Ruthenium Half-Sandwich Type Complexes with Bidentate Monosaccharide Ligands Show Antineoplastic Activity in Ovarian Cancer Cell Models through Reactive Oxygen Species Production

Ruthenium complexes are developed as substitutes for platinum complexes to be used in the chemotherapy of hematological and gynecological malignancies, such as ovarian cancer. We synthesized and screened 14 ruthenium half-sandwich complexes with bidentate monosaccharide ligands in ovarian cancer cell models. Four complexes were cytostatic, but not cytotoxic on A2780 and ID8 cells. The IC50 values were in the low micromolar range (the best being 0.87 µM) and were similar to or lower than those of the clinically available platinum complexes. The active complexes were cytostatic in cell models of glioblastoma, breast cancer, and pancreatic adenocarcinoma, while they were not cytostatic on non-transformed human skin fibroblasts. The bioactive ruthenium complexes showed cooperative binding to yet unidentified cellular target(s), and their activity was dependent on reactive oxygen species production. Large hydrophobic protective groups on the hydroxyl groups of the sugar moiety were needed for biological activity. The cytostatic activity of the ruthenium complexes was dependent on reactive species production. Rucaparib, a PARP inhibitor, potentiated the effects of ruthenium complexes.


Introduction
Metal-based drugs used in cancer therapy include square planar platinum(II) complexes, e.g., cisplatin (Figure 1, I), oxaliplatin (II), and carboplatin (III), which are registered worldwide. Although Pt complexes are versatile tools of the trade, their applicability has shortcomings, such as the development of platinum resistance in tumors [1,2], ototoxicity [3], and nephrotoxicity [4,5], with the latter two being very characteristic for cisplatin.
Ruthenium complexes can be promising alternatives to platinum complexes due to their similar chemical characteristics in terms of ligand exchange [6]. In addition, Rucomplexes have enhanced delivery properties compared to Pt-based drugs. For example, Ru-based drugs have improved cellular entry under hypoxic conditions that characterize aggressively growing tumors [7,8], or they can be delivered by binding to transferrin [6]. Among the Ru(III)-and Ru(II)-based complexes, a great number of derivatives have been tested as anticancer metallodrugs, and two of them, the imidazolium salt of tetrachlorido(dimethylsulfoxide)(imidazole)ruthenium(III) (Figure 1, IV) and sodium tetrachloridobis(indazole)ruthenium(III) (V), reached clinical trials [6,[13][14][15][16]. The half-sandwich type Ru(II)-arene organometallics [17], due to a high structural variability, represent one of the most widely investigated compound classes in the development of new candidates of multitargeted metallodrugs [6,18]. The presence of an η 6 -arene or η 5 -arenyl residue in the coordination sphere contributes to the stabilization of the +2 oxidation state of the metal ion and to the maintenance of the hydrophilic/lipophilic balance of the whole molecule. The remaining three coordination sites of the Ru(II) ion are usually occupied by at least one leaving group and mono-or bidentate ligands, as can be seen in RAPTA-T and -C (VI) and RM175 (VII), respectively, which, among others, have become leads for Ru(II)-based complexes [15,17,19].
Due to the biological relevance of sugars, the incorporation of a carbohydrate-containing ligand into platinum group metal complexes in general [6,20] and into Ru(II)-arene/arenyl complexes in particular [15] seems a rational choice for further drug design. Thus, several functional features of carbohydrates, such as their contribution to different cellular phenomena (e.g., to cell-cell recognition and adhesion), their crucial role in cellular energy supply, and their binding capacities to carbohydrate-specific proteins (e.g., lectins, glucose transporters, and glycoenzymes), can be exploited to obtain new platinum group metal complexes with anticancer potential [15,20].
Unlike organic drug molecules, the biological effects of metal complexes can be modulated by a wider array of parameters including size and charge of the species, hardsoft character of the metal ion, stability, inertness, and geometry of the complex, just to mention a few. In this regard, a more comprehensive study of Ru-sugar conjugatebased complexes, in which (A) the role of the metal chelating part of the ligand, (B) the basicity and binding strength of the coordinating donor atoms, and (C) the effect of the lipophilic/hydrophilic character of the complex, tuned by the presence/absence of various protecting groups at the sugar moiety, were explored, may provide a more detailed outlook of the structure-activity relationship (SAR) of these types of complexes.
As mentioned earlier, a rationale for the design and application of ruthenium complexes is to replace platinum compounds by ruthenium complexes in clinical settings. To the best of our knowledge, no real Cand N-glycopyranosyl heterocyclic ligands as potential bidentate chelators have so far been used to obtain Ru(II) arene/arenyl complexes. In an ongoing project focused on these types of sugar derivatives capable of forming either five-or six-membered chelates with Ru(II), herein we report on the synthesis and comprehensive characterization of a set of Cand N-glycopyranosyl azoles and their half-sandwich Ru-complexes (Figure 1, XIV). Since platinum complexes are widely used in the treatment of hematological and gynecological malignancies, the anticancer potential of the above ligands and their complexes were studied in comparison with Pt-complexes I-III against various human ovarian cancer cell lines. Table 1. Synthesis of 1-(β-D-glucopyranosyl)-4-hetaryl-1,2,3-triazoles.

Chemistry
wich Ru-complexes (Figure 1, XIV). Since platinum complexes are widely used in the treatment of hematological and gynecological malignancies, the anticancer potential of the above ligands and their complexes were studied in comparison with Pt-complexes I-III against various human ovarian cancer cell lines.

Chemistry
For the formation of the planned Ru(II) complexes, the sugar-based heterocyclic N,Nchelating ligands were prepared first.

Chemistry
For the formation of the planned Ru(II) complexes, the sugar-based heterocyclic N,Nchelating ligands were prepared first.

