Coumarin-Based Triapine Derivatives and Their Copper(II) Complexes: Synthesis, Cytotoxicity and mR2 RNR Inhibition Activity

A series of thiosemicarbazone-coumarin hybrids (HL1-HL3 and H2L4) has been synthesised in 12 steps and used for the preparation of mono- and dinuclear copper(II) complexes, namely Cu(HL1)Cl2 (1), Cu(HL2)Cl2 (2), Cu(HL3)Cl2 (3) and Cu2(H2L4)Cl4 (4), isolated in hydrated or solvated forms. Both the organic hybrids and their copper(II) and dicopper(II) complexes were comprehensively characterised by analytical and spectroscopic techniques, i.e., elemental analysis, ESI mass spectrometry, 1D and 2D NMR, IR and UV–vis spectroscopies, cyclic voltammetry (CV) and spectroelectrochemistry (SEC). Re-crystallisation of 1 from methanol afforded single crystals of copper(II) complex with monoanionic ligand Cu(L1)Cl, which could be studied by single crystal X-ray diffraction (SC-XRD). The prepared copper(II) complexes and their metal-free ligands revealed antiproliferative activity against highly resistant cancer cell lines, including triple negative breast cancer cells MDA-MB-231, sensitive COLO-205 and multidrug resistant COLO-320 colorectal adenocarcinoma cell lines, as well as in healthy human lung fibroblasts MRC-5 and compared to those for triapine and doxorubicin. In addition, their ability to reduce the tyrosyl radical in mouse R2 protein of ribonucleotide reductase has been ascertained by EPR spectroscopy and the results were compared with those for triapine.

A series of potentially bidentate (NS), tridentate (ONS or CNS, in particular, for Pd(II) and Ru(II)) Schiff bases is well documented in the literature [40][41][42][43][44][45][46][47][48][49][50]. These have been obtained by condensation reactions of acetyl-, formyl-or trifluoroacetyl-coumarins with substituted thiosemicarbazides. In the case of tridentate ONS Schiff bases, the starting coumarins contain a second carbonyl or hydroxyl function in a position suitable for chelation to a transition metal. Here, we propose a new synthetic pathway to coumarinthiosemicarbazone hybrids with NNS binding site, as is the case for triapine, which is considered the most favourable for the design of potent metal-based anticancer drugs [11]. A series of novel antitumor TSC-coumarin hybrids and their Cu(II) complexes were prepared, by attachment of 3-AP-related moiety to a coumarin fragment at position 3 via a piperazine spacer. Copper(II) complexes with potentially tetradentate piperazine ligands bearing pendant pyridyl groups were reported to effectively cleave DNA and to be cytotoxic [51]. An additional 3-AP-based moiety was attached at position 7 (for both type of structures see Chart 2). Chart 2. TSC-coumarin hybrids (HL 1 -HL 3 and H 2 L 4 ) and their copper(II) complexes studied in this work, 1 was studied by X-ray crystallography.
It is well known that these two frameworks separately exhibit anticancer activity, therefore the aim was to determine if they can work in synergy. It was hypothesised that the novel hybrid molecules could display multiple biological activities with an improved selectivity profile, and reduced side effects. The novel compounds were characterised by spectroscopic and spectroelectrochemical techniques, and their antiproliferative activity was screened by a colorimetric MTT assay in various cancer cell lines (human breast adenocarcinoma (MDA-MB-231), human colorectal adenocarcinoma cell lines (COLO-205 and COLO-320)) and human healthy lung fibroblasts . The inhibition of mouse R2 RNR protein, a likely target for some TSCs, by the metal-free ligand HL 1 and copper(II) complexes 1 and 3 (see Chart 2) was also investigated, and compared with that of triapine.

Chart 3.
The structure of Zand E-isomer of HL 1 .

