“Half-Sandwich”/RuII Anticancer Complexes Containing Triphenylphosphine and p-Substituted Benzoic Acids

Mononuclear and binuclear Ru/arene/triphenylphosphine complexes with p-substituted benzoic acid derivatives were prepared and characterized. These monocationic complexes of type [Ru(η-p-cymene)(PPh3)L] (L = benzoic acid (1), p-hydroxybenzoic acid (2), p-nitrobenzoic acid (3) and terephthalic acid (4)) were characterized using various techniques, such as nuclear magnetic resonance (NMR) and matrix-assisted laser desorption/ionization-time of flight (MALDI-TOF) mass spectrometry, and the crystal structure of 1, 3 and 4 were determined by X-ray diffraction analysis. The cytotoxicity of the complexes was evaluated, in vitro, against tumorigenic [MDA-MB-231, MCF-7 (breast), A549 (lung) and DU-145 (prostate)] and non-tumorigenic [MCF-10A (breast), MRC-5 (lung) and PNT-2 (prostate)] cells. The binuclear complex (4) was inactive due to its low solubility. Complexes 1, 2 and 3 showed similar cytotoxicity, however, complex 1 presented better selectivity index against MDA-MB-231 than compounds 2 and 3. Cellular ruthenium absorption was explored by inductively coupled plasma mass spectrometry (ICP-MS) analyzing the whole cells and the culture medium. Complementary studies showed that complex 1 inhibited colony formation, induced morphology changes in cells and promoted cell cycle arrest in the Sub-G1 phase for the MDA-MB-231 cells.


Introduction
In the medicinal inorganic chemistry field, research on new metal-based anticancer drugs has received considerable attention since the discovery of anticancer properties of cisplatin and analog drugs. Pt-based coordination compounds have been present in most conventional cancer treatments in recent years (40% approximately). Moreover, they are just as important now as they were when they first reached the market fifty years ago. 1,2 However, these drugs are restricted by severe dose-limiting due to side-effects. For instance, a cancer patient treated with platinum drugs can experience around 40 specific side-effects, such as severe nausea, vomiting and tissue damage as neurotoxicity, nephrotoxicity and others. [3][4][5][6] Ideally, an effective cancer drug is active against tumor cells not causing damage to healthy tissues. These facts are one of the most important motivator aspects to develop either new cisplatin analogs or other metal-based anticancer drugs. Recently, many tools have been used to develop new drugs/metallodrugs, such as those based on the structure-activity relationship (SAR), 7,8 "Half-Sandwich"/Ru II Anticancer Complexes Containing Triphenylphosphine and p-Substituted Benzoic Acids pharmacophore models, 9 pharmacokinetic approaches [10][11][12] and nanopreparations. 3,5,13,14 In addition, another way to improve the metallodrugs used in the cancer treatment is to search for new active and selective complexes based on different metal centers. Among all the transition metals present in the periodic table, great emphasis has been given to the ruthenium complexes, which have a great structure variety. The first ruthenium complex with anticancer activity reported in the literature 15 was the cis-[Ru(NH 3 ) 4 (Cl) 2 ]Cl (Figure 1), the cisplatin analog, for which the hypothesis of "activation-by-reduction" of Ru III /Ru II was suggested. This proposition was confirmed for one of the most studied ruthenium complexes, the NAMI-A (Figure 1), an efficient anti-metastatic drug. Alessio and co-workers 2 have demonstrated that to be active, in vivo, the NAMI-A acts as prodrugs under physiological conditions. Researchers [15][16][17][18][19] have suggested that the reduction of the Ru III /Ru II is important to produce a more labile complex, which rapidly reacts with specific sites of biomolecules, improving its antitumor activity.