Chemistry
For the formation of the planned Ru(II) complexes, the sugar-based heterocyclic N,Nchelating ligands were prepared first.
The complexation reactions of the Ru-dimer with an equimolar amount or a slight excess of the O-peracylated (L-1 and L-2) and the O-unprotected (L-3) 1-(β-D-glucopyranosyl)-4-hetaryl-1,2,3-triazoles in the presence of the halide abstraction reagent TlPF 6 in a CH 2 Cl 2 -MeOH solvent mixture were smoothly accomplished at room temperature to give the PF 6 − salts of the expected [(η 6 -p-cym)Ru II (N-N)Cl] + complexes Ru-1-Ru-3 in excellent yields (Scheme 2). The complexes containing the O-peracylated glucosyl-1,2,3-triazole ligands (Ru-1a,b and Ru-2a) were stable and inert enough to be purified by column chromatography on silica gel, while the isolation of the highly polar complexes having the O-deprotected heterocyclic chelators (Ru-3a,b) could be effected by crystallization. Due to the formation of a new stereogenic center on the metal ion and the chiral nature of the glucose unit, diastereomers of the complexes were formed in each case, whose separation could be achieved neither by column chromatography nor by crystallization.
The formation of the complexes and the existence of the diastereomeric pairs were confirmed by 1 H-and 13 C-NMR spectroscopy in each case. As a representative, the superposition of the 1 H-and 13 C-NMR spectra of Ru-7, the free ligand L-7, and the Rudimer, respectively, are presented in Figure 2. Generally, the conversion of Ru-dimer into the corresponding [(η 6 -p-cym)Ru II (N-N)Cl]PF 6 complexes (Ru-1-Ru-12) led to remarkable downfield shifts of the aromatic p-cymene signals in both the 1 H-and the 13 C-NMR spectra. No significant changes in the chemical shifts of the proton and carbon resonances of the sugar moiety were observed, except for those which were close to the coordination sphere. Thus, H-1 protons of the sugar-based heterocyclic ligands usually displayed noticeable downfield shifts up to 0.05-0.27 ppm upon coordination. On the other hand, e.g., in the case of the O-acyl protected glycosyl-1,3,4-oxadiazole derivatives L-4-L-8 the coordination resulted in either a downfield or an upfield shift of the H-2 resonances depending on the diastereomers formed. The formation of the 5-membered chelates with the participation of heterocyclic aglycone parts was also shown by the change in the chemical shift of several heteroaromatic signals. For example, in case of the 2-pyridyl substituted derivatives, welltraceable and consistent changes were observed in the appearance of the proton and carbon signals of the pyridine ring. For example, the H-6 and C-6 signals of the pyridine ring showed downfield shifts up to 0.48-0.83 ppm ( 1 H-NMR) and 5.7-7.8 ppm ( 13 C-NMR), respectively, as a result of the coordination. A more detailed collection of these data can be found in Tables S1-S8 in the Supplementary Materials. To get the desired cationic half-sandwich Ru(II) complexes, the above heterocyclic monosaccharide derivatives were reacted with the commercially available dichloro(η 6 -pcymene)ruthenium(II) dimer ([(η 6 -p-cym)RuCl2]2, Ru-dimer).

Ru-6 85
(3:2) Ru-9 94 (1:1) Ru-12 50 (1:1) The formation of the complexes and the existence of the diastereomeric pairs were confirmed by 1 H-and 13 C-NMR spectroscopy in each case. As a representative, the superposition of the 1 H-and 13 C-NMR spectra of Ru-7, the free ligand L-7, and the Ru-dimer, respectively, are presented in Figure 2. Generally, the conversion of Ru-dimer into the  (1:1)

Ru-6 85
(3:2) Ru-9 94 (1:1) Ru-12 50 (1:1) The formation of the complexes and the existence of the diastereomeric pairs were confirmed by 1 H-and 13 C-NMR spectroscopy in each case. As a representative, the superposition of the 1 H-and 13 C-NMR spectra of Ru-7, the free ligand L-7, and the Ru-dimer, respectively, are presented in Figure 2. Generally, the conversion of Ru-dimer into the corresponding [(η 6 -p-cym)Ru II (N-N)Cl]PF6 complexes (Ru-1-Ru-12) led to remarkable downfield shifts of the aromatic p-cymene signals in both the 1 H-and the 13 C-NMR spectra. No significant changes in the chemical shifts of the proton and carbon resonances of (1:1)

Ru-6 85
(3:2) Ru-9 94 (1:1) Ru-12 50 (1:1) The formation of the complexes and the existence of the diastereomeric pairs were confirmed by 1 H-and 13 C-NMR spectroscopy in each case. As a representative, the superposition of the 1 H-and 13 C-NMR spectra of Ru-7, the free ligand L-7, and the Ru-dimer, respectively, are presented in Figure 2. Generally, the conversion of Ru-dimer into the corresponding [(η 6 -p-cym)Ru II (N-N)Cl]PF6 complexes (Ru-1-Ru-12) led to remarkable downfield shifts of the aromatic p-cymene signals in both the 1 H-and the 13 C-NMR spectra. No significant changes in the chemical shifts of the proton and carbon resonances of (1:1)

Ru-6 85
(3:2) Ru-9 94 (1:1) Ru-12 50 (1:1) The formation of the complexes and the existence of the diastereomeric pairs were confirmed by 1 H-and 13 C-NMR spectroscopy in each case. As a representative, the superposition of the 1 H-and 13 C-NMR spectra of Ru-7, the free ligand L-7, and the Ru-dimer, respectively, are presented in Figure 2. Generally, the conversion of Ru-dimer into the corresponding [(η 6 -p-cym)Ru II (N-N)Cl]PF6 complexes (Ru-1-Ru-12) led to remarkable downfield shifts of the aromatic p-cymene signals in both the 1 H-and the 13 C-NMR spectra. No significant changes in the chemical shifts of the proton and carbon resonances of  The aqueous stability of the complexes was also studied over time. As an example, the time dependence of the NMR spectra of Ru-3a is shown in Figure S1. The small shifts of signals attributable to the structural change of the complex can be observed; however, by adding KCl to the 2 day old equilibrated sample, the original signals could be recovered. This unambiguously proves that only the exchange of the coordinating chloride ion by a water molecule occurred in a reversible manner without affecting the dissociation of the five-membered N,Nchelate.
For comparative biological studies, two additional Ru(II) complexes containing nonsugar based ligands (Scheme 3, Ru-13 and Ru-14) were also synthesized, starting from the Ru-dimer with 1-phenyl-4-(pyridine-2-yl)-1,2,3-triazole [34] (L-13) and 2-phenyl-5-(pyridine-2-yl)-1,3,4-oxadiazole [35,36] (L-14), respectively. The aqueous stability of the complexes was also studied over time. As an example, the time dependence of the NMR spectra of Ru-3a is shown in Figure S1. The small shifts of signals attributable to the structural change of the complex can be observed; however, by adding KCl to the 2 day old equilibrated sample, the original signals could be recovered. This unambiguously proves that only the exchange of the coordinating chloride ion by a water molecule occurred in a reversible manner without affecting the dissociation of the five-membered N,Nchelate.

