Organoruthenium Complexes with Benzo-Fused Pyrithiones Overcome Platinum Resistance in Ovarian Cancer Cells

Simple Summary Ovarian cancer is the fifth most common cancer in the developing world, with many front-line treatments combining paclitaxel with platinum-based anticancer agents cisplatin/carboplatin. However, increased incidence of platinum resistance demands the development of new chemotherapeutic agents. The aim of our study was to explore a new family of organoruthenium(II) pyrithione complexes for their efficacy towards platinum-resistant ovarian cancer cells. We confirmed that this new class of compounds remain highly potent towards platinum-insensitive cells and appear to work by a mechanism that is not common to platinum agents. Abstract Drug resistance to existing anticancer agents is a growing clinical concern, with many first line treatments showing poor efficacy in treatment plans of some cancers. Resistance to platinum agents, such as cisplatin, is particularly prevalent in the treatment of ovarian cancer, one of the most common cancers amongst women in the developing world. Therefore, there is an urgent need to develop next generation of anticancer agents which can overcome resistance to existing therapies. We report a new series of organoruthenium(II) complexes bearing structurally modified pyrithione ligands with extended aromatic scaffold, which overcome platinum and adriamycin resistance in human ovarian cancer cells. The mechanism of action of such complexes appears to be unique from that of cisplatin, involving G1 cell cycle arrest without generation of cellular ROS, as is typically associated with similar ruthenium complexes. The complexes inhibit the enzyme thioredoxin reductase (TrxR) in a model system and reduce cell motility towards wound healing. Importantly, this work highlights further development in our understanding of the multi-targeting mechanism of action exhibited by transition metal complexes.


Introduction
Ovarian cancer (OC), also known as "the silent killer", belongs to the most lethal gynecologic malignancies [1][2][3][4][5] and is the fifth most common cancer type in women in the developed world [6]. Although the mortality of patients with an early-stage OC is low, diagnosis at this stage is rather poor due to less obvious and non-specific symptoms. However, once diagnosed with the advanced OC, the survival rate decreases immensely, also due to development of chemotherapy resistance [5]. Thus far, five histological subtypes of OC are known, of which epithelial tumors are the most common, accounting for approximately 90% [7]. Several genetic (BRCA1/2 gene mutations, DNA mismatched repair gene mutations) and epidemiological risk factors (age, obesity, talcum powder use/asbestos exposure) contribute to the development and the progression of OC [5,8].

Chemicals
Starting materials and other reagents as well as the solvents for the synthesis of ligands a-b and complexes 1a-b and 2a-b were purchased from commercial suppliers (Fluorochem, Hadfield, UK; Strem Chemicals, Inc., Newburyport, MA, USA; Merck, Darmstadt, Germany; Honeywell, Muskegon, MI, USA) and used as received, except from CHCl3 used for N-oxidation, which was dried over Na2SO4 prior to use. Pta ligand (1,3,5triaza-7-phosphaadamantane) was prepared as reported [45]. The progress of the reaction was followed by pre-coated TLC sheets ALUGRAM ® SIL G/UV254 (Macherey-Nagel, Düren, Germany) under UV light, and column chromatography was carried out on Merck Silica gel 60 (35-70 µm) as stationary phase.

Physicochemical Characterization
1 H and 31 P NMR spectra were obtained with Bruker Avance III 500 spectrometer at room temperature at 500 MHz and 202 MHz, respectively. Chemical shifts (δ) are reported in ppm. 1 H NMR spectra are referenced to residual peaks of NMR solvent CD2Cl2 or D2O at 5.32 (referenced against the central line of triplet) or 4.79 ppm, respectively, and chemical shifts of 31 P NMR spectra relative to external standard. The multiplicities are abbreviated as s = singlet, d = doublet, dd = doublet of the doublets, sept = septet, and m = multiplet. Coupling constants (J) are given in Hz. MestReNova version 11.0.4 was used for NMR data processing. Infrared spectra were recorded with Bruker FTIR Alpha Platinum ATR spectrometer. High resolution mass spectra (HRMS) were carried out on an Agilent 6224 Accurate Mass TOF LC/MS instrument. Elemental analyses were obtained on a Perkin-Elmer 2400 II instrument (C, H, N). UV-Vis spectra of complexes were obtained on

Chemicals
Starting materials and other reagents as well as the solvents for the synthesis of ligands a-b and complexes 1a-b and 2a-b were purchased from commercial suppliers (Fluorochem, Hadfield, UK; Strem Chemicals, Inc., Newburyport, MA, USA; Merck, Darmstadt, Germany; Honeywell, Muskegon, MI, USA) and used as received, except from CHCl 3 used for N-oxidation, which was dried over Na 2 SO 4 prior to use. Pta ligand (1,3,5-triaza-7-phosphaadamantane) was prepared as reported [45]. The progress of the reaction was followed by pre-coated TLC sheets ALUGRAM ® SIL G/UV254 (Macherey-Nagel, Düren, Germany) under UV light, and column chromatography was carried out on Merck Silica gel 60 (35-70 µm) as stationary phase.