Physical Measurements
Elemental analyses for 1-4 were carried out in a Carlo Erba microanalyser at the Microanalytical Laboratory of the University of Vienna. TFA content was measured by using free zone capillary electrophoresis (Agilent Scientific Instruments, Santa Clara, CA 95051,United States). Separation of ions was achieved at −30 kV along a 45 cm silica capillary with 50 µm bore width. The electrolyte was a Good's buffer solution prepared from 50 mM cyclohexylaminosulfonic acid and 20 mM arginine (pH = 9.1). Trimethyl(tetradecyl)ammonium hydroxide was added as EOF modifier also influencing the separation of TFA from acetate. Electrospray ionisation mass spectrometry (ESI-MS) was carried out with an amaZon speed ETD Bruker (Bruker Daltonik GmbH, Bremen, Germany, m/z range 0-900, ion positive/negative mode, 180 • C, heating gas N2 (5 L/min), Capillary 4500 V, End Plate Offset 500 V) instrument for all compounds. Expected and experimental isotopic distributions were compared. UV-vis spectra of HL 1 -HL 3

Crystallographic Structure Determination
X-ray diffraction measurement of Cu(L 1 )Cl·1.775H 2 O (1 ·1.775H 2 O) was performed on a Bruker D8 (Karlsruhe, Germany) Venture diffractometer. A single crystal was positioned at 45 mm from the detector, and 1591 frames were measured, each for 60 s over 0.7 • scan width. Crystal data, data collection parameters, and structure refinement details are given in Table 1. The structure was solved by direct methods and refined by fullmatrix least-squares techniques. Non-H atoms were refined with anisotropic displacement parameters except those belonging to disordered fragments. H atoms were inserted in calculated positions and refined with a riding model. The following computer programs and hardware were used: structure solution, SHELXS-2014 and refinement, SHELXL-2014 [53] molecular diagrams, ORTEP [54] computer, Intel CoreDuo. CCDC 2047825.

Electrochemistry and Spectroelectrochemistry
Cyclic voltammetry experiments with 0.5 mM solutions of Cu(II) complexes and ligands in 0.1 M nBu 4 NPF 6 DMSO were performed under argon atmosphere using a threeelectrode arrangement with glassy carbon 1 mm disc working electrode (from Ionode, Australia), platinum wire as counter electrode, and silver wire as pseudo-reference electrode. Ferrocene (Fc) served as the internal potential standard. A Heka PG310USB (Lambrecht, Germany) potentiostat with a PotMaster 2.73 software package served as the potential control in voltammetry studies. In situ ultraviolet-visible-near-infrared (UV-vis-NIR) spectroelectrochemical measurements were performed on a spectrometer Avantes (Apeldoorn, The Netherlands) (Model AvaSpec-2048x14-USB2) in the spectroelectrochemical cell kit (AKSTCKIT3) with the Pt-microstructured honeycomb working electrode (1.7 mm optical path length), purchased from Pine Research Instrumentation (Lyon, France). The cell was positioned in the CUV-UV Cuvette Holder (Ocean Optics, Ostfildern, Germany) connected to the diode-array UV-vis-NIR spectrometer by optical fibres. UV-vis-NIR spectra were processed using the AvaSoft 7.7 software package. Halogen and deuterium lamps were used as light sources (Avantes, Model AvaLight-DH-S-BAL). The in situ EPR spectroelectrochemical experiments were carried out under an argon atmosphere in the EPR flat cell (0.5 mm thickness of the inner space) equipped with a large platinum mesh working electrode.
The freshly prepared solutions were carefully purged with argon and the electrolytic cell was polarised in the galvanostatic mode directly in the cylindrical EPR cavity TM-110 (ER 4103 TM) and the electron paramagnetic resonance (EPR) spectra were measured in situ. The EPR spectra were recorded at room temperature or at 77 K with the EMX Bruker spectrometer (Rheinstetten, Germany).
where n is the number of reflections and p is the total number of parameters refined.