Identification of Ruthenium Compounds with Antineoplastic Properties
We screened 14 ligands and their ruthenium complexes, bringing up the number of compounds to 28 in a concentration range of 100-0.0017 µ M. The compounds were tested in an assay aiming to assess short-term toxicity (methylthiazolyldiphenyl-tetrazolium bromide (MTT) reduction assay, 4 h) and long-term cytostasis or cytotoxicity (sulforhodamine B (SRB) proliferation assay, 48 h) on A2780 human ovarian carcinoma cells. We found four complexes, Ru-2a, Ru-4, Ru-6, and Ru-8, possessing cytostatic properties in both long-term SRB and short-term MTT assays ( Figure 3A

Identification of Ruthenium Compounds with Antineoplastic Properties
We screened 14 ligands and their ruthenium complexes, bringing up the number of compounds to 28 in a concentration range of 100-0.0017 µM. The compounds were tested in an assay aiming to assess short-term toxicity (methylthiazolyldiphenyl-tetrazolium bromide (MTT) reduction assay, 4 h) and long-term cytostasis or cytotoxicity (sulforhodamine B (SRB) proliferation assay, 48 h) on A2780 human ovarian carcinoma cells. We found four complexes, Ru-2a, Ru-4, Ru-6, and Ru-8, possessing cytostatic properties in both long-term SRB and short-term MTT assays ( Figure 3A,B). The uncomplexed ligands L-2a, L-4, L-6, and L-8 had no biological activity ( Figure 3A,B).