Physicochemical Characterization
1 H and 31 P NMR spectra were obtained with Bruker Avance III 500 spectrometer at room temperature at 500 MHz and 202 MHz, respectively. Chemical shifts (δ) are reported in ppm. 1 H NMR spectra are referenced to residual peaks of NMR solvent CD 2 Cl 2 or D 2 O at 5.32 (referenced against the central line of triplet) or 4.79 ppm, respectively, and chemical shifts of 31 P NMR spectra relative to external standard. The multiplicities are abbreviated as s = singlet, d = doublet, dd = doublet of the doublets, sept = septet, and m = multiplet. Coupling constants (J) are given in Hz. MestReNova version 11.0.4 was used for NMR data processing. Infrared spectra were recorded with Bruker FTIR Alpha Platinum ATR spectrometer. High resolution mass spectra (HRMS) were carried out on an Agilent 6224 Accurate Mass TOF LC/MS instrument. Elemental analyses were obtained on a Perkin-Elmer 2400 II instrument (C, H, N). UV-Vis spectra of complexes were obtained on PerkinElmer LAMBDA 750 UV/Vis/near-IR spectrophotometer and UV- Vis stability spectra on UV-Vis spectrophotometer UV-2600 Shimadzu. X-ray diffraction data were collected on an Oxford Diffraction SuperNova diffractometer with Mo/Cu microfocus X-ray source (K α radiation, λ Mo = 0.71073 Å, λ Cu = 1.54184 Å) with mirror optics and an Atlas detector at 150(2) K. The structures were solved in Olex 2 graphical user interface [46] by direct methods implemented in SHELXT and refined by a full-matrix least-squares procedure based on F 2 using SHELXL [47]. All non-hydrogen atoms were refined anisotropically. The hydrogen atoms were placed at calculated positions and treated using appropriate riding models. The crystal structures were submitted to the CCDC and were allocated the deposition numbers 2065161-2 065162 for ligand a and complex 2b, respectively. Cyclic voltammetry (CV) was performed using an Autolab PGSTAT100N potentiostat, processed with Nova 2.0 software, cleaned with Microsoft Excel, and graphs were created using SigmaPlot 13.0. Measurements for complexes 1a and 2a and ligand a (1.0 mM, CH 3 CN containing tetrabutylammonium hexafluorophosphate (0.1 M) as supporting electrolyte) were carried out using degassed solutions under nitrogen, and voltammograms were scanned from +2.0 V to -2.0 V. Alternatively, voltammograms were scanned from +0.8 V to -0.8 V to mimic biological redox potentials. In a typical electrochemical experimental set up, a three-electrode system was used: a glassy carbon electrode as the working electrode, Ag/AgCl as the reference electrode ((+7 mV vs. Fc + /Fc) + 205 mV vs. N.H.E.), and platinum wire as the counter electrode. For each electrode, CV was performed at a ν = 100 mVs -1 .

Syntheses of Ligands and Complexes
Ligands a-b as well as organoruthenium(II) chlorido complexes 1a-b were resynthesized according to published protocols [44,48], while organoruthenium(II) pta complexes 2a-b were newly prepared following our published protocol [41] as described below (general synthesis pathways presented in Supplementary Materials, Figure S1) and physiochemically characterized ( 1 H NMR spectra can be found in Supplementary Materials, Figures S13 and S14).
Appropriate chlorido complex 1a-b (50 mg, 0.112 mmol, 1 mol. equiv.) was dissolved in dichloromethane (30 mL) and finely ground pta ligand (0.168 mmol, 1.5 mol. equivalents) as well as NH 4 PF 6 (0.168 mmol, 1.5 mol. equivalents) were added and stirred in the dark for 48 h at ambient temperature. After the reaction was completed, the solvent was partly evaporated and filtered through a pad of Celite to remove side product NH 4 Cl and excess of NH 4 PF 6 and pta. Then, the mother liquor was concentrated on a rotary evaporator to approximately 2 mL. Cold diethyl ether was added to the oily residue, which resulted in precipitation of orange solid, which was washed with diethyl ether and dried at 45 • C overnight.

Aqueous Stability
For NMR stability, approximately 4 mg of compound was first dissolved in (CD 3 ) 2 SO to which D 2 O was added to obtain 5% (CD 3 ) 2 SO/D 2 O final solution (600 µL). In case of the stability experiments with NaCl, NaCl was first dissolved in D 2 O, and a latter solution was consequently added to (CD 3 ) 2 SO solution with the complex to obtain 5% (CD 3 ) 2 SO/D 2 O final solution containing 140 mM NaCl (600 µL). 1 H NMR spectra were recorded immediately after the preparation of the solutions and later at the selected timepoints.
UV-Vis stability was studied over 24 h using UV-visible spectroscopy in biologically relevant matrixes: a) PBS, b) RPMI-1640, c) fully prepared RPMI-1640 which included the addition of 10% v/v fetal calf serum and 1% v/v pen/strep antibiotics, and d) human blood plasma. DMSO stock solutions of tested complexes were prepared and further diluted in the mentioned matrixes. UV-Vis spectra were obtained immediately after the preparation of solutions and after 24 h between 250 and 900 nm using single beam scans with background correction. Between these two measurement timepoints, samples were kept in sealed cuvettes at 37 • C.

Determination of Antiproliferative Activity
Using flat-bottom 96-well plates, 5 × 10 3 cells were seeded per well in drug-free media and incubated at 37 • C K for 48 h. Prior to the treatment of cells, a stock solution of a test compound was prepared using a 1:1 mixture of cell culture medium and saline (0.9%) and a starting concentration of 5% v/v DMSO to aid complex solubility. Serial dilutions of the stock solution were prepared to ensure the final working DMSO concentration did not exceed 0.5% v/v. Cells were treated with six concentrations of test complex between 200 and 0.01 µM for 24 h at 37 • C, then the supernatant solution was aspirated, cells were washed with PBS, and cells were re-incubated in fresh (drug-free) media. After 72 h further incubation at 37 • C (so-called "recovery time"), cell viability was determined using the MTT assay (4 h dark exposure to MTT reagent). After this time, dye was solubilized in DMSO, and absorbance measurements were obtained using a microplate reader. Cell viabilities and respective IC 50 concentrations (the concentration of tested complex which caused 50% growth inhibition) were calculated relative to untreated controls as part of duplicate of triplicate experiments in two independent experiments.

Induction of Apoptosis
Briefly, 2 × 10 5 A2780 human ovarian cancer cells were seeded in flat-bottom 6-well plates and incubated at 37 • C for 24 h in a CO 2 humidified atmosphere. Cells were treated Cancers 2021, 13, 2493 6 of 21 with 1 µM concentrations of complex 1a in cell culture media (DMSO not exceeding 0.5% v/v), prepared as described previously. After 24 h incubation, the supernatant solution was removed by aspiration, and adherent cells were washed with PBS and collected using trypsin. After quenching of trypsin activity with culture media (containing FCS), cells were resuspended as a single cell solution in buffer containing propidium iodide (PI, Ex. 560 nm, Em. 595 nm) and Annexin-V-FITC conjugate (Ex. 485 nm, Em. 535 nm). Separately, negative (untreated) and positive (1 µg/mL staurosporine) control samples were prepared. No cellular fixation protocols were employed to circumvent non-specific binding of Annexin-V-FITC. Samples were analyzed using flow cytometry with gating determined using positive and negative control samples. The experiment was carried out as three independent biological replicates.