Cell Lines and Culture Conditions
Human breast adenocarcinoma cells MDA-MB-231, sensitive COLO-205 and multidrug resistant COLO-320 colorectal adenocarcinoma cells and normal human lung fibroblasts MRC-5 were obtained from ATCC. COLO-205 and COLO-320 cells were cultured in RPMI-1640 medium containing 10% foetal bovine serum, MRC-5 cells were cultured in EMEM medium containing 10% foetal bovine serum. MDA-MB-231 cells were cultured in DMEM/high glucose medium containing 10% FBS. Adherent MDA-MB-231 cells were grown in Falcon tissue culture 75 cm 2 flasks and all other cells were grown in tissue culture 25 cm 2 flasks (BD Biosciences, Singapore). All cell lines were grown at 37 • C in a humidified atmosphere of 95% air and 5% CO 2 . The stock solutions of copper(II) complexes were prepared in DMSO.

Inhibition of Cell Viability Assay
The cytotoxicity of the compounds was determined by colorimetric MTT assay. The cells were harvested from culture flasks by trypsinisation and seeded into Cellstar 96-well microculture plates at the seeding density of 6000 cells per well (6 × 10 4 cells/mL, MDA-MB-231) or 10,000 cells per well (1× 10 5 cells/mL cells/mL, COLO-205, COLO-320 and MRC-5). The cells were allowed to resume exponential growth for 24 h, and subsequently were exposed to drugs at different concentrations in media for 72 h. The drugs were diluted in complete medium at the desired concentration and added to each well (100 µL) and serially diluted to other wells. After exposure for 72 h, the media were replaced with MTT in media (5 mg/mL, 100 µL) and incubated for additional 50 min. Subsequently, the media were aspirated and the purple formazan crystals formed in viable cells were dissolved in DMSO (100 µL). Optical densities were measured at 570 nm using the BioTek H1 Synergy (BioTek, Singapore) microplate reader. The quantity of viable cells was expressed in terms of treated/control (T/C) values in comparison to untreated control cells, and 50% inhibitory concentrations (IC 50 ) were calculated from concentration-effect curves by interpolation.
Evaluation was based on at least three independent experiments, each comprising six replicates per concentration level.

Tyrosyl Radical Reduction in Mouse R2 RNR Protein
The 9.4 GHz EPR spectra were recorded at 30 K on a Bruker Elexsys II E540 EPR spectrometer with an Oxford Instruments ER 4112HV helium cryostat, essentially as described previously [13]. The concentration of the tyrosyl radical in mouse R2 ribonucleotide reductase protein (mR2) was determined by double integration of EPR spectra recorded at non-saturating microwave power levels (3.2 mW) and compared with the copper standard [55] mR2 protein was expressed, purified, and iron-reconstituted as described previously [56] and passed through a 5 mL HiTrap desalting column (GE Healthcare) to remove excess iron. The purified, iron-reconstituted mR2 protein resulted in the formation of 0.76 tyrosyl radical/polypeptide. Samples containing 20 µM mR2 in 50 mM Hepes buffer, pH 7.50/100 mM NaCl, and 20 µM HL 1 , complex 1, or complex 3 in 1% (v/v) DMSO/H 2 O, and 2 mM dithiothreitol (DTT) were incubated for indicated times and quickly frozen in cold isopentane. The same samples were used for repeated incubations at room temperature. The experiments were performed in duplicates.

Redox Activity of [Fe II (L 1 ) 2 ]
The generation of paramagnetic intermediates was monitored by cw-EPR spectroscopy using the EMX spectrometer (Bruker). Deionised water was used for the preparation of ethanol/water solutions in which HL 1 was dissolved (ethanol was used to increase the solubility of HL 1 ). The spin trapping agent (DMPO; Sigma-Aldrich) was distilled prior to use.