Ruthenium Compounds Have Similar Inhibitory Characteristics to Platinum Compounds
Next, we assessed the four active complexes, Ru-2a, Ru-4, Ru-6, and Ru-8, and the corresponding ligands, L-2a, L-4, L-6, and L-8, in detail. All compounds were tested on two ovarian cancer cell lines (A2780 and ID8) and on human primary skin fibroblasts (non-transformed, primary cells) in MTT and SRB assays. MTT assays were performed 4 h post treatment and indicated rapid toxicity of the compounds, while SRB assays were performed 2 days post treatment and represented long-term cytostasis or toxicity.
None of the free ligands exerted rapid toxicity on any of the cell lines in short-term MTT assays ( Figure 4). Nevertheless, the application of Ru-2a and Ru-8 above 10 µM concentration reduced the MTT signal in A2780 and ID8 cells. Ru-6 reduced the MTT signal in ID8 cells at 100 µM, and a similar trend was observed on A2780 cells. Although it was not statistically significant, Ru-4 led to a similar reduction in MTT signal at 100 µM in A2780 and ID8 cells. Complexes Ru-2a, Ru-4, Ru-6, and Ru-8 had no effect on primary fibroblasts in MTT assays.  Statistical significance was determined using a two-way ANOVA test, and all measurement points were compared with each other; *, **, and *** indicate statistically significant differences between vehicle-treated (control) and free ligand/ruthenium complex-treated cells at p < 0.05, p < 0.01, and p < 0.001, respectively; # , ## , and ### indicate statistically significant differences between free ligand-treated and ruthenium complex-treated cells at p < 0.05, p < 0.01, and p < 0.001, respectively. Values were normalized to vehicle-treated cells, where the absorbance of vehicle-treated cells was equal to 1. Statistical significance was determined using a two-way ANOVA test, and all measurement points were compared with each other; *, **, and *** indicate statistically significant differences between vehicle-treated (control) and free ligand/ruthenium complex-treated cells at p < 0.05, p < 0.01, and p < 0.001, respectively; # , ## , and ### indicate statistically significant differences between free ligand-treated and ruthenium complex-treated cells at p < 0.05, p < 0.01, and p < 0.001, respectively. Values were normalized to vehicle-treated cells, where the absorbance of vehicle-treated cells was equal to 1.
Ruthenium complexes Ru-2a, Ru-4, Ru-6, and Ru-8 were cytostatic, while none of the corresponding ligands (L-2a, L-4, L-6, and L-8) had cytostatic properties in long-term SRB assays on A2780 and ID8 cells ( Figure 5). On primary human fibroblasts, the ligands, L-2a, L-4, L-6, and L-8, as well as two complexes, Ru-4 and Ru-8, had no effect. In contrast, Ru-2a and Ru-6 reduced the SRB signal at 100 µM on fibroblasts. Statistical significance was determined using a two-way ANOVA test, and all measurement points were compared with each other; * and *** indicate statistically significant differences between vehicle-treated (control) and free ligand/ruthenium complex-treated cells at p < 0.05, p < 0.01, and p < 0.001, respectively; # and ### indicate statistically significant differences between free ligand-treated and ruthenium complex-treated cells at p < 0.05, p < 0.01, and p < 0.001, respectively. Nonlinear regression was performed on the data. Values were normalized to vehicle treated cells, where the absorbance of vehicle-treated cells was equal to 1. Statistical significance was determined using a two-way ANOVA test, and all measurement points were compared with each other; * and *** indicate statistically significant differences between vehicletreated (control) and free ligand/ruthenium complex-treated cells at p < 0.05, p < 0.01, and p < 0.001, respectively; # and ### indicate statistically significant differences between free ligand-treated and ruthenium complex-treated cells at p < 0.05, p < 0.01, and p < 0.001, respectively. Nonlinear regression was performed on the data. Values were normalized to vehicle treated cells, where the absorbance of vehicle-treated cells was equal to 1.
Cytostasis in the long term can be due to enhanced cell death. To exclude that possibility, we performed annexin V-propidium iodide (PI) double staining. Treating A2780 cells with Ru-2a, Ru-4, Ru-6, and Ru-8 did not largely increase the proportions of PI-positive, annexin V-positive, and double-positive cells in contrast to hydrogen peroxide that was used as a positive control whether at 2 h, 4 h, or 48 h post treatment ( Figure 6). The 4 h and 48 h timepoints were chosen to be the same as those used in the previous assays. The 2 h timepoint was found in our previous study to be optimal for the induction of apoptosis, marked by phosphatidylserine exposure [37][38][39]. , and cells were subjected to flow cytometry as described in Section 5. The percentages of cells in the quadrants are plotted. Data are presented as the average ± SEM from three biological replicates; individual assays were performed in duplicate. Normality was tested; on all datasets, Box-Cox normalization was performed to achieve a normal distribution. Statistical significance was determined using a two-way ANOVA test, and all measurement points were compared with each other; ** and *** indicate statistically significant differences between vehicle-treated (control) and treated cells (ruthenium complexes or 300 µ M H2O2) corresponding to the same quadrant (e.g., vehicle-treated double-negative cells vs. Ru-2a-treated double-negative cells) at p < 0.01 and p < 0.001, respectively.
Since ruthenium complexes are regarded as alternatives to platinum-centered drugs [6], we used the currently therapeutically available platinum-based drugs cisplatin (I), oxaliplatin (II), and carboplatin (III) as reference compounds and tested them on A2780 and ID8 cells, as well as on primary human fibroblasts. Platinum drugs had no effect in MTT assays ( Figure 7A), while they were cytostatic on A2780 and ID8 cells in SRB assays ( Figure 7B). Carboplatin and oxaliplatin had no effect on primary human fibroblasts in SRB assays, while cisplatin was cytostatic above 10 µ M (Figure 7). , and cells were subjected to flow cytometry as described in Section 5. The percentages of cells in the quadrants are plotted. Data are presented as the average ± SEM from three biological replicates; individual assays were performed in duplicate. Normality was tested; on all datasets, Box-Cox normalization was performed to achieve a normal distribution. Statistical significance was determined using a two-way ANOVA test, and all measurement points were compared with each other; ** and *** indicate statistically significant differences between vehicle-treated (control) and treated cells (ruthenium complexes or 300 µM H 2 O 2 ) corresponding to the same quadrant (e.g., vehicle-treated double-negative cells vs. Ru-2a-treated double-negative cells) at p < 0.01 and p < 0.001, respectively.
Since ruthenium complexes are regarded as alternatives to platinum-centered drugs [6], we used the currently therapeutically available platinum-based drugs cisplatin (I), oxaliplatin (II), and carboplatin (III) as reference compounds and tested them on A2780 and ID8 cells, as well as on primary human fibroblasts. Platinum drugs had no effect in MTT assays ( Figure 7A), while they were cytostatic on A2780 and ID8 cells in SRB assays ( Figure 7B). Carboplatin and oxaliplatin had no effect on primary human fibroblasts in SRB assays, while cisplatin was cytostatic above 10 µM (Figure 7).
In addition to the IC 50 value, regression analysis yielded the Hill slope. The Hill slope is a readout, characterized by the slope of the curve, that provides information on the binding characteristics of a compound [40]. The Hill slope of platinum compounds on A2780 and ID8 cells was in the range of 0.6-1.6. In stark contrast to that, the Hill slope of the ruthenium complexes was in the range of 1.5-3.7 (Table 3). A higher Hill slope suggests higher cooperativity upon the binding of the ruthenium complexes to target biomolecules than in the case of platinum compounds.  Table 3). The IC50 values of Ru-2a, Ru-4, Ru-6, and Ru-8 fell between 0.9 and 9 µ M, and the A2780 cells were more sensitive to ruthenium complexes as compared to ID8 cells (Table 3). Ru-2a had the lowest IC50 value in A2780 and ID8 cells (0.9 and 1.5 µ M, respectively). The IC50 values of the ruthenium complexes (0.9-9 µ M) were comparable to or lower than those of the platinum compounds (0.1-28 µ M) ( Table 3).
In addition to the IC50 value, regression analysis yielded the Hill slope. The Hill slope is a readout, characterized by the slope of the curve, that provides information on the binding characteristics of a compound [40]. The Hill slope of platinum compounds on A2780 and ID8 cells was in the range of 0.6-1.6. In stark contrast to that, the Hill slope of the ruthenium complexes was in the range of 1.5-3.7 (Table 3). A higher Hill slope suggests higher cooperativity upon the binding of the ruthenium complexes to target biomolecules than in the case of platinum compounds.  Table 3. The lipophilicity parameters (logD) and kinetic properties of the biologically active ruthenium complexes identified in the study and the reference platinum compounds (IC 50 (µM)). N/A-could not be calculated/had no effect.