Cell Cycle Analysis
Briefly, 1 × 10 6 A2780 human ovarian cancer cells were seeded using flat-bottom 6-well plates and incubated at 37 • C for 24 h. Cells were treated with 1 µM concentrations of complex 1a in cell culture media (DMSO not exceeding 0.5% v/v), prepared as described previously. After 24 h incubation at 37 • C, the supernatant solution was removed by aspiration, and adherent cells were washed with PBS and collected using trypsin. Negative (untreated) control samples were also prepared. After quenching of trypsin activity with culture media (containing FCS), cells were resuspended in ice-cold ethanol for 2 h. Ethanol was removed by centrifugation, and cell pellets were suspended in fixation buffer containing propidium iodide (PI, Ex. 560 nm, Em. 595 nm) and RNAse A. Cells were analyzed using flow cytometry, and data were processed using Flowjo for Windows. Experiments were carried out as three independent biological replicates.

Wound Healing Assay
Briefly, 24-well plates were seeded using 1 × 10 4 A2780 human ovarian cancer cells per well and incubated at 37 • C for 24 h. After this time, a sterile pipette tip was used to create two "wounds" per well, and cells were treated with 1 µM complex 1a for 24 h. The supernatant solution was aspirated, and cells were washed with PBS and stained using crystal violet solution (10% ethanol). After staining, excess stain was removed using PBS, and cells were visualized using a transmission microscope using 4× optical zoom. Values were calculated out of 36 random measurements of three wounds (12 measurements in each wound).

Colony Formation Assay
A2780 cancer cells were seeded in P100 flasks at a density of 1 × 10 3 and incubated for 24 h at 37 • C before administration of 1 µM concentration of complex 1a in culture medium (DMSO not exceeding 0.5% v/v), prepared as described previously. After this time, cells were washed and allowed a further seven days of growth at 37 • C in fresh media. Then, colonies were counted. This experiment was carried out in duplicate with independent triplicate experiments.

Induction of Reactive Oxygen Species (ROS)
Briefly, 1 × 10 4 A2780 human ovarian cancer cells were seeded using 96-well black plates and incubated at 37 • C for 24 h in a humidified atmosphere. Cells were treated with 1 µM concentrations of complex 1a in cell culture media (DMSO not exceeding 0.5% v/v), prepared as described previously. After 24 h incubation at 37 • C, the supernatant solution was removed by aspiration, adherent cells washed with PBS, and to each well was added 100 µL of 50 µM of 2 ,7 -dichlorofluorescein diacetate (DCFH-DA), after which time, plates were incubated in the dark for 2 h. After staining, cells were washed thoroughly with PBS to remove excess stain before adding ROS inducers to positive control samples (1 mM hydrogen peroxide or 0.5 mM tert-butyl hydroperoxide) for 2 h in the dark. Fluorescence readings were obtained using a microplate reader (λ Ex = 485 nm, λ Em = 530 nm). Additional control samples included negative (untreated) controls, controls only treated with metal complexes (to eliminate auto-fluorescence), and complex-treated cells with ROS inducers. Data were obtained as biological triplicates. Generation of ROS was also investigated using fluorescence microscopy. Samples were prepared as described for quantification above with the following modification: cells were seeded using 8-well microscopy chambers using 5 × 10 4 cells per well. Measurements were obtained using an EVOS FL microscope.

Evaluation of Mitochondrial Function
Briefly, 5 × 10 3 A2780 human ovarian cancer cells were seeded in 8-well microscopy chambers and incubated at 37 • C for 24 h. Cells were treated with 1 µM concentrations of complex 1a in cell culture media (DMSO not exceeding 0.5% v/v), prepared as described previously. After 24 h incubation at 37 • C, the supernatant solution was removed by aspiration, and adherent cells washed with PBS and stained using DAPI/Rhodamine 123 (Rh-123) in buffer for 1 h. This experiment included negative (untreated) controls. Measurements were obtained as biological triplicates using an EVOS FL microscope.

Thioredoxin Reductase Assay
Assay was carried out as in previously reported detailed protocols [49,50]. In short, commercially available rat liver TrxR (Sigma-Aldrich, St. Louis, MO, USA) was exposed to graded concentrations of the test compounds over 75 min. After this exposure period, the reduction of 5,5 -dithiobis (2-nitrobenzoic acid) mediated by NADPH and TrxR to 5-thio-2nitrobenzoic acid was monitored photometrically at 405 nm. The IC 50 values represent the concentrations, which were required to reduce the enzyme catalyzed turnover by 50%.

Cellular Accumulation of Ruthenium
Briefly, 1 × 10 6 cells were seeded using P100 plates and incubated at 37 • C for 24 h. Cells were treated with IC 50 concentrations of complex 1a or 2a in cell culture media (DMSO not exceeding 0.5% v/v), prepared as described previously. After 24 h incubation at 37 • C, the supernatant solution was removed by aspiration, and adherent cells were washed with PBS and harvested using trypsin. Trypsin activity was quenched using culture medium (containing FCS), cells were counted, and cell pellets were obtained by centrifugation. Cell pellets were re-washed with PBS to remove excess culture medium and re-centrifuged to afford the final cell pellet for analysis. The protein content of the cell pellets was determined by the Bradford method, and the ruthenium levels were measured using a high-resolution continuum source atomic absorption spectrometer (HRCS-AAS; contrAA 700 AnalytikJena) according to previously reported protocols [50,51].

Statistical Analysis
In all cases, independent two-sample t-tests with unequal variances, Welch's tests, were carried out to establish statistical significance of the variations (p < 0.01 for **, and p < 0.05 for *).