One specific class of ruthenium(II) compounds that has become very popular in recent years is the organometallic complexes of type "half-sandwich". Reports on Ru II /η 6 -arene complexes with the general formula [Ru(η 6 -arene)(L−L)(X)] n can be found in the literature, where "L−L" (bidentate) and "X" (monodentate) are neutral or charged ligands of various types. Usually, X is a halide and, similar to NAMI-A, the activation of these complexes requires the hydrolysis of these ligands. However, the half maximal inhibitory concentration (IC 50 ) values in cancer cells for these compounds are generally high (> 100 μM), which is attributed to the high reactivity of these complexes, as a result of the low stability in organic solvents or under physiological conditions. [20][21][22][23] In addition, there are many papers 22,23 which currently report the use of monophosphines or monopyridines as the "X" ligand, resulting in complexes with high toxicity (< 5 μM), thanks to the lipophilicity enhancement of the compounds, when at least one of these molecules is present in their structures. Furthermore, when "L−L" is N−N, 24-26 P−P, 27-32 N−S, 33,34 24,31,32 O−O 24,35,36 donor ligand, a wide range of IC 50 values is obtained. Ruthenium arene 1,3,5-triaza-7-phosphaadamantane (RAPTA)-like ( Figure 1) compounds are other groups of half-sandwich Ru II compounds that show good anticancer properties. Many of them present the phosphane ligand, known as PTA (1,3,5-triaza-7-phosphaadamantane), to form water-soluble complexes due to the high hydrophilic character of the PTA ligand. Thus, NAMI-A and RAPTA types of compounds exert an antimetastatic activity, and in parallel, present a good cytotoxic activity, and low toxicity, in vivo. This fact is surprising, considering the differences between the complexes in terms of ruthenium oxidation state, complex charge, ligands and geometry.
Concerning the O−O type of ligands coordinated in the bidentate form to the Ru II /η 6 -arene complex, it should be emphasized that the biological properties of carboxylatebased compounds are underexplored. Probably, the tensioned four-membered ring formed by metal/carboxylate anion moiety may present low stability when compared with the correspondent five/six-membered ones. Thus, the main problem of these kinds of Ru II /η 6 -arene/carboxylate complexes can be their low stability in coordinating solvents or physiological medium, limiting their biological application. 37,38 Benzoic acid and its analogs belong to an important class of ligands, displaying a structural diversity, allowing them to access the ortho, meta and para positions with a wide possibility of substitution with many organic functions. Due to the promising chemical and biological results obtained with the Ru II /η 6 -arene-based complexes, this paper reports on the reactivity of p-substituted benzoic acid derivatives with Ru II /η 6 -arene/triphenylphosphine complexes with general formula [Ru(η 6 -p-cymene)(PPh 3 ) (L 3 )]BF 4 (benzoic acid (L 1 ); p-hydroxybenzoic acid (L 2 ); p-nitrobenzoic acid (L 3 ) and terephthalic acid (L 4 )), in order to obtain active and stable compounds. The complexes were characterized by various techniques, such as Fourier transform infrared spectroscopy (FTIR), multinuclear nuclear magnetic resonance (NMR) spectroscopy and X-ray diffraction (XRD) experiments. Afterwards, the cytotoxicity of the complexes against four human tumor cell lines (MDA-MB-231/MCF-7 (breast), DU-145 (prostate) and A549 (lung)) and three non-tumor human cell lines (MRC-5 (lung), MCF-10A (breast) and PNT2 (prostate)) were evaluated. Therefore, to better understand the action mode of these complexes, we studied the cellular uptake and cell death mechanism, including cell cycle disturbances and cell migration inhibition.
Alternatively, an attempt to obtain complex 1, using sodium benzoate and [Ru(η 6 -p-cymene)(PPh 3 )Cl 2 ] in methanolic solution under reflux for 24 h, was monitored by 31 P{ 1 H} nuclear magnetic resonance. As a result, the formation of several by-products was observed, in which the most abundant one (occurring at 51.6 ppm in 31 P{ 1 H} NMR spectrum) was isolated by crystallization (methanol/hexane mixture 1:1 v/v). The crystal structure of this complex was determined by X-ray diffraction ( Figure 2). Surprisingly, the arene ligand is absent in the complex. As confirmed by previous literature, 39,40 the reflux of ruthenium/arene complexes in methanolic solutions can promote the arene loss, making three coordination points available. As can be seen in Figure 2, the complex displays an interesting structural behavior observed for the first time. In addition to the coordinated water bridged between the ruthenium atoms, the structure obtained presented two triphenylphosphines (one coordinate for each ruthenium), one monodentate coordinated methanol molecule and four differently coordinated benzoates, resulting in a complex of formula [Ru 2 (L 1 ) 4 (CH 3 OH)(PPh 3 ) 2 (H 2 O)], where L 1 = benzoic acid. The crystal structure displays the three possible coordination modes for carboxylate anion, monodentate, bidentate and bridging bidentate (see Figure 2). The absence of Ru III signals in the electron paramagnetic resonance (EPR) spectrum confirms the oxidation state of the metal (Ru II ).