The Biological Activity of Ruthenium Complexes Is Dependent on Reactive Oxygen Species Production
Currently available ruthenium complexes with antineoplastic activity have diverse modes of action involving (mitochondrial) reactive oxygen species production [9,15,41,42] and the induction of DNA damage [43,44]. We detected an increase in the expression of an oxidative stress marker, 4-hydroxy-nonenal (4HNE), at the level of the whole lane, as well as when a specific band was assessed, in A2780 cells treated with the active ruthenium complexes Ru-2a, Ru-4, Ru-6, and Ru-8 corresponding to the IC 50 values (Figure 8).   We assessed whether oxidative stress had a role in the cytostatic effects exerted by Ru-2a, Ru-4, Ru-6, and Ru-8. To that end, we tried to revert the cytostatic effects of ruthenium complexes using strong reductants such as reduced glutathione (GSH) and N-acetyl-cysteine (NAC). Furthermore, we also tested MitoTEMPO, a mitochondrially targeted antioxidant that can efficiently detoxify mitochondria-derived reactive oxygen species [45,46]. GSH and NAC cotreatment attenuated the cytostatic effects induced by Ru-2a, Ru-4, Ru-6, and Ru-8 ( Figure 9A) pointing to the causative role of reactive oxygen species production in cytostasis. Nevertheless, MitoTEMPO did not modulate the effects of Ru-2a, Ru-4, Ru-6, and Ru-8 ( Figure 9A) suggesting that mitochondria are not the source of the reactive oxygen species.
As thiols are soft Lewis bases and ruthenium is a soft Lewis acid, excess amounts of GSH or NAC can lead to the disassembly of ruthenium complexes. To provide evidence against this scenario, we applied another antioxidant, vitamin E, that does not have thiol groups. The application of vitamin E, similar to GSH and NAC, attenuated the cytostatic effects of the bioactive ruthenium complexes ( Figure 9B), providing further evidence on the involvement of reactive oxygen species. Interestingly, treatment of the cells with Trolox, a derivative of vitamin E lacking the apolar phytyl chain, did not provide protection against ruthenium compounds ( Figure 9C).
Reactive oxygen species production leads to DNA damage and poly(ADP-ribose) polymerase (PARP) activation [47,48]. Certain ruthenium complexes were shown to potentiate the effects of PARP inhibitors [41,42,49,50]. We assessed whether Ru-2a, Ru-4, Ru-6, and Ru-8 have similar properties by treating cells with 3 µM rucaparib, a clinically available potent PARP inhibitor. Rucaparib reduced cell proliferation in agreement with previous publications (e.g., [51]) ( Figure 10). When we performed nonlinear regression analysis to assess the IC 50 values, we found that rucaparib potentiated the effects of all ruthenium complexes marked by a decrease in the IC 50 values in combination treatment ( Figure 10). Nevertheless, it is of note that the potentiation was not as marked as the effects of antioxidants; hence, PARP-mediated effects probably have minor importance. Figure 8. Ruthenium complexes induce reactive oxygen species production. A total of 1 × 10 6 A2780 cells were plated on six-well plates. Cells were treated with Ru-2a at 1.3 µ M, Ru-4 at 6.2 µ M, Ru-6 at 8.5 µ M, and Ru-8 at 3.7 µ M for 48 h. Then, the cells were harvested, and the protein lysate was prepared. Subsequently, 20 µ g protein was separated by SDS-PAGE on an 8% gel, followed by Western blotting, and the blot was stained with 4HNE and actin antibodies. Densitometry values were normalized to the control. Data are represented as the average ± SEM from three biological replicates. Since there was no variance in the control group, we did not perform statistical analysis. Values were normalized to vehicletreated cells, where the absorbance of vehicle-treated cells was equal to 1.  . Reduced thiols and vitamin E can block the cytostatic effects of ruthenium complexes. (A) A total of 3 × 10 3 A2780 cells were plated on 96-well plates. Cells were treated with the reduced thiol compounds in the concentrations indicated for 48 h, and then the SRB assay was performed. Data are presented as the average ± SEM from three biological replicates. The individual SRB assays were performed in duplicate. Normality was tested; the Ru-4 dataset had a normal distribution, whereas Ru-2a, Ru-6, and Ru-8 dataset normality was achieved by logarithmic transformation. Statistical significance was determined using a two-way ANOVA test, and all measurement points were compared with each other. (B) A total of 3 × 10 3 A2780 cells were plated on 96-well plates. Cells were treated with the compounds in the concentrations indicated for 48 h, and then the SRB assay was performed. Data are presented as the average ± SEM from three biological replicates. The individual SRB assays were performed in duplicate. Normality was tested; the Ru-4 and Ru-8 datasets had a normal distribution, whereas Ru-2a and Ru-6 dataset normality was achieved by logarithmic transformation. (C) A total of 3 × 10 3 A2780 cells were plated on 96-well plates. Cells were treated with Trolox in the concentration indicated for 48 h, and then the SRB assay was performed. Data are presented as the average ± SEM from three biological replicates. The individual SRB assays were performed in duplicate. Normality was tested; the Ru-2a, Ru-6, and Ru-8 datasets had a normal distribution, whereas Ru-4 dataset normality was achieved by logarithmic transformation. Statistical significance was determined using a two-way ANOVA test, and all measurement points were compared with each other; *, **, and *** indicate statistically significant differences between vehicle-treated (control) and antioxidant (GSH, NAC, vitamin E, or Trolox)-treated cells at p < 0.05, p < 0.01, and p < 0.001, respectively; # , ## , and ### indicate statistically significant differences between vehicle-treated (control) and ruthenium-treated cells at p < 0.05, p < 0.01, and p < 0.001, respectively. Values were normalized to vehicle-treated cells, where the absorbance of vehicle-treated cells was equal to 1. available potent PARP inhibitor. Rucaparib reduced cell proliferation in agreement with previous publications (e.g., [51]) ( Figure 10). When we performed nonlinear regression analysis to assess the IC50 values, we found that rucaparib potentiated the effects of all ruthenium complexes marked by a decrease in the IC50 values in combination treatment ( Figure 10). Nevertheless, it is of note that the potentiation was not as marked as the effects of antioxidants; hence, PARP-mediated effects probably have minor importance.

Bioactive Ruthenium Complexes Can Cause Cytostasis in Breast Carcinoma, Pancreatic Adenocarcinoma, and Glioblastoma Cell Lines
Prior studies [22,23,52,53] have suggested that sugar-based ruthenium complexes may be active on other cell lines and, therefore, potentially in other neoplastic diseases. We assessed breast cancer (modelled by MCF7 cells) as prior data [22] suggested the potential effectiveness of ruthenium complexes. Pancreatic adenocarcinoma (modelled by Capan2 cells) and glioblastoma (modelled by U251 cells), similar to ovarian cancer, are usually malign diseases where treatment options are limited [54][55][56]. All complexes exerted cytostatic effects on all cell lines in proliferation assays (SRB) (Figure 11). The IC 50 values of the ruthenium complexes on U251, MCF7, and Capan2 were higher than on A2780 or ID8 cells (Table 3).