Synthesis and Characterisation of HL 1 -HL 3 and H 2 L 4 , Their Copper(II) Complexes 1-4 and 1
The synthesis of ligands HL 1 -HL 3 and H 2 L 4 was realised in 12 steps. Two building blocks were prepared first, namely 4-chloromethyl-2-dimethoxymethylpyridine (E) starting from 2,4-pyridinedicarboxylic acid in five steps, as shown in Scheme S1, and 7-hydroxy-3-(piperazine-1-carbonyl)-2H-chromen-2-one (H) starting from 2,4-dihydroxybenzaldehyde in four steps, as shown in Scheme S2 and discussed in detail in ESI †. The atom labelling of the precursors for NMR resonances assignment is given in Schemes S3 and S4. Then these building blocks E and H were combined to give aldehydes or aldehyde precursors, which entered condensation reactions with thiosemicarbazide derivatives with formation of HL 1 -HL 3 and H 2 L 4 in 3 steps as shown in Scheme 1.

Synthesis of 3-(4-((2-(Dimethoxymethyl)pyridin-4-yl)methyl)piperazine-1-carbonyl)-7-hydroxy-2H-chromen-2-one (I 1 ) and 7-((2-(Dimethoxymethyl)pyridin-4-yl)methoxy)-3-(4-((2-(dimethoxymethyl)pyridin-4-yl)methyl)piperazine-1-carbonyl)-2H-chromen-2-one (I 2 )
The condensation of two building blocks H and E (Scheme 1) was first attempted in the presence of Et 3 N (pK a = 10.75) to yield the desired product I 1 in low yield (12-17%). By using a much stronger base, namely 1,1,3,3-tetramethylguanidine (TMG, pK a = 13), and optimising the molar ratio of reactants the I 1 was produced in good to very good yield (50-80%) (Table S1, ESI †). We also noticed that the synthesis of I 1 was accompanied by the formation of another product of condensation, i.e., I 2 , with two pyridine rings and one coumarin-piperazine moiety even if a double excess of H in comparison to E was used. Interestingly, the same reaction performed in the presence of K 2 CO 3 as a base resulted only in I 2 . The two products I 1 and I 2 were separated by column chromatography with MeOH/EtOAc 2:1 as eluent (for details see Experimental part). The identity of I 1 and I 2 was confirmed by positive ESI mass spectra, which showed peaks at m/z 440. 18 13 C NMR spectra were also in agreement with the structures proposed. Assignment of proton resonances was carried out based on 2D NMR experiments. The presence of the second pyridine ring-containing fragment instead of the hydroxyl group at the position 7 of the coumarin ring in I 2 was evidenced by an additional group of signals due to 2dimethoxymethylpyridine moiety (Atom 24 -Atom 31 , Scheme S5, ESI †), as well as by the downfield shifts of the nearest to the condensation place protons H 6 and H 8 at 6.81, 6.73 ppm in I 1 and at 7.10 and 7.13 ppm in I 2 . The rest of protons (Atom 4 -Atom 23 ) of I 2 revealed the same resonances in terms of chemical shifts as the corresponding protons in I 1 . Even the methylene protons H 16 in I 1 and I 2 resonate at almost the same magnetic field, namely at 3.57 and 3.58 ppm, respectively. The methylene protons H 24 in I 2 are seen at 5.36 ppm and can be used as diagnostic signature for identification of I 2 . The corresponding methylene carbon atoms in I 2 also showed noticeable differences in resonances with C 16 at 60.43 ppm and C 24 at 68.22 ppm. The downfield shift of C 24 resonance is in accordance with the vicinity of a more electronegative oxygen atom next to Atom 24 in comparison to nitrogen atom next to Atom 16 (Scheme S5, ESI †).