The single crystals of complexes 1, 3 and 4 were obtained by slow evaporation of a dichloromethane solution. Complexes 1 and 3 crystallized with the addition of the KPF 6 salt, presenting the PF 6 − as a counterion, while complex 4 crystallized with the original BF 4 − counterion, with no need to add KPF 6 salt. The coordination mode of carboxylic acid to the metal as bidentate via O−O atoms was unambiguously confirmed by the X-ray technique, differing from the former binuclear structure with diverse coordination modes. The summary of crystal data and Scheme 1. Scheme for the synthesis of complexes 1-3 and 4 with the respective ligands: benzoic acid (L 1 ); p-hidroxybenzoic acid (L 2 ); p-nitrobenzoic acid (L 3 ) and terephthalic acid (L 4 ). structural refinements are shown in the Supplementary Information (SI) section (Table S1). All complexes crystallized in a triclinic space group, in which complex 3 presents one molecule per asymmetric unit, complex 1 crystallizes with two equivalent molecules, while complex 4 shows one-half molecule being the complete structure generated by the inversion center located in the terephthalic ring.
All complexes present a "piano-stool" structure, displaying the ruthenium center in a pseudo-octahedral geometry, as can be seen in Figure 3. The six carbon atoms of the p-cymene ligand form the seat, whilst the three legs of the "piano-stool" are constituted by the phosphorus (triphenylphosphine) and oxygen atoms of carboxylate ion. The distorted geometry is confirmed by the P 1 −Ru−O 1 and P 1 −Ru−O 2 bond angles at around 90° and the low angle value for O 1 −Ru−O 2 (near 60°), as expected, which occurs due to the high tensioned four-membered ring composed by the carboxylate group coordinated to the metal center. Selected bond lengths and bond angles are represented in Table 1. The bond lengths C 1 −O 2 and C 1 −O 1 differ from the free ligands due to the electron delocalization between these chemical bonds after the deprotonation and subsequent coordination to the metal center. Thus, the C−O bond lengths are between the C=O double and C−O single bond lengths (Table 1).
The infrared (IR) spectra of the free p-substituted benzoic ligands exhibit the νO−H stretching vibrations in the range of 3700-3250 cm −1 , while in the spectra of the complexes (see Figure S22 of SI section) this band disappears due to the deprotonation and coordination of the carboxylate anion to the metal center, except for complex 2, which has a p-substituted hydroxyl group. Another important change in the spectra of the complexes was the shift of the symmetric (ν sym COO − ) and asymmetric (ν asy COO − ) stretching of carboxylate anion, when   compared with the spectra of the free ligands. According to Honorato et al. 41 and Nakamoto, 42 complexes with COO − coordinated bidentate to the metal exhibit Δ value (Δ = ν sym COO − − ν asy COO − ) significantly lower than the values found for the free ligands. All complexes exhibit the same behavior, in which the ν sym COO − shift to high frequencies, and the ν asy COO − shift to low frequencies, 41,[43][44][45] as can be seen in Table 2. Thus, the coordination of the carboxyl group to the ruthenium metal by bidentate mode can be also confirmed by IR spectra, as confirmed by X-ray crystallographic structures. An intense band near 1000 cm −1 in the complexes refers to νB−F, showing the presence of the BF 4 − counter anion, as supported by molar conductance values. Weak intensity bands observed at around 495-501 and 336-377 cm −1 are assigned to νRu−P and νRu−O stretching vibrations, respectively.
All complexes were characterized by 1D multinuclear 31 P{ 1 H}, 1 H, and 13 C{ 1 H} NMR experiments and by 2D correlation spectroscopy (COSY) ( 1 H-1 H), heteronuclear single quantum correlation (HSQC) ( 13 C-1 H) and heteronuclear multiple bond correlation (HMBC) ( 13 C-1 H). The 31 P{ 1 H} spectra of complexes display just one signal, a singlet for coordinated phosphorus of triphenylphosphine at around 35 ppm (see Table 2). As can be seen, the 31 P{ 1 H} signals in the complexes shifted to high frequencies when compared with the precursor [Ru(η 6 -p-cymene) (PPh 3 )Cl 2 ] (24 ppm) as a consequence of carboxylate coordination. The main cause of this deshielding effect on the phosphorus signal is the exchange of chlorido ligands (π-and σ-donor) by the carboxylate ligand (π-acceptor). 1 H and 13 C{ 1 H} NMR spectra of all synthesized complexes display the number of peaks according to the expected structures. In the 1 H NMR spectra of 1-4 (Figures S1, S6, S11 and S16 of SI section), three sets of signals were identified: the first one at low frequencies (d 1-3 ppm) refers to the hydrogens of the methyl and propyl group of p-cymene. The second at around d 5-6 ppm are signals assigned to the hydrogens of the ring of p-cymene, and the last refers to the hydrogens of triphenylphosphine and of p-substituted benzoic acids. All attributions were confirmed by the 1 H-1 H COSY experiment, showing the respective vicinal coupling of these chemical shifts.