The Carbohydrate Moiety Is Important for the Bioactivity of the Ruthenium Complexes
To assess the role of the monosaccharide moiety in the bioactive compounds, we synthesized and assessed two molecules where the monosaccharide unit was substituted with a phenyl group (Ru-13/L-13 and Ru-14/L-14). Neither of the free ligands, L-13 or L-14, had cytostatic activity on A2780 cells in SRB assays ( Figure 12). The triazole-containing Ru-13 was cytostatic on A2780 cells with an IC 50 value of~12 µM, which was higher by one order of magnitude than the IC 50 value of the sugar-containing molecule Ru-2a ( Figure 12, Table 3). Furthermore, the Hill coefficient of Ru-13 was 1.5, suggesting no or little cooperativity. The oxadiazole-containing Ru-14 had no cytostatic activity ( Figure 12).
We assessed breast cancer (modelled by MCF7 cells) as prior data [22] suggested the po-tential effectiveness of ruthenium complexes. Pancreatic adenocarcinoma (modelled by Capan2 cells) and glioblastoma (modelled by U251 cells), similar to ovarian cancer, are usually malign diseases where treatment options are limited [54][55][56]. All complexes exerted cytostatic effects on all cell lines in proliferation assays (SRB) (Figure 11). The IC50 values of the ruthenium complexes on U251, MCF7, and Capan2 were higher than on A2780 or ID8 cells (Table 3). Figure 11. Assessment of the bioactive ruthenium complexes on models for glioblastoma, breast cancer, and pancreatic adenocarcinoma for cytostatic activity. A total of 2 × 10 3 U251 cells (glioblastoma model), 3 × 10 3 MCF7 cells (breast cancer), and 2 × 10 3 Capan2 (pancreatic adenocarcinoma) were plated on 96-well plates. Cells were treated with the compounds in the concentrations indicated for 48 h, and then the SRB assay was performed. Data are presented as the average ± SEM from three biological replicates; individual assays were performed in duplicate. Normality was tested; the Ru-4/L-4 U251, Capan2 dataset normality was achieved by logarithmic transformation, and the Ru-6/L-6 MCF7 dataset was normalized using the Box-Cox normalization method. Statistical significance was determined using a two-way ANOVA test, and all measurement points were compared with each other; *, **, and *** indicate statistically significant differences between vehicle-treated (control) and free Figure 11. Assessment of the bioactive ruthenium complexes on models for glioblastoma, breast cancer, and pancreatic adenocarcinoma for cytostatic activity. A total of 2 × 10 3 U251 cells (glioblastoma model), 3 × 10 3 MCF7 cells (breast cancer), and 2 × 10 3 Capan2 (pancreatic adenocarcinoma) were plated on 96-well plates. Cells were treated with the compounds in the concentrations indicated for 48 h, and then the SRB assay was performed. Data are presented as the average ± SEM from three biological replicates; individual assays were performed in duplicate. Normality was tested; Capan2 dataset normality was achieved by logarithmic transformation, and the Ru-6/L-6 MCF7 dataset was normalized using the Box-Cox normalization method. Statistical significance was determined using a two-way ANOVA test, and all measurement points were compared with each other; *, **, and *** indicate statistically significant differences between vehicle-treated (control) and free ligand/ruthenium complex-treated cells at p < 0.05, p < 0.01, and p < 0.001, respectively; # , ## , and ### indicate statistically significant differences between free ligand-treated and ruthenium complex-treated cells at p < 0.05, p < 0.01, and p < 0.001, respectively. Nonlinear regression was performed on the data. Values were normalized to vehicle-treated cells, where the absorbance of vehicle-treated cells was equal to 1. with a phenyl group (Ru-13/L-13 and Ru-14/L-14). Neither of the free ligands, L-13 or L-14, had cytostatic activity on A2780 cells in SRB assays ( Figure 12). The triazole-containing Ru-13 was cytostatic on A2780 cells with an IC50 value of ~12 µ M, which was higher by one order of magnitude than the IC50 value of the sugar-containing molecule Ru-2a (Figure 12, Table 3). Furthermore, the Hill coefficient of Ru-13 was 1.5, suggesting no or little cooperativity. The oxadiazole-containing Ru-14 had no cytostatic activity (Figure 12).

Figure 12.
Assessment of non-sugar ruthenium complexes on A2780 cells for cytostatic activity. A total of 3 × 10 3 A2780 cells were plated on 96-well plates. Cells were treated with the compounds in the concentrations indicated for 48 h, and then the SRB assay was performed. Data are represented as the average ± SEM, from three biological replicates; individual assays were performed in duplicate. Normality was tested; both datasets were normalized using the Box-Cox normalization method. Statistical significance was determined using a two-way ANOVA test, and all measurement points were compared with each other; *** indicates statistically significant differences between vehicle-treated (control) and free ligand/ruthenium complex-treated cells at p < 0.001. Nonlinear regression was performed on the data. Values were normalized to vehicle-treated cells, where the absorbance of vehicle-treated cells was equal to 1. Figure 12. Assessment of non-sugar ruthenium complexes on A2780 cells for cytostatic activity. A total of 3 × 10 3 A2780 cells were plated on 96-well plates. Cells were treated with the compounds in the concentrations indicated for 48 h, and then the SRB assay was performed. Data are represented as the average ± SEM, from three biological replicates; individual assays were performed in duplicate. Normality was tested; both datasets were normalized using the Box-Cox normalization method. Statistical significance was determined using a two-way ANOVA test, and all measurement points were compared with each other; *** indicates statistically significant differences between vehicle-treated (control) and free ligand/ruthenium complex-treated cells at p < 0.001. Nonlinear regression was performed on the data. Values were normalized to vehicle-treated cells, where the absorbance of vehicle-treated cells was equal to 1.