Syntheses and Crystal Structures of the Ligands and Complexes
Compounds that were evaluated for their anticancer properties are presented in Figure 1 (B-ligands, C-complexes). Isostructural ligands a and b ( Figure 1B) are isoquinoline and quinoline-derived pyrithione analogues, prepared in a two-step synthesis, involving N-oxidation followed by thiolation ( Figure S1A) [44,48]. With ligands in hand organoruthenium(II) chlorido complexes 1a-b ( Figure 1C) were prepared by stirring ruthenium precursor [Ru(p-cymene)Cl 2 ] 2 , selected pyrithionato-ligand a or b, and a base sodium methoxide in dichloromethane ( Figure S1B). After the purification of reaction mixture by column chromatography, complexes 1a-b were isolated upon precipitation from DCM/nheptane [44]. Reactions to afford organoruthenium(II) pta complexes 2a-b ( Figure 1C; Figure S1B) were carried out using chlorido complexes 1a or 1b, respectively, which were stirred with ammonium hexafluorophosphate (NH 4 PF 6 ) and ground pta ligand, both of which were present in excess. The role of NH 4 PF 6 was to facilitate the abstraction of the chlorido ligand from ruthenium (II) species and the subsequent binding of the phosphine ligand pta. After the reaction was completed, the excesses of NH 4 PF 6 and pta as well as the insoluble byproduct NH 4 Cl were filtered off through a pad of Celite, and the mother liquor was concentrated on a rotary evaporator. The addition of cold diethyl ether resulted in the precipitation of the yellow solid complexes, which were isolated by filtration.
Crystal structures of chlorido complexes 1a-b were already reported [44]. However, we additionally obtained suitable crystals of ligand a and pta complex 2b for single crystal X-ray diffraction ( Figure S2). Ligand a crystallized in the thione form (Figure 2), similar to all other previously reported pyrithione analogues [41]. The packing of molecules was characterized by the formation of dimers through O-H···S hydrogen bonds of the thiohydroxamate functional groups and π-stacking interactions of the isoquinoline rings with the distance between planes defined by the 10 ring atoms of 3.403 Å ( Figure S3). The crystal structure of pta complex 2b showed pseudooctahedral geometry of the ruthenium (II) coordination sphere with the bidentately bound ligand b forming a cationic structure ( Figure 2, Table S1). dium methoxide in dichloromethane ( Figure S1B). After the purification of reaction ture by column chromatography, complexes 1a-b were isolated upon precipitation DCM/n-heptane [44]. Reactions to afford organoruthenium(II) pta complexes 2a-b (Fi 1C; Figure S1B) were carried out using chlorido complexes 1a or 1b, respectively, w were stirred with ammonium hexafluorophosphate (NH4PF6) and ground pta ligand, of which were present in excess. The role of NH4PF6 was to facilitate the abstraction o chlorido ligand from ruthenium (II) species and the subsequent binding of the phosp ligand pta. After the reaction was completed, the excesses of NH4PF6 and pta as we the insoluble byproduct NH4Cl were filtered off through a pad of Celite, and the mo liquor was concentrated on a rotary evaporator. The addition of cold diethyl ether resu in the precipitation of the yellow solid complexes, which were isolated by filtration.
Crystal structures of chlorido complexes 1a-b were already reported [44]. How we additionally obtained suitable crystals of ligand a and pta complex 2b for single cr X-ray diffraction ( Figure S2). Ligand a crystallized in the thione form (Figure 2), simil all other previously reported pyrithione analogues [41]. The packing of molecules characterized by the formation of dimers through O-H···S hydrogen bonds of the thi droxamate functional groups and π-stacking interactions of the isoquinoline rings the distance between planes defined by the 10 ring atoms of 3.403 Å ( Figure S3). The tal structure of pta complex 2b showed pseudooctahedral geometry of the ruthenium coordination sphere with the bidentately bound ligand b forming a cationic structure ure 2, Table S1).

NMR Stability in Solution
NMR stability studies were carried out for chlorido 1a-b and pta 2a-b complex follow possible structural changes in simplified aqueous solutions of 5% (CD3)2SO/D2 the presence or the absence of 140 mM NaCl ( Figures S4-11). The addition of the salt m ics high concentrations of extracellular fluids of the body, while studies in water a observation of possible hydrolysis events. It is reported that Ru-Cl bond in structu similar complexes is prone to undergo substitution with the neutral water molecules form positively charged Ru-OH2 species which can further interact with biological tar [52]. Moreover, in our previous studies, we found that the behavior of various org ruthenium complexes in solutions depends on the type of the ligand and affects int tions with biomolecules and, consequently, their biological activity [41,53,54].

NMR Stability in Solution
NMR stability studies were carried out for chlorido 1a-b and pta 2a-b complexes to follow possible structural changes in simplified aqueous solutions of 5% (CD 3 ) 2 SO/D 2 O in the presence or the absence of 140 mM NaCl ( Figures S4-S11). The addition of the salt mimics high concentrations of extracellular fluids of the body, while studies in water allow observation of possible hydrolysis events. It is reported that Ru-Cl bond in structurally similar complexes is prone to undergo substitution with the neutral water molecules and form positively charged Ru-OH 2 species which can further interact with biological targets [52]. Moreover, in our previous studies, we found that the behavior of various organoruthenium complexes in solutions depends on the type of the ligand and affects interactions with biomolecules and, consequently, their biological activity [41,53,54].
Interestingly, ruthenium pyrithione chlorido complexes 1a and 1b were observed to remain stable in both water and saline solutions, with only minor release of p-cymene ligand (around 2%) from the metal observed after 24 h (Figures S4-S7), and the chlorido complex remained the major species in solution. In contrast, ruthenium pta complexes 2a and 2b appeared less stable in both water and saline ( Figures S8-S11), and after 4 days, the free p-cymene ligand dominated over those of the coordinated pta complex. Additionally, new peaks were identified (4.4-4.2 ppm) which can be likely attributed to an uncoordinated pta derivate (pta oxide, 1,3,5-triaza-7-phosphaadamantane-7-oxide) as previously described for similar pta complexes [41]. However, we cannot unambiguously confirm the presence of pta oxide, as these new peaks do not precisely coincide with a referenced spectrum of free pta oxide.