A different displacement was observed for the hydrogens of p-substituted benzoic acid ring. The activating/deactivating effect of substituent exerts a direct influence on the chemical shift of these signals. As described in Figure 4, complex 3, that contains the most deactivating group (NO 2 group) has the signal shifted to high frequencies (red arrow), while in complex 2, with the most activating group (OH group), chemical shifts occur to lower frequencies (blue arrow). This effect is induced by the electron donation of the OH group, shielding the aromatic hydrogens, while the electron withdrawing effect of the NO 2 group deshields these signals. 46,47 The 13 C NMR spectra of all complexes show one signal around 180 ppm, which is assigned to the carbon from the carboxylate group of the coordinated ligands. This deshielded signal, assigned to the coordinated carboxylate group, is far from that one observed in the free ligand (ca. 173 ppm) ( Table 2). These results are in agreement with the literature report 41 that attributed this pronounced displacement to π-electron delocalization of the chelate ring formed after deprotonation and coordination, promoting a deshielding effect. This is also important evidence of the  metal coordination to the ligands through the carboxyl group. All the other signals were attributed using the 2D experiments ( 13 C-1 H HSQC and HMBC) (see the NMR spectra in the SI section, Figures S1 to S20). The matrix-assisted laser desorption/ionizationtime of flight mass spectra (MALDI-TOF-MS) for complexes 1-3 (see Figure S21 in the SI section) showed a signal corresponding to molecular mass without the tetrafluoroborate ion [M] +  The stability of complexes 1-4 was studied using 31 P{ 1 H} NMR in acetone, dimethyl sulfoxide (DMSO) and in the mixture of these solvents with Roswell Park Memorial Institute (RPMI) biological medium (solvent/ RPMI = 2:1). In DMSO and DMSO/RPMI, all compounds were not stable, forming several by-products from the labilization of the p-substituted benzoic acid or arene ligands, and were replaced by DMSO or chloride ions. This happened instantaneously after solubilization, as already described in the literature. 48 However, in the acetone/ RPMI (2:1) mixture, the complexes were stable within the experiment time (48 h), displaying two signals at 0 ppm (phosphate of RPMI medium) and 34 ppm (complex 1) ( Figure 5). Previous reports 49,50 demonstrate that acetone was the most favorable solvent to dissolve compounds in cell growth in vitro experiments, displaying a non-toxic profile (0.1-1% v/v), which is a compatible solvent with the examined cells, as the MCF-7 studied in this work.

Biological anticancer experiments
The in vitro cytotoxicity assays of 1-4 and free ligands were carried out against four human tumor cell lines (MDA-MB-231/MCF-7 (breast), DU-145 (prostate) and A549 (lung)) and three non-tumor human cell lines (MRC-5 (lung), MCF-10A (breast) and PNT2 (prostate)), adopting a conventional tetrazolium colorimetric (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT)) method. The IC 50 values and selective indexes (SI) are summarized in Table 3. The higher concentration tested for complex 4 was 30 μM, due to its low solubility. Under this condition, complex 4 does not display cytotoxic activity as it is inactive. On the other hand, complex 3 is more toxic than complexes 1, 2 and the precursor. Given that complexes 1, 2 and 3 were more cytotoxic than the respective precursor and free ligands for all evaluated cell lines, this suggests that the coordination of p-substituted benzoic acids to ruthenium may improve the cytotoxicity. As can be seen in Table 3, complex 1 was the most selective in all the tested cells, and was 3.5 times more active in the breast cancer cell MDA-MB-231 than the respective non-tumorigenic MCF-10A cells. These results are good evidence of the influence of the triphenylphosphine ligand on cellular toxicity. The literature [20][21][22][23] cites many examples of Ru IIarene complexes, without phosphine, with very high IC 50 values in various tumorigenic cell lines. One of the causes that leads to low toxicity of these compounds is the lack of lipophilic groups, decreasing the cellular uptake of these complexes. 33,51,52 To investigate the effects on cell morphology of the most selective compound, complex 1 was incubated with the MDA-MB-231 cells at different concentrations ( Figure S23 of SI section) proportional to the IC 50 values. Negative control cells have a spindle-shaped phenotype in both strains, while cells treated with complex 1, especially at 24 μM for MDA-MB-231 (2 × IC 50 , 24 h), have spherically shaped cells (which may be indicative of apoptosis), 33,[53][54][55][56][57] in addition to damaged cell bodies, ongoing cell death, loss of adhesion and confluence. For concentrations lower than IC 50 , no considerable changes were observed, showing that complexes cannot cause visible damage at these concentrations.