Discussion
In this study, we screened a set of carbohydrate-based half-sandwich type organoruthenium complexes. Four compounds were identified to display long-term cytostatic effects but little rapid toxicity on two different ovarian cancer cell lines. These compounds were not toxic toward primary human fibroblasts. The practical absence of toxicity on non-transformed fibroblasts aligns well with the low toxicity of other ruthenium complexes [9][10][11][12]. The activity of the compounds was dependent on the presence of ruthenium(II), as the ligands had no biological activity. The IC 50 values of the active compounds were comparable or superior to the currently applied platinum compounds (cisplatin, oxaliplatin, and carboplatin) and other sugar-containing ruthenium complexes [22,23,52,53]. The identified Ru(II) compounds were active upon long-term application in SRB assays, similar to platinum compounds.
We assessed the relationship between the biological activity and the structure of the active molecules. We compared all complexes to Ru-4 that had considerable cytostatic potential in SRB assays on ovarian cancer cells but no activity on primary fibroblasts ( Figure 13). An important structural feature that plays a key role in the biological activity of the molecules is the presence and size of the protecting groups on the hydroxyl groups of the carbohydrate moieties. All bioactive molecules (Ru-2a, Ru-4, Ru-6, and Ru-8) have O-benzoyl groups (Figure 13). Changing the O-benzoyl groups to smaller O-acetyl groups (Ru-4 vs. Ru-5) or complete deprotection (Ru-4 vs. Ru-10) abrogated the inhibitory activity. These findings are similar to the observations of Hamala and colleagues [22], who showed that the inhibitory activity of carbohydrate-based ruthenium complexes (Figure 1, X) was enhanced when the hydroxyl groups of the sugar units were protected by esters. Furthermore, increasing the length of the acyl chain improved the inhibitory efficacy (acetyl < propionyl < butyryl). In good agreement with that, the replacement of the protected monosaccharide in the molecule with a single phenyl group lowered the cytostatic capacity of the ruthenium complex  or fully abrogated it . This finding also corroborates the inevitable role of the sugar moiety in determining the biological activity. Apparently, increasing the lipophilic character of the compounds improves the cytostatic properties of sugar-based ruthenium complexes. This is underlined by the logD values in Tables 3 and S9 showing a significant difference in lipophilicity in favor of the benzoylated derivatives. Hanif and colleagues [52] also had similar findings with RAPTA analogs as a function of the size of the arene cap and the acetal protecting group. It is tempting to speculate that these large, apolar protective groups could facilitate the membrane permeability of these compounds. In good agreement with that, Trolox, a derivative of vitamin E that lacks the long apolar phytyl chain, was ineffective in protecting cells against cytostasis induced by ruthenium compounds. This finding suggests oxidative stress in apolar, lipid-containing compartments. Furthermore, we found two other structural features that impacted bioactivity: (A) modification of the carbohydrate moiety from glucose to xylose by a formal removal of the hydroxymethyl group at position 5  or changing the configuration of the C-4 center (glucose to galactose as in Ru-8), and (B) replacement of the 1,3,4-oxadiazole ring by 1,2,3-triazole (Ru-2a). These changes increased the rapid toxicity of the molecules or rendered the molecules toxic on primary cells (Figure 13). In addition, a triazole ring in the molecule ensured better cytostatic properties than an oxadiazole ring (Ru-13 vs. Ru-14 and Ru-2a vs. Ru-4).
It was a surprising observation that the binding of the active compounds showed a high level of cooperativity, as suggested by the value of the Hill slope, being 2-3 for Ru-2a, Ru-4, Ru-6, and Ru-8. This suggests that the binding of a ruthenium complex facilitates the binding of the subsequent molecules. The Hill slope was also determined for the reference platinum-based drugs; however, these molecules did not show signs of cooperative binding as the Hill slope was ~1. Apparently, the binding and, probably, the mode of action of the ruthenium complexes identified in this study differ from those of the reference platinum compounds. In the study of Hanif et al. [52] assessing RAPTA analogs (general Furthermore, we found two other structural features that impacted bioactivity: (A) modification of the carbohydrate moiety from glucose to xylose by a formal removal of the hydroxymethyl group at position 5  or changing the configuration of the C-4 center (glucose to galactose as in Ru-8), and (B) replacement of the 1,3,4-oxadiazole ring by 1,2,3-triazole (Ru-2a). These changes increased the rapid toxicity of the molecules or rendered the molecules toxic on primary cells (Figure 13). In addition, a triazole ring in the molecule ensured better cytostatic properties than an oxadiazole ring (Ru-13 vs. Ru-14 and Ru-2a vs. Ru-4).
It was a surprising observation that the binding of the active compounds showed a high level of cooperativity, as suggested by the value of the Hill slope, being 2-3 for Ru-2a, Ru-4, Ru-6, and Ru-8. This suggests that the binding of a ruthenium complex facilitates the binding of the subsequent molecules. The Hill slope was also determined for the reference platinum-based drugs; however, these molecules did not show signs of cooperative binding as the Hill slope was~1. Apparently, the binding and, probably, the mode of action of the ruthenium complexes identified in this study differ from those of the reference platinum compounds. In the study of Hanif et al. [52] assessing RAPTA analogs (general formula VIII in Figure 1), one compound was identified with a steep inhibitory curve, suggesting cooperative binding. This observation implicates that cooperative binding is not characteristic for all sugar-containing ruthenium complexes.
Our primary aim was to assess the applicability of ruthenium complexes in ovarian cancer, as these compounds were intended to be used as substitutes to platinum complexes [6]. Nevertheless, we provided evidence that the ruthenium complexes are also active in cell models of glioblastoma, breast cancer, and pancreatic adenocarcinoma, although the complexes showed preference toward ovarian cancer cells. Other sugar-containing ruthenium complexes were also active in ovarian cancer cell lines (A2780 [22,53], SK-OV-3 [22], CH1 [52,53]) further underlying the applicability of such compounds in ovarian cancer. Nevertheless, Ru-sugar complexes showed a cytostatic response in MDA-MD-231 breast cancer cells [22], colon cancer [52,53], non-small-cell lung cancer [52], and cervical carcinoma (HeLa) cells [23], suggesting a wider applicability of such compounds.
Apparently, the mode of action of ruthenium complexes is pleiotropic and involves binding to polynucleotides [44] or the production of reactive oxygen species [9,15,41,42]. The actual mode of action or the dominant pathway to induce cytostasis differs for different classes of ruthenium complexes [15]. Carbohydrate-ruthenium complexes were shown to induce apoptosis that we did not detect for our compounds in concentrations corresponding to the IC 50 values [22]. The active compounds identified in this study led to oxidative stress marked by increased 4HNE expression, a marker for lipid peroxidation. The functional role of reactive oxygen species production was confirmed by the application of GSH and NAC, two strong reductants, that reverted the cytostatic effects of the bioactive ruthenium complexes, as well as confirmed by using vitamin E. The source of reactive oxygen species was other than the mitochondria. Reactive oxygen species production was shown to be cytostatic in numerous carcinomas [9,15,41,42,57]. In addition to oxidative stress, we showed that the effects of ruthenium complexes is potentiated by a PARP inhibitor, rucaparib, similar to others [41,42,49,50]. Nevertheless, the level of potentiation was lower than the anticytostatic effects of antioxidants, pointing to a lower importance for PARP-mediated pathways.
In the biological study of the complexes in ovarian adenocarcinoma cells, antineoplastic properties characterized by little acute toxicity but long-term cytostasis were identified. The bioactive ruthenium complexes had micromolar IC 50 values on A2780 and ID8 cells, while they had little or no activity on primary, non-transformed human fibroblasts, highlighting the low toxicity and selectivity of these compounds toward transformed cancer cells. The presence of the sugar moiety equipped with large hydrophobic protective groups on the hydroxyl groups of was found to be crucial for the biological activity. The bioactive ruthenium complexes, identified herein, showed cooperative binding to yet unidentified cellular target(s) and induced oxidative stress that was essential for their cytostatic activity.