Antiproliferative Activity
We first established the anticancer activities of (iso)quinoline-derived pyrithione ligands a-b, their organoruthenium(II) chlorido 1a-b, and pta complexes 2a-b towards six human cancer cell lines-A2780 (ovarian), A549 (lung), HCT116 (colorectal), OE19 (oesophageal), HEPG2 (hepatocellular), and PC3 (prostate)-compared to the established platinum anticancer agent, cisplatin. Our results demonstrate that organoruthenium(II) chlorido complexes 1a and 1b showed moderate antiproliferative activity across a panel of human cancer cell lines (Table 1), however, they were particularly active towards ovarian cancer cells (IC 50 1.0-2.2 µM). Ligands a and b from which the complexes were derived also exhibited modest antiproliferative activities towards all tested cancer cell lines (Table 1), however, to a lesser extent than their chlorido complexes. The ruthenium p-cymene precursor [Ru(p-cymene)Cl 2 ] 2 of complexes 1a and 1b is known to be inactive towards cancer cells [22,[55][56][57][58], and our group determined its activity on A2780 and A549 cells with similar results ( Figure S15). Hence, the enhancement of ligand anticancer activity can be attributed to the formation of the new complex. In contrast, pta complexes 2a and 2b were found to be inactive towards all cell lines in the concentration range investigated. This is particularly interesting when considering the established mechanism of action (MoA) of some structurally-similar ruthenium and osmium piano stool complexes, for which hydrolysis and/or displacement of the metal-halide bond at the monodentate site is known to be crucial to in-cell activation [59,60]. It appears that the pta moiety disrupts this process, thereby significantly reducing the potency of these complexes compared to their chlorido analogues, 1a and 1b. Interestingly, substitution of the chlorido ligand for pta in previously reported methyl-substituted pyrithione complexes did not affect anticancer activities [41], however, our data suggest that the nature of the monodentate ligand is of great importance in the MoA of ruthenium pyrithione complexes bearing an extended aromatic system. Platinum therapies are widely used in the first line treatment of ovarian cancers, however, resistance is an ongoing clinical concern. Thus, anticancer activities were next determined towards three ovarian epithelial cancer cell lines, A2780, SKOV3, and SW626 (Table 2). Activity trends between all tested compounds were highly similar, with comparable activities observed towards A2780 ( Figure S16) and SW626 and slightly lower activities determined towards SKOV3 cancer cells. A2780 ovarian cancer cells are highly sensitive to cisplatin, whereas ectopic expression of the CA125 tumor antigen C-terminal domain in SKOV3 cells decreases their sensitivity to cisplatin [61]. Interestingly, SW626 cancer cells are also thought to be primary ovarian cells, though studies suggested that they might be derived from an ovarian metastasis of a colorectal adenocarcinoma [62]. All ligands and chlorido complexes appeared slightly less active towards SKOV3 cells compared to platinum-sensitive A2780 cells. The observed variations in activity could include a contribution from intrinsic cellular differences between A2780 and SKOV3. These cells differ in proliferation rates, receptor expression, and morphology. Therefore, we also determined antiproliferative activities of the most promising complexes, 1a and 1b, towards a platinum-resistant variant of A2780 (A2780Cis) to provide direct comparison between platinum-sensitive and platinum-resistant cells ( Table 2). Strikingly, the activities of 1a and 1b were statistically unchanged between these two cell lines, suggesting that these complexes show promise in the fight against platinum resistance.
In addition to platinum resistance in the clinic, some cancers may acquire resistance to a range of chemotherapeutics, becoming so-called "multi-drug resistant" (MDR) cancers. Thus, the antiproliferative activities of 1a and 1b were also determined towards A2780ADR, an Adriamycin-resistant variant of the parent A2780 cell line (Table 2). Both complexes achieved comparably low micromolar IC 50 concentrations, between 1.0-1.6 µM for the complex 1a and 2.2-2.8 µM for the complex 1b. Given the likely importance of the Ru−Cl bond in the mechanism of action and how orientation of either isomer does not inflict steric interference at the monodentate site, it is perhaps not surprising that, overall, 1a and its structural isomer 1b exhibited comparable antiproliferative activities across all human cell lines investigated.

Cellular Accumulation of Ruthenium
Metal complexes may be transported across cell membranes by passive diffusion, transport proteins, or endocytosis [63], where subsequent in-cell activation and anticancer activity may occur [59,64]. In addition, efflux mechanisms should also be considered. In combination, both influx and efflux pathways can contribute to drug resistance [63]. For example, cisplatin-resistant A2780Cis cells exhibit downregulation of a copper uptake transporter protein (CTR1) and upregulation of efflux transporters multi-drug resistance protein 2 (MRP-2), ATPase copper transporting α (ATP7A), and ATPase copper transporting β [65].
Cellular accumulations of ruthenium in cells treated with either active complex 1a or 2a were determined after 24 h of incubation with equimolar concentrations (1.0 µM) of metal complex. Ruthenium was found to be internalized by A2780 cancer cells treated with either chlorido 1a or its analogue pta complex 2a, with average accumulations of 1.30 nmol (131.4 ng) Ru/mg protein and 1.02 nmol (103.1 ng) Ru/mg protein, respectively (Table 3). Since A2780 cells were incubated with equimolar concentrations of 1a and 2a, it is perhaps unsurprising that metal accumulations were found to be highly comparable, suggesting that these structurally similar complexes may share a common mechanism of cellular influx. However, what is striking is how similar levels of internalized complex differed in their anticancer activity between (IC 50 = 1.0 µM and > 50 µM for 1a and 2a, respectively). Thus, while the nature of the monodentate site does not appear to influence accumulation, these data provide further evidence for its importance in the overall mechanism of action. It was previously demonstrated for structurally similar organoruthenium(II) chlorido complexes that cellular uptake does not corelate with the cytotoxicity of the compounds towards A2780 cells [66]. Interestingly, other organoruthenium(II) pta complexes were reported to be inactive on TS/A adenocarcinoma cancer cells despite their cellular uptake [67], as observed with complex 2a.