Synthesis of HL 1 -HL 3 and H 2 L 4
The synthesis of TSCs HL 1 -HL 3 and H 2 L 4 was performed by condensation reactions of the aldehyde J 1 with three different thiosemicarbazides in the 1:1 molar ratio in boiling ethanol, while H 2 L 4 starting from J 2 and 4N-dimethylthiosemicarbazide in the 1:2 molar ratio in boiling ethanol, in good yields (65-84%) (Scheme 1 and Scheme S8, ESI †). Other attempts described in ESI † and related to Scheme S8 were less successful. The obtained TSCs HL 1 -HL 3 and H 2 L 4 were characterised by elemental analysis, multinuclear NMR, UV-vis, IR spectroscopy and ESI mass spectrometry. Mass spectra measured in positive ion mode gave peaks attributed to [M + H] + and [M + Na] + ions. One-and two-dimensional 1 H and 13 C NMR spectra confirmed the expected structures for HL 1 -HL 3 and H 2 L 4 and the presence of Eand/or Z-isomers in DMSO-d 6 . Whereas in the case of TSCs with the bulky N-substituents (R 2 = Ph in HL 2 or R 2 = 4-hydroxy-3,5-dimethylphenyl in HL 3 ) only E-isomer was observed, for HL 1 (R 1 = R 2 = Me) both isomers in the molar ratio of Z/E 1:3.4 are present. These two isomers, shown in Chart 3 for HL 1 , are formed due to hydrogen bond formation between the NH group as proton donor (H 23 in Scheme S9, ESI †), and pyridine nitrogen as proton acceptor, and can be easily distinguished by the resonance of the proton H 23 in 1 H NMR spectra. The NH signal in Z-isomer is downfield shifted in comparison to proton H 23 of the E-isomer. Therefore, the proton H 23 in Z-HL 1 and E-HL 1 resonates at 15.12 and 11.16 ppm, respectively. In the NMR spectra of HL 2 and HL 3 , only one set of signals is present, and proton H 23 gave singlets at 12.01 and 11.85 ppm, respectively, which allowed for their assignment to E-isomers. The most difficult was the assignment of all resonances for H 2 L 4 (Scheme S10). In a similar way to HL 1 (R 1 = R 2 = Me), H 2 L 4 gave two sets of signals, originated from the two isomers, namely Z,Zand E,E-isomers. Immediately after dissolution in DMSO-d 6 , the molar ratio of Z,Z/E,E was 1:2, but in several hours a new equilibrium was reached, in which the molar ratio between the two isomers was 1:4.3. The NH protons involved in the hydrogen bonds (H 31 , H 34 in Scheme S10, ESI †) resonate at 15.12 and 15.06 ppm in Z,Z-H 2 L 4 and at 11.16 and 11.19 ppm in E,E-H 2 L 4 . The presence of Eand Z-isomers is typical for thiosemicarbazones, and our data are in good agreement with those reported in the literature [24,60].
All obtained coumarin-based TSCs (HL 1 -HL 3 and H 2 L 4 ) showed very similar emission spectra in H 2 O, 1% DMSO/H 2 O solutions with a maximum at 451-454 nm (λ em ) when irradiated with 377-386 nm (λ ex ) ( Figure S3, ESI †). Comparison with the fluorescence spectra of triapine (in water (0.5% DMSO), λ ex = 360 nm, λ em = 457 nm) [58], and 7-hydroxycoumarin-3-carboxylic acid (G, λ ex = 350 and 395 nm, λ em = 450 nm, in phosphate buffer (PB), pH 7.4) [59] showed that the combination of two pharmacophores (triapine and coumarin) in the new hybrids HL 1 -HL 3 and H 2 L 4 resulted in a partial quenching of fluorescence, even though the same emission bands with almost the same maxima in the cancer cell-free medium were observed. This fact did not allow for the monitoring of the intracellular distribution of HL 1 -HL 3 and H 2 L 4 .

Synthesis of Complexes
The synthesis of copper(II) complexes was carried out in alcohol (1-3, Cu(HL 1-3 (2) showing that demethylation at the end nitrogen atom of the thiosemicarbazide moiety increases slightly the complex stability [60].