General Procedure III for the Zemplén Deacylation
An O-acyl protected glycosyl-azole was dissolved in a 1:1 mixture of dry MeOH and dry CHCl 3 (1 mL/25 mg substrate), and a few drops of a~1 M solution of NaOMe in MeOH was added (pH = 8-9). The reaction mixture was left at r.t. until the TLC indicated total conversion of the starting material. The neutralization of the solution was carried out by the addition of a cation exchange resin (Amberlyst 15, H + form). The resin was then filtered off, and the solution was evaporated under reduced pressure. The crude product was purified by column chromatography or crystallization.

General Procedure IV for the Formation of the [(η 6 -p-cym)Ru II (N-N))Cl]PF 6 Type Complexes Containing O-peracylated Glycosyl Azole Ligands
To a solution of ([(η 6 -p-cym)RuCl 2 ] 2 ) dimer (Ru-dimer) in CH 2 Cl 2 (1 mL/10 mg dimer), the corresponding O-peracylated glycosyl azole (2-2.3 eq.) and TlPF 6 (2 eq.) were added. Under stirring, 1 mL of methanol was added to this reaction mixture in order to accelerate the precipitation of the TlCl. The initial red solution turned to yellow, indicating the formation of the [(η 6 -p-cym)Ru II (N-N)Cl]PF 6 type complex. The reaction mixture was then stirred at r.t. for an additional hour, and the total disappearance of the Ru-dimer was judged by TLC (9:1 CHCl 3 -MeOH). After completion of the complexation reaction, the precipitated TlCl was filtered off, and the solution was evaporated in vacuo. The crude complex was purified by column chromatography or crystallization. In a solution of ([(η 6 -p-cym)RuCl 2 ] 2 ) dimer (Ru-dimer) in CH 2 Cl 2 (1 mL/10 mg dimer), the corresponding unprotected glycosyl azole (2 eq.) and TlPF 6 (2 eq.) were suspended. Under stirring, 1 mL of methanol was added to this reaction mixture in order to dissolve the heterocyclic sugar derivative and accelerate the precipitation of the TlCl. The initial red solution turned to yellow, indicating the formation of the [(η 6 -p-cym)Ru II (N-N)Cl]PF 6 type complex. The reaction mixture was then stirred at r.t. for an additional hour, and the total disappearance of the Ru-dimer was judged by TLC (9:1 CHCl 3 -MeOH). After completion of the complexation reaction, the precipitated TlCl was filtered off, and the solution was evaporated in vacuo. The crude complex was purified by crystallization. The 1-(β-D-glucopyranosyl)-4-(pyridin-2-yl)-1,2,3-triazole (L-3a, 20.0 mg, 0.065 mmol) was suspended in dry pyridine (0.5 mL), and benzoyl chloride (36 µL, 0.310 mmol) was added. The reaction mixture was stirred at 60 • C until the TLC (3:2 EtOAc-hexane) showed complete disappearance of the starting material (1 h). The solvent was removed under diminished pressure. The residue was dissolved in CH 2 Cl 2 (20 mL) and extracted with sat. aq. NaHCO 3 (10 mL) and then with water (10 mL). The organic phase was dried over     The 2-(β-D-galactopyranosyl)-5-(pyridin-2-yl)-1,3,4-oxadiazole (L-10, 20.0 mg, 0.065 mmol) was suspended in dry pyridine (0.5 mL), and benzoyl chloride (36 µL, 0.310 mmol) was added. The reaction mixture was stirred at 60 • C until the TLC (3:2 EtOAc-hexane) showed complete disappearance of the starting material (1 h). The solvent was removed under diminished pressure. The residue was dissolved in CH 2 Cl 2 (20 mL) and extracted with sat. aq. NaHCO 3 (10 mL) and then with water (10 mL). The organic phase was dried over  released by adding 100 µL of 10 mM Tris base. Plates were measured in a plate photometer (Thermo Scientific Multiscan GO spectrophotometer, Waltham, MA, USA) at 540 nm. On each plate, wells were designed to contain untreated cells. In calculations, the readings for these wells were considered as 1, and all readings were expressed relative to these values.

Annexin V-FITC, PI Double Staining
Annexin V-PI double staining was applied to assess apoptotic and necrotic cell death similar to [37,38,69]. A2780 cells were treated with the indicated compounds at the concentration corresponding to their IC 50 value for 2, 4, and 48 h. The 4 and 48 h time points correspond to the time points for the MTT and SRB assays, while the 2 h time point reflects the optimum time point for the detection of apoptotic or necrotic cell death [37,38,69]. Quadrants were set on the basis of the FITC and PI values observed for the vehicle-treated cells.

Western Blotting
The preparation of protein extracts, the separation of protein extracts in SDS-polyacrylamide gel electrophoresis, and the subsequent Western blotting were performed as described in [70] using the antibodies in Table 4. Enhanced chemiluminescence was developed using ChemiDoc Imager, (Bio-Rad, Hercules, CA, USA). Densitometry was performed using Image Lab Touch Software, Bio-Rad, Hercules, CA, USA).