Interactions in Biological Media
NMR stability studies suggested that ruthenium pta complexes 2a-b may undergo greater structural changes in solution (Figures S8-S11) compared to their chlorido analogues 1a-b (Figures S4-S7). To gain further insight into the stability of the compounds in a more complex and biologically relevant environment, the compounds were further studied by UV-visible spectroscopy using (i) phosphate buffer solution PBS, (ii) cell culture medium RPMI-1640, (iii) cell culture medium RPMI-1640 supplemented with 10% (v/v) fetal calf serum and 1% (v/v) penicillin/streptomycin antibiotics, and (iv) human blood plasma by UV-Vis spectrometry. As similarly observed in preliminary NMR stability studies, chlorido complexes 1a and 1b ( Figure 3A and Figure S12A, respectively) remained stable in all four media. While pta complexes 2a and 2b retained modest stability in phosphate-buffered saline, as observed in NMR studies, spectral changes were observed in all other media ( Figure 3B and Figure S12B, respectively). The observed UV-Vis spectral changes of the pta complexes could relate to a commonality between RPMI-1640 culture medium (in the fetal calf serum supplemented and non-supplemented samples) and human blood plasma. Hence, addition of fetal calf serum to the supplemented media sample was not essential to causing this change. Both RPMI-1640 and human plasma contain significant levels of nucleophilic components (e.g., 1 mM reduced glutathione in RPMI-1640, with similar levels measured in human plasma [68]) which may interfere with coordinated pta. More generally, UV-Vis studies together with NMR stability experiments point to more possible structural changes of investigated pta than chlorido complexes that might contribute to the lack of observed in vitro activity of 2a and 2b in the presence of cell culture medium.

Explorations of the Mechanism of Action
We previously reported similar pyrithione ligands which became effective inhibitors of thioredoxin reductase (TrxR) when associated with the architecture of a metal complex [41]. This enzyme (alongside thioredoxin and NADPH) makes up the thioredoxin system, which is involved in processes such as control of the intracellular redox environment, cellular growth, and the regulation of apoptosis [69]. As such, the thioredoxin system is an interesting target in the development of anticancer drugs [70]. In fact, overexpressed thioredoxin in some cancer cells was implicated as a contributor to platinum resistance due to its involvement in the regulation of apoptosis [71]. Similarly, inhibition of thioredoxin reductase by the gold(I) drug auranofin was proposed to enhance cell death in cisplatinresistant human ovarian cancer cells [72].
As complex 1a retained activity towards both platinum sensitive (A2780) and resistant (A2780Cis) cancer cells, interactions between complex 1a and thioredoxin reductase were investigated. During the preliminary assay, a concentration of 10 µM complex 1a caused 94% inhibition of TrxR, with an IC50 of 4.1 ± 0.16 µM determined in dose-response experiments. We also observed that ligand a (from which complex 1a was derived) did not inhibit TrxR. Considering that the determined inhibitory concentration of TrxR (4.1 µM) was also comparable with the determined anticancer growth inhibition concentration (1.0 µM), TrxR is a possible subcellular target of 1a and may contribute to the in vitro mechanism of action. Involvement of TrxR in the MoA of a multi-targeting metal complex presents a possible explanation for the ability of 1a to overcome platinum resistance in A2780 cancer cells.
Since complex 1a appears to inhibit thioredoxin reductase activity in model experiments, and TrxR is linked to cellular redox regulation, we explored evidence for this observation in vitro by fluorescence microscopy. While low to moderate concentrations of reactive oxygen species (ROS) are reported to be important for normal functioning of some physiological processes, elevated ROS species can damage proteins, nucleic acids, lipids, and membranes, which may activate either non-physiological (necrotic) or regulated (apoptotic) cell death pathways [73,74]. Many metal complexes are known to cause enhanced ROS formation [75][76][77][78]. We therefore evaluated the induction of reactive oxygen species (ROS) in A2780 cells by complex 1a. However, our results showed that the compound 1a only induced a small increase in intracellular ROS levels (fluorescence intensity ca. 700 fluorescence units) compared to the untreated control (ca. 650 fluorescence units; Figure 4), with measured ROS levels in treated cells only slightly above basal signaling levels in untreated cells. In contrast, positive controls hydrogen peroxide and Luperox