The clonogenic assay is an in vitro cell survival test based on the ability of a single cell to proliferate retaining its reproductive ability to form a large colony (> 50 cells). 58,59 The colony area percentage, commonly associated with cell survival and colony intensity, shows the ability of the cells to grow in a densely populated colony. Complex 1 at a concentration of 6.0 μM (2 × IC 50 ) inhibited the number  (Figure 6a). There was a concentration-response tendency, displaying the ability of 1 to perform cytotoxic and cytostatic effects, indicating the continuous loss of clonogenicity of MDA-MB-231 after the treatment. The metastatic process is a multiple step procedure, which includes clonogenic survival, angiogenic process, cell invasion and migration, resulting in a metastasis. Thus, MDA-MB-231, after exposed to 1 (48 h), diminished the ability of single cells (clones) to form colonies, demonstrating a very important property in therapy against metastatic cancer. Different from this, a large number of reports 33,41,60-63 demonstrate the ability of ruthenium-based complexes to eliminate the clonogenicity at low concentrations, not exhibiting a concentration-response tendency.
Cell migration is an important step of angiogenesis, which is associated to the metastasis. The ability of 1 to inhibit the migration of MDA-MB-231 cells was evaluated by the wound healing assay (Figure 7a   or equal to the IC 50 , as compared to the control, as shown in Figure 7b. This result is consistent with the cytotoxic activity of complexes in MDA-MB-231 cells. Inhibition at the Sub-G1 stage indicates apoptosis cell death by DNA cleavage or by an indirect route. For the concentrations lower than the IC 50 , corroborating with the morphological observations, no cell cycle arrest was observed. The ruthenium concentration was determined by inductively coupled plasma mass spectrometry (ICP-MS) in adherent tumor cells MDA-MB-231, in the initial and final culture medium. Figure 8 shows the 102 Ru concentration, expressed in μg L −1 , for the treated cells and for the exposed media. As can be seen, the adherent tumor cells showed approximately 30% of cellular uptake in relation to the initial culture medium. The remaining ruthenium concentration was found in the final culture medium. The analyte concentrations, determined in the samples, did not present significant differences (t-paired test with 95% of confidence). The average concentration of Ru after 24 h of incubation with complex 1 was 250 ± 16 fg per MDA-MB-231 cell. Our results are in agreement with previous publications 64-66 regarding ruthenium uptake in tumor cell lines. The published results revealed an average of 0.9 fg of Ru per A2780 cell incubated with RAPTA-C, 64

General procedure
In a dark room, the [Ru(η 6 -p-cymene)(PPh 3 )Cl 2 ] (0.180 mmol; 0.100 g) was dissolved in 15 mL of acetone for subsequent addition of the respective p-substituted benzoic acid ligand (0.198 mmol for 1-3, 0.090 mmol for 4) and (0.378 mmol) of AgBF 4 . The mixture was kept under inert atmosphere and was stirred for 1.0 h, then this solvent was evaporated. 15 mL dichloromethane was added and filtered off in a Celite column. The resulting solution was dried under reduced pressure solubilized in 2 mL of methanol and precipitated with the addition of distilled water. The solid was filtered off, washed with water (3 × 5 mL) and dried under vacuum.