X-ray Crystallography
The result of SC-XRD study of [Cu(L 1 )Cl] (1 ) is shown in Figure 1. The complex crystallised in the triclinic centrosymmetric space group P1 with two crystallographically independent molecules of the complex in the asymmetric unit. The crystal contains co-crystallised water (3.55 molecules per asymmetric unit), which is disorded over several positions. The copper(II) ion adopts a square-planar coordination geometry. The thiosemicarbazone acts as a monoanionic tridentate ligand, which is bound to copper(II) via pyridine nitrogen atom N1, hydrazinic nitrogen atom N2 and thiolato sulfur atom S. The fourth position is occupied by the chlorido co-ligand. The Cu to ligands bond lengths are quoted in the legend to Figure 1. The established coordination mode is typical for copper(II) complexes with pyridine-2-carboxaldehyde thiosemicarbazones [61][62][63]. We refrain from making any comparison of the metrics parameters in this molecule with those in related copper(II) thiosemicarbazonates, taking into account the low-resolution SC-XRD data collected for this compound. The anticancer activity of the copper(II) complexes of TSCs is often related to their redox reactions with physiological reductants. Therefore, the redox properties of Cu(II) complexes were investigated by cyclic voltammetry (CV) and EPR/UV/Vis-NIR spectroelectrochemistry (SEC) in solution, at room temperature.

Electrochemistry and Spectroelectrochemistry
Cyclic voltammograms (CVs) of 1-4 show one irreversible or quasi-reversible oneelectron reduction wave between −0.87 and −0.91 V vs. Fc + /Fc (Figure 2a and Figure S4a in ESI †) that corresponds to the reduction of the Cu(II) to Cu(I) [26,64]. The irreversible oxidation wave at E pa = 0.1 V vs. Fc + /Fc (Figure 2b) can be only seen for complex 3 with potentially redox-active 2,6-dimethyl-4-aminophenol unit. A similar response was observed for the corresponding proligand HL 3 ( Figure S4b, ESI †), likely indicating oxidation of the 2,6-dimethyl-4-aminophenol moiety [65,66]. In situ spectroelectrochemistry in 0.1 M nBu 4 NPF 6 /DMSO solution provides further evidence for chemical irreversibility of the reduction process, as shown for 2 in Figure 3. UV-vis spectrum of Cu(II) complex 2 exhibits absorption maxima at 338 nm and in the 400-430 nm region. The first maximum can be attributed to intraligand π→π* electronic transition, while other bands at higher wavelengths are due to ligand-to-metal chargetransfer (CT) processes [67]. Absorption spectra measured upon cathodic reduction of 2 in the region of the first reduction peak revealed an increase of the initial optical bands at 338 nm and simultaneous decrease of the of the S→Cu(II) CT band (~420 nm) (Figure 3a). There is only partial recovery of the initial optical band after scan reversal (Figure 3b), which provides further evidence that Cu(I) state is not stable, and the complex decomposes with the ligand (or new ligand product) release [64]. Characteristic for Cu(II), the intensity of the room temperature X-band EPR spectrum of 2 decreased upon stepwise application of a negative potential at the first reduction peak, indicating the formation of a d 10 EPRsilent Cu(I) species (see inset in Figure 3b). Endogenous thiols are likely able to reduce copper(II) complexes of TSCs [68] to less stable copper(I) species that may dissociate. The released proligand can act as an Fe-chelator as was already unambiguously confirmed for other TSCs, including morpholine-thiosemicarbazone hybrids [26] and 3-amino-2pyridinecarboxaldehyde-S-methylisothiosemicarbazone [64], and consequently may reduce the tyrosyl radical in R2 RNR. Additionally, thiols may react with Cu(II) ions to form mixtures of Cu(I) and Cu(II) complexes, which could behave either as antioxidants or pro-oxidants, depending on the molar ratio of reactants [69]. Potential dependence of UV−vis spectra with respective cyclic voltammogram (Pt-microstructured honeycomb working electrode, scan rate v = 5 mV s −1 ); (b) evolution of UV−vis spectra in 2D projection in backward scan; blue line represents the initial optical spectrum (inset: EPR spectra measured at the first reduction peak using Pt mesh working electrode).