Explorations of the Mechanism of Action
We previously reported similar pyrithione ligands which became effective inhibitors of thioredoxin reductase (TrxR) when associated with the architecture of a metal complex [41]. This enzyme (alongside thioredoxin and NADPH) makes up the thioredoxin system, which is involved in processes such as control of the intracellular redox environment, cellular growth, and the regulation of apoptosis [69]. As such, the thioredoxin system is an interesting target in the development of anticancer drugs [70]. In fact, overexpressed thioredoxin in some cancer cells was implicated as a contributor to platinum resistance due to its involvement in the regulation of apoptosis [71]. Similarly, inhibition of thioredoxin reductase by the gold(I) drug auranofin was proposed to enhance cell death in cisplatinresistant human ovarian cancer cells [72].
As complex 1a retained activity towards both platinum sensitive (A2780) and resistant (A2780Cis) cancer cells, interactions between complex 1a and thioredoxin reductase were investigated. During the preliminary assay, a concentration of 10 µM complex 1a caused 94% inhibition of TrxR, with an IC 50 of 4.1 ± 0.16 µM determined in dose-response experiments. We also observed that ligand a (from which complex 1a was derived) did not inhibit TrxR. Considering that the determined inhibitory concentration of TrxR (4.1 µM) was also comparable with the determined anticancer growth inhibition concentration (1.0 µM), TrxR is a possible subcellular target of 1a and may contribute to the in vitro mechanism of action. Involvement of TrxR in the MoA of a multi-targeting metal complex presents a possible explanation for the ability of 1a to overcome platinum resistance in A2780 cancer cells.
Since complex 1a appears to inhibit thioredoxin reductase activity in model experiments, and TrxR is linked to cellular redox regulation, we explored evidence for this observation in vitro by fluorescence microscopy. While low to moderate concentrations of reactive oxygen species (ROS) are reported to be important for normal functioning of some physiological processes, elevated ROS species can damage proteins, nucleic acids, lipids, and membranes, which may activate either non-physiological (necrotic) or regulated (apoptotic) cell death pathways [73,74]. Many metal complexes are known to cause enhanced ROS formation [75][76][77][78]. We therefore evaluated the induction of reactive oxygen species (ROS) in A2780 cells by complex 1a. However, our results showed that the compound 1a only induced a small increase in intracellular ROS levels (fluorescence intensity ca. 700 fluorescence units) compared to the untreated control (ca. 650 fluorescence units; Figure 4), with measured ROS levels in treated cells only slightly above basal signaling levels in untreated cells. In contrast, positive controls hydrogen peroxide and Luperox  Reactive oxygen species are the byproduct of oxidative phosphorylation, and enhanced ROS concentrations are often correlated with mitochondrial dysfunction [79]. Since levels of ROS were relatively unchanged by treatment of A2780 cells with 1a, mitochondrial function in treated cells was also evaluated to provide further evidence that ROS are unlikely to play a major role in the in-cell mechanism of complex 1a ( Figure 5). In both untreated controls and treated cells, blue fluorescence (DAPI) suggested that cells did not exhibit significant nuclear morphology changes, consistent with previous experiments. Similarly, green fluorescence (Rhodamine-123) showed that mitochondrial function was unaffected by treatment with 1a compared to the untreated control ( Figure S17) under the conditions investigated. Further evidence of the lack of redox involvement of complex 1a was provided by cyclic voltammetry studies, including ligand a, and analogous pta complex 2a. While all three compounds are redox active between -2.0 V to +2.0 V (Figure 6A), only uncoordi- Reactive oxygen species are the byproduct of oxidative phosphorylation, and enhanced ROS concentrations are often correlated with mitochondrial dysfunction [79]. Since levels of ROS were relatively unchanged by treatment of A2780 cells with 1a, mitochondrial function in treated cells was also evaluated to provide further evidence that ROS are unlikely to play a major role in the in-cell mechanism of complex 1a ( Figure 5). In both untreated controls and treated cells, blue fluorescence (DAPI) suggested that cells did not exhibit significant nuclear morphology changes, consistent with previous experiments. Similarly, green fluorescence (Rhodamine-123) showed that mitochondrial function was unaffected by treatment with 1a compared to the untreated control ( Figure S17) under the conditions investigated. Reactive oxygen species are the byproduct of oxidative phosphorylation, and enhanced ROS concentrations are often correlated with mitochondrial dysfunction [79]. Since levels of ROS were relatively unchanged by treatment of A2780 cells with 1a, mitochondrial function in treated cells was also evaluated to provide further evidence that ROS are unlikely to play a major role in the in-cell mechanism of complex 1a ( Figure 5). In both untreated controls and treated cells, blue fluorescence (DAPI) suggested that cells did not exhibit significant nuclear morphology changes, consistent with previous experiments. Similarly, green fluorescence (Rhodamine-123) showed that mitochondrial function was unaffected by treatment with 1a compared to the untreated control ( Figure S17) under the conditions investigated. Further evidence of the lack of redox involvement of complex 1a was provided by cyclic voltammetry studies, including ligand a, and analogous pta complex 2a. While all three compounds are redox active between -2.0 V to +2.0 V ( Figure 6A), only uncoordinated ligand a exhibited redox activity between the biologically-relevant range of -0.8 V Further evidence of the lack of redox involvement of complex 1a was provided by cyclic voltammetry studies, including ligand a, and analogous pta complex 2a. While all three compounds are redox active between -2.0 V to +2.0 V ( Figure 6A), only uncoordinated ligand a exhibited redox activity between the biologically-relevant range of -0.8 V and +0.8 V ( Figure 6B). Thus, while ruthenium pyrithione complexes 1a and 2a are unlikely to be involved in cellular redox processes, it is possible that cellular degradation pathways may mediate release of free ligand a, which may account for the observations of slightly elevated ROS levels compared to the untreated control ( Figure 4). Such degradation pathways were previously identified for structurally similar osmium-arene complexes utilizing combination studies of ICP-MS and X-ray fluorescence microscopy, involving endosomal degradation and glutathione [80].
Cancers 2021, 13, x 14 of 21 and +0.8 V ( Figure 6B). Thus, while ruthenium pyrithione complexes 1a and 2a are unlikely to be involved in cellular redox processes, it is possible that cellular degradation pathways may mediate release of free ligand a, which may account for the observations of slightly elevated ROS levels compared to the untreated control ( Figure 4). Such degradation pathways were previously identified for structurally similar osmium-arene complexes utilizing combination studies of ICP-MS and X-ray fluorescence microscopy, involving endosomal degradation and glutathione [80]. Upon excluding ROS and mitochondrial targeting from the likely major mechanism of complex 1a in cells, a cell cycle analysis of A2780 cancer cells treated with compound 1a was determined (Table 4). In platinum-sensitive cells, platinum-based therapies exhibit S + G2/M-phase cell cycle arrest as a result of binding DNA, thereby preventing DNA synthesis. This results in the activation of p53/p21 pathways, leading to cellular apoptosis [81]. However, in platinum-resistant cells, DNA repair mechanisms such as the nucleotide excision repair systems (NER) can repair platinated DNA adducts, preventing S+G2/M cell cycle arrest and promoting platinum resistance [82]. In order to overcome platinum resistance, multi-targeted compounds must identify unique subcellular targets that are not common to platinum drugs. After incubation of A2780 cancer cells with complex 1a, the cell cycle G1 population was significantly increased relative to the untreated control (72.5% compared to 62.5%, respectively). As such, it appears that the mechanism of action of 1a is not common to platinum-based compounds. It is well established that metal complexes often exhibit multi-targeting mechanisms [24], and we suggest that the compound 1a is likely to interact with various sub-cellular components other than DNA. On detecting significant G1 arrest caused by complex 1a, the induction of apoptosis (programmed cell death) in A2780 cancer cells was investigated to determine whether the Upon excluding ROS and mitochondrial targeting from the likely major mechanism of complex 1a in cells, a cell cycle analysis of A2780 cancer cells treated with compound 1a was determined (Table 4). In platinum-sensitive cells, platinum-based therapies exhibit S + G 2 /M-phase cell cycle arrest as a result of binding DNA, thereby preventing DNA synthesis. This results in the activation of p53/p21 pathways, leading to cellular apoptosis [81]. However, in platinum-resistant cells, DNA repair mechanisms such as the nucleotide excision repair systems (NER) can repair platinated DNA adducts, preventing S+G 2 /M cell cycle arrest and promoting platinum resistance [82]. In order to overcome platinum resistance, multi-targeted compounds must identify unique subcellular targets that are not common to platinum drugs. After incubation of A2780 cancer cells with complex 1a, the cell cycle G 1 population was significantly increased relative to the untreated control (72.5% compared to 62.5%, respectively). As such, it appears that the mechanism of action of 1a is not common to platinum-based compounds. It is well established that metal complexes often exhibit multi-targeting mechanisms [24], and we suggest that the compound 1a is likely to interact with various sub-cellular components other than DNA. On detecting significant G 1 arrest caused by complex 1a, the induction of apoptosis (programmed cell death) in A2780 cancer cells was investigated to determine whether the mechanism of 1a was cytotoxic or cytostatic (Table 5). Cells were stained with propidium iodide (PI) and Annexin-V (FITC conjugate; ANN) to identify loss of membrane integrity (PI able to enter cells and bind to DNA) and the translocation of phosphatidylserine to the outer membrane, a trait commonly associated with late-stage apoptosis. A2780 cells treated with complex 1a only showed partial evidence for apoptotic cell death (8.4% of cell population), with a greater cell population determined as non-viable cells (11.2% of cell population), indicative of a major non-apoptotic pathway in the mechanism of action. Apoptotic cell death pathway is frequently identified with platinum agents, and its activation is mediated by a variety of mechanisms. For example, while platinumassociated apoptosis is typically dependent on the activation of tumor suppressor protein p53, ruthenium complexes were reported to induce ERK1/2-mediated apoptosis via a p53-indpendent pathway [83]. While it appears that this new class of ruthenium complexes does cause slight apoptosis, this level is unlikely to contribute to the main mechanism of action of such complexes towards A2780 cancer cells. Owing to the large population of non-viable cells after treatment with 1a, the membrane integrity of exposed cells was evaluated. Cells exhibiting compromised membranes in the absence of apoptotic cell death may indicate the involvement of a necrotic cell death pathway. Interestingly, while cells did not exhibit significant nuclear morphology changes (viable blue DAPI florescence), complex 1a appeared to disrupt the cell membrane, allowing propidium iodide (PI) to enter cells, causing red florescence upon binding cellular DNA ( Figure 7). Further, this finding is consistent with the observation of a greater population of cells in the PI+ quadrant during the investigation of apoptosis ( Table 5).
The influence of complex 1a on the migration of A2780 cancer cells was next investigated as part of a wound-healing assay (Figure 8). A wound was created in a monolayer of A2780 cells using the pipette tip (1282 ± 59 µm). Cells treated with complex 1a were observed and compared to the untreated (negative) control to determine whether the presence of the ruthenium complex would influence wound healing or cell motility. After 24 h exposure to 1a, the wound healing was significantly hindered (p < 0.05) compared to the untreated cell population, exhibiting a wound width of 1035 ± 89 µm (19% reduction) compared to 319 ± 35 µm (75% reduction), respectively. Cells treated with 1a showed a reduction in motility towards wound healing, which is consistent with previously observed G 1 arrest. Since such cell cycle arrest extends the growth phase and delays DNA replication and subsequent cell division, these findings suggest that the mechanism of action appears to be cytostatic, as opposed to cytotoxic. In addition, cell colonies treated with complex 1a for 24 h were found to be significantly smaller and fewer in number (24 ± 5 colonies) compared to untreated colonies (45 ± 3 colonies) after 7 days growth (p < 0.01). This result is again consistent with the observation of G 1 cell cycle arrest and a cytostatic (rather than cytotoxic) mechanism of action. pathway. Interestingly, while cells did not exhibit significant nuclear morphology changes (viable blue DAPI florescence), complex 1a appeared to disrupt the cell membrane, allowing propidium iodide (PI) to enter cells, causing red florescence upon binding cellular DNA (Figure 7). Further, this finding is consistent with the observation of a greater population of cells in the PI+ quadrant during the investigation of apoptosis (Table  5). The influence of complex 1a on the migration of A2780 cancer cells was next investigated as part of a wound-healing assay (Figure 8). A wound was created in a monolayer Cancers 2021, 13, x of A2780 cells using the pipette tip (1282 ± 59 µm). Cells treated with complex observed and compared to the untreated (negative) control to determine whether ence of the ruthenium complex would influence wound healing or cell motility h exposure to 1a, the wound healing was significantly hindered (p < 0.05) compa untreated cell population, exhibiting a wound width of 1035 ± 89 µm (19% r compared to 319 ± 35 µm (75% reduction), respectively. Cells treated with 1a s reduction in motility towards wound healing, which is consistent with previ served G1 arrest. Since such cell cycle arrest extends the growth phase and del replication and subsequent cell division, these findings suggest that the mech action appears to be cytostatic, as opposed to cytotoxic. In addition, cell colonie with complex 1a for 24 h were found to be significantly smaller and fewer in nu ± 5 colonies) compared to untreated colonies (45 ± 3 colonies) after 7 days gro 0.01). This result is again consistent with the observation of G1 cell cycle arrest an static (rather than cytotoxic) mechanism of action.