Instrumentation
All manipulations were carried out under an argon atmosphere using standard Schlenk techniques. The microanalyses were performed using a FISIONS CHNS, model EA 1108. The IR spectra were recorded on a FTIR Bomem-Michelson 102 spectrometer in the 4000-250 cm −1 region, in KBr pellets. Conductivity values were obtained using a Meter Lab CDM2300 instrument. The MALDI-TOF-MS experiments were performed adopting the MALDI-TOF/TOF Bruker Daltonics Autoflex Speed mass spectrometer equipment, using dihydroxybenzoic acid as a calibrator. The NMR experiments ( 1 H, 13  (85% v/v) and performed in acetone or reaction medium, using a capillary containing D 2 O. The 31 P{ 1 H} registered spectra for the stability studies employing as solvents acetone (67%) and RPMI (33%) supplemented with fetal bovine serum (FBS, 10%).

X-ray structure determination
Complexes 1, 3 and 4 were crystallized from methanolic solution resulting in red single crystals. The measurements of a single crystal by X-ray diffraction were performed on a Rigaku XtaLAB mini II or APEX DUO (Bruker) diffractometer with graphite monochromator and Mo Kα radiation (λ = 0.71073 Å). The structures were obtained by the intrinsic phasing method using SHELXT. 69 The Gaussian method was used for the absorption corrections. The table and structure representations were generated by OLEX2 70 and MERCURY, 71 respectively. The main crystal data collections and structure refinement parameters for all complexes are summarized in the SI section (Table S1). non-tumor human prostate cells were maintained in the RPMI-1640 medium also supplemented with 10% of FBS. The non-tumor human breast cell line, MCF-10A (ATCC No. CRL-10317), was cultivated in DMEM/F12 medium containing 5% of horse serum, epidermal growth factor (EGF) (0.02 mg mL −1 ); hydrocortisone (0.05 mg mL −1 ); cholera toxin (0.001 mg mL −1 ); insulin (0.01 mg mL −1 ); penicillin (100 UI mL −1 ); streptomycin (100 mg mL −1 ) and L-glutamine (2 mM). All cell lines were maintained at 37 °C in a humidified 5% CO 2 atmosphere. For cell viability assay by the MTT method, 1.5 × 10 4 cells well −1 were seeded in 150 μL of the supplemented medium into 96-well plates. After 24 h, the cells were treated with different concentrations of the complexes (dissolved in acetone, final concentration of 1% v/v) and incubated for 48 h. MTT (1 mg mL −1 ) was added (30 μL well −1 ) and the plates were incubated again for 4 h. After removing the medium, the formazan crystals were dissolved in isopropanol. Negative control experiments were loaded with the addition of acetone (1%). The optical density was measured at 540 nm using a 96-well multiscanner autoreader (ELISA).

Cell morphology
MDA-MB-231 or MCF-10A cells were seeded (10.5 × 10 5 cells well −1 ) in a 12-well plate and incubated in the supplemented medium at 37 °C in 5% CO 2 for 24 h. Cell morphology was examined after treatment of the cells with different concentrations of complexes for 48 h, in an inverted microscope (Nikon, T5100) with amplification of 10× of magnification.

Colony formation
A density of 300 cells well −1 of the MDA-MB-231 cell line was seeded into a 6-well plate and maintained in the supplemented medium at 37 °C in 5% CO 2 for 24 h. Cells were treated with different concentrations of the complexes for 48 h. The medium was replaced by a fresh medium without any complex and the plates incubated for 10 days. The cells were washed with phosphate-buffered saline (PBS), fixed with methanol and acid acetic (3:1) for 5 min and stained with Panotic dye for 5 min. Relative survival was calculated from the number of single cells that formed colonies (aggregate of at least 50 cells).

Migration
In the wound healing assay, MDA-MB-231 cells (0.5 × 10 5 well −1 ) were plated in 12-well plates and incubated until the culture reached around 90% of confluence. A straight scratch was made with a sterile pipette tip and cells were washed with PBS and fresh medium. Cells were incubated with the complex (lower than IC 50 ). Images were taken using a Moticam, 1000-S camera coupled to an inverted microscope (Nikon, T5100) at 4× total magnification at 0 and 24 h. For the transwell assay, a density of 0.5 × 10 5 cells well −1 of MDA-MB-231 was incubated with complexes, in different concentrations, and seeded on the upper chamber in a DMEM medium without FBS. In the lower chamber, DMEM medium with 10% FBS was added. The eventual migration process occurred for 22 h at 37 °C in 5% CO 2 . Cells that remained in the upper chamber were removed with a cotton swab and cells that migrated through the upper chamber membrane were fixed with methanol and stained with 1% toluidine blue. Migrated cells were quantified by manual counting.