Anticancer Activity
The cytotoxic potential of ligands HL 1 -HL 3 and H 2 L 4 and their respective Cu(II) complexes was determined against a range of highly resistant cancer cell lines, including triple negative breast cancer cell line MDA-MB-231, colorectal adenocarcinoma cell line COLO-205 and its multi-drug-resistant analogue COLO-320, as well as healthy human lung fibroblasts MRC-5 by a colorimetric MTT assay, and compared to doxorubicin. The results are presented in Table 2 and Figure S5, ESI †, and represent the mean IC 50 values with standard deviations. As can be seen from Table 2, in all tested cell lines the ligands and their respective complexes demonstrated antiproliferative activity in a micromolar concentration range, similar to previously published TSC-coumarin hybrids [41]. The cytotoxicity of the ligands showed the following trend in all cancer cell lines: HL 2 ≈ HL 3 < HL 1 ≤ H 2 L 4 , indicating detrimental effects of aromatic substituents in the thioamide group. On the contrary, the addition of the second TSC fragment has improved the cytotoxicity of HL 1 by a factor of 1.5-2. The cytotoxicity of the ligand H 2 L 4 was comparable to the clinically used doxorubicin in all cancer cells lines, demonstrating its high therapeutic potential. However, this modification has also led to the increased toxicity towards healthy fibroblasts in agreement with high toxicity of 3-AP (selectivity factors (SF) towards MDA-MB-231 and COLO-320 cell lines are 1.3 and 0.9, respectively) [24,25]. In general, HL 1 -HL 3 demonstrated higher selectivity towards MRC-5 cells over MDA-MB-231 cells (SF = 3-10). On the contrary, their selectivity towards COLO cell lines was quite poor and only HL 1 demonstrated reasonable selectivity reflected by SF = 6.1. The coordination of HL 1 and H 2 L 4 to Cu(II) centre resulted in a decrease of anticancer activity in all cancer cell lines. Similarly, Cu(II) complexes of 3-AP demonstrated lower cytotoxicity than metal-free 3-AP [70]. However, Cu(II) complexes of HL 2 and HL 3 retained the cytotoxicity of the ligands in COLO-205 and COLO-320, while being more toxic to MRC-5 (SF = 0.1-0.9). This study demonstrates how slight changes in the 3-AP structure markedly affect the balance between anticancer activity and toxicity to healthy cells. Since the antiproliferative activity of TSCs is often related to inhibition of the R2 RNR protein, and may involve the formation of iron complexes of TSCs, first the ability of iron(II) complex with HL 1 to generate ROS was investigated, followed by the potential of selected compounds prepared in this work to reduce the tyrosyl radical in mR2 RNR.