Conclusions
A new series of organoruthenium(II) pyrithione complexes were describe show promising potency towards ovarian cancer cells. The nature of the monode was confirmed to be crucial in the activation and/or the mechanism of action of th plexes, as substitution of the chlorido moiety for pta rendered the complexes in

Conclusions
A new series of organoruthenium(II) pyrithione complexes were described, which show promising potency towards ovarian cancer cells. The nature of the monodentate site was confirmed to be crucial in the activation and/or the mechanism of action of these complexes, as substitution of the chlorido moiety for pta rendered the complexes inert under the test conditions of our experiments. This work suggests that the mode of action of 1a is cytostatic in nature, involving G 1 cell cycle arrest, the absence of ROS generation of mitochondrial dysfunction, reduced cell motility, reduced cell colony size, and limited induction of apoptosis. Similarly, cells treated with 1a were found to exhibit damaged cell membranes, in agreement with the increased non-viable cell population determined by flow cytometry as part of apoptosis experiments. Importantly, the mechanism of action appears to differ from the well-established DNA binding mechanism of action of platinum anti-cancer agents, and our findings offer an explanation as to how complex 1a appears to overcome platinum resistance in ovarian cancer cells. In addition, complex 1a also inhibits TrxR, enzyme involved in cancer development. Our findings highlight how a multi-targeting mechanism of action offered by such metal complexes can contribute to future development of cancer therapies to combat drug resistance.