecki Y, Palamiuc L, Hritcu L, Gross I, et al. Complex regulation of p73 isoforms after alteration of amyloid precursor polypeptide (APP) function and DNA damage in neurons. J Biol Chem 2011;286(50):43013–25. [33] Bergamo A, Gaiddon C, Schellens JH, Beijnen JH, Sava G. Approaching tumour therapy beyond platinum drugs Status of the art and perspectives of ruthenium drug candidates. J Inorg Biochem 2012;106(1):90–9. [34] Klajner M, Hebraud P, Sirlin C, Gaiddon C, Harlepp S. DNA binding to an anticancer organo-ruthenium complex. J Phys Chem B 2010;114(44): 14041–47. [35] Gaiddon C, de Tapia M, Loefﬂer JP. The tissue-speciﬁc transcription factor Pit1/GHF-1 binds to the c-fos serum response element and activates c-fos transcription. Mol Endocrinol 1999;13(5):742–51. [36] Nicoletti I, Migliorati G, Pagliacci MC, Grignani F, Riccardi C. A rapid and simple method for measuring thymocyte apoptosis by propidium iodide staining and ﬂow cytometry. J Immunol Methods 1991;139(2):271–9. [37] Sohm F, Gaiddon C, Antoine M, Boutillier AL, Loefﬂer JP. The retinoblastoma susceptibility gene product/Sp1 signalling pathway is modulated by Ca2+/ calmodulin kinases II and IV activity. Oncogene 1999;18(17):2762–9. [38] Gaiddon C, Tian J, Loefﬂer JP, Bancroft C. Constitutively active G(S) alphasubunits stimulate Pit-1 promoter activity via a protein kinase A-mediated pathway acting through deoxyribonucleic acid binding sites both for Pit-1 and for adenosine 30 ,50 -monophosphate response element-binding protein. Endocrinology 1996;137(4):1286–91. [39] Ryabov AD, Kurova VS, Ivanova EV, Le Lagadec R, Alexandrova L. Redox mediation and photomechanical oscillations involving photosensitive cyclometalated Ru(II) complexes: glucose oxidase, and peroxidase. Anal Chem 2005;77(4):1132–9. [40] Fetzer L, Boff B, Ali M, Xiangjun M, Collin JP, Sirlin C, et al. Library of secondgeneration cycloruthenated compounds and evaluation of their biological properties as potential anticancer drugs: passing the nanomolar barrier. Dalton Trans 2011;40(35):8869–78. [41] McGhee JD, von Hippel PH. Theoretical aspects of DNA–protein interactions: co-operative and non-co-operative binding of large ligands to a one-dimensional homogeneous lattice. J Mol Biol 1974;86(2):469–89. [42] Heffeter P, Bock K, Atil B, Reza Hoda MA, Korner W, Bartel C, et al. Intracellular protein binding patterns of the anticancer ruthenium drugs KP1019 and KP1339. J Biol Inorg Chem 2010;15(5):737–48. [43] Jakupec MA, Reisner E, Eichinger A, Pongratz M, Arion VB, Galanski M, et al. Redox-active antineoplastic ruthenium complexes with indazole: correlation of in vitro potency and reduction potential. J Med Chem 2005;48(8):2831–7. [44] Le Lagadec R, Alexandrova L, Estevez H, Pfeffer M, Laurinavius V, Razumiene J, et al. Bis-ruthena(III)cycles [Ru((CN)-N-boolean AND)(2)((NN)-N-boolean AND)]PF6 as low-potential mediators for PQQ alcohol dehydrogenase ((CN)-N-boolean AND = 2-phenylpyridinato or 4-(2-tolyl)pyridinato, (NN)N-boolean AND = bpy or phen). Eur J Inorg Chem 2006;(14):2735–8.