Cell cycle analysis
Flow-cytometric analysis of the cell cycle phase distribution of MDA-MB-231 cells, treated with ruthenium complex, was carried out seeding the quantity of 4.0 × 10 5 cells well −1 , in 6-well plates in DMEM/10% FBS medium. The cells were incubated at 37 °C in a humidified 5% CO 2 atmosphere for 24 h and treated with different concentrations of complexes and camptothecin (positive control). After 48 h of incubation, the cells were collected, washed with PBS and resuspended in 70% ice cold ethanol and frozen at −20 °C for 12 h. After this period, the cells were centrifuged and resuspended in a solution of RNase A (0.02 mg mL −1 ) and propidium iodide (PI) (1 μg mL −1 ) in PBS, incubated at 37 °C for 30 min and the DNA content was determined by flow cytometry. The experiment was performed in triplicate.

Cellular uptake experiments
One day before the treatment, 1.5 × 10 6 MDA-MB-231 cells were seeded in a Corning Costar 125 cm 2 flask. The medium was replaced with a new one containing 5 μM of complex 1 and incubated with the cells for 24 h at 37 °C. A separate flask with medium free metal was used as the control. At the end of the incubation period, the medium was transferred to a pre-cleaned tube, centrifuged (1200 rpm, 5 min) to remove floating cells, and the supernatant was collected for further analysis. The cells were washed twice with 10 mL of ice-cold ultrapure water, trypsinized with 4.0 mL trypsin-ethylenediamine tetraacetic acid (EDTA) solution for 2 min, and 6 mL of culture medium was added. Then, the cell suspension was transferred to a pre-cleaned tube, centrifuged (1200 rpm, 5 min, 2 °C), and the supernatant solution was carefully removed using a pipette. Finally, the cells were washed with 2 mL of cooled ultrapure water, pelleted, and stored at −20 °C for further analysis.

ICP-MS measurements
Ultrapure water (resistivity ≥ 18.2 MΩ cm at 25 °C) obtained from a Milli-Q Element water purification system (Merck Millipore, Bedford, MA, USA) was used throughout. HNO 3 (14 M) for trace metal analysis was previously purified by sub-boiling distillation in a Distillacid TM BSB-939-IR apparatus (Berghof, Eningen, Germany). Tetramethylammonium hydroxide solution (TMAH, Sigma-Aldrich Brazil, Cotia, São Paulo, Brazil) 25% was used as received. All determinations of metal content were conducted by monitoring the 102 Ru signal on an ICP-MS Agilent 7800 equipped with a concentric nebulizer and a Scott double pass spray chamber. A singleelement Ru standard solution used for ICP-MS calibrations was prepared by diluting 1000 mg L −1 of Ru (Qhemis, São Paulo, SP, Brazil) in 0.14 mol L −1 HNO 3 medium, as well as Rh and Ir used as internal standards. The analytical solutions for calibration contained from 0.010 to 200 μg L −1 of each analyte and the internal standards were added at 10.0 μg L −1 to analytical calibration solutions, analytical blanks and samples. Cell pellets were digested using 200 μL of TMAH and acidified with 200 μL of HNO 3 . Subsequently, the digests were diluted with water to 5% v/v HNO 3 . The culture medium was digested using 1 mL of the initial or final medium and 250 μL of HNO 3 directly in the perfluoroalkoxy alkane (PFA) digestion vessels and microwave-assisted digested using a single reaction chamber oven (UltraWave TM , Milestone, Sorisole, Italy). The chamber was pressurized with nitrogen gas (99.9%, White Martins, Praxair) to 40 bar. The microwave heating program was applied as follows: (i) 10 min to reach 180 °C and (ii) 10 min to hold it at 180 °C. Subsequently, digests were diluted to 20.0 mL with distilled-deionized water and an aliquot of each solution was appropriately 50-fold diluted, followed by quantification by ICP-MS. The initial and the final cultured mediums are defined as the medium prepared with 5 μM of complex 1, before and after the cellular exposition.
Supplementary data (NMR, XRD, FTIR and MALDI-TOF spectra) are available free of charge at http://jbcs.sbq.org.br as PDF file.