Redox Activity of [Fe II (L 1 ) 2 ]
To investigate whether the iron complex of the lead HL 1 hybrid is redox active and therefore able to generate ROS in the aqueous environment, the EPR spin trapping experiments were performed using 5,5-dimethyl-1-pyrrolin-N-oxide (DMPO) as the spin trapping agent. The complex [Fe II (L 1 ) 2 ] was prepared by the reaction of an anoxic aqueous solution of FeSO 4 ·7H 2 O with ethanolic solution of HL 1 at 1:2 molar ratio. The formation of [Fe II (L 1 ) 2 ] species was confirmed by electronic absorption spectrum of [Fe II (L 1 ) 2 ] (Figure 4a), which is characteristic for Fe(II) bis-TSC complexes [24,60] and also by positive ion ESI mass spectrometry of the iron(II) complex under air which revealed the presence of a peak at m/z 1042 attributed to [Fe III (L 1 ) 2 ] + .
The EPR spectra of Fe(II)/HL 1 /DMPO/H 2 O-EtOH were recorded either in the presence or absence of H 2 O 2 . In the system containing H 2 O 2 , a four-line EPR signal characteristic for the •OH-DMPO spin adduct was observed (Figure 4b, signal marked with circles). However, also in the absence of H 2 O 2 , the formation of carbon-centred radicals was detected. In this case, both coordinated HL 1 and ethanol [71] may act as HO• scavengers, generating carbon-centred radicals which are trapped by DMPO (Figure 4b, signal marked with stars). It is most probable that the coumarin moiety of HL 1 can readily react with HO• radicals producing carbon-centred radicals. It is important to note that no radicals were formed in the system HL 1 /H 2 O 2 /DMPO/H 2 O-EtOH (4:1, v/v) (Figure 4b, blue line), which indicates the crucial role of Fe(II) for ROS generation, in both systems (+/− H 2 O 2 ). Thus, it may be concluded that complex [Fe II (L 1 ) 2 ] is redox active, which may account for the observed antiproliferative activity of HL 1 in cancer cell lines.

Tyrosyl Radical Reduction in Mouse R2 RNR Protein
The kinetics of tyrosyl radical reduction in mR2 by the proligand HL 1 , and complexes 1 and 3, were investigated by EPR spectroscopy at 30 K ( Figure 5). The results show that HL 1 and 1 exhibit similar tyrosyl radical reduction, approximately 25-30% after 10 min of incubation at 1:1 protein-to-compound mole ratio at 298 K. Furthermore, in the presence of an external reductant (DTT) the reduction was increased to 70-75% in 10 min. On the other hand, complex 3 was not as efficient, resulting in only 20% reduction. It is worth noting that the presence of DTT did not have an effect on radical destruction by this complex, indicating that, although the diferric center of mR2 becomes reduced by DTT and therefore causes the tyrosyl radical to be more susceptible for reduction [13,72], complex 3 may be too large to enter the mR2 hydrophobic pocket that harbours the tyrosyl radical, due to the bulky dimethyl-protected phenolic group of the proligand HL 3 . Furthermore, when compared to triapine, which reduces 100% tyrosyl radical in 3 min, [64] neither HL 1 nor 1 are shown to be efficient mR2 inhibitors. Although HL 1 may chelate iron from mR2 and form the redox active [Fe II (L 1 ) 2 ] complex, it is evident that this species is devoid of the potent inhibitory activity observed previously for [Fe II (3-AP) 2 ] [12,13]. This indicates that the introduction of the coumarin/piperazine moiety decreases the radical-reducing potency of 3-AP towards mR2, which may be attributed to the antioxidative properties of the coumarin-moiety of these hybrid compounds.

Conclusions
An entry to a new series of thiosemicarbazone-coumarin hybrids and their copper(II) complexes via multistep chemical transformations has been provided by this work. The hybrids prepared exhibited antiproliferative activity in human breast adenocarcinoma (MDA-MB-231) and human colorectal adenocarcinoma (COLO-205 and COLO-320) cell lines, with IC 50 in a micromolar concentration range, in the following rank order: HL 2 ≈ HL 3 < HL 1 ≤ H 2 L 4 . These results indicate that aromatic substituents at the thioamide group decrease the antiproliferative activity of the ligand, while the addition of the second TSC fragment (in H 2 L 4 ) leads to its improvement. However, this addition also led to increased toxicity towards healthy fibroblasts, as previously observed for 3-AP [22,23]. Cu(II) complexes of HL 1 and H 2 L 4 demonstrated reduced anticancer activities in all cancer cell lines. However, Cu(II) complexes of HL 2 and HL 3 retained the cytotoxicities of their respective ligands in COLO-205 and COLO-320, while being more toxic to MRC-5. The results from this study confirm once again that the choice of the substituent on 3-AP may greatly affect the balance between anticancer activity and toxicity to healthy cells. It is important to point out that the antiproliferative activity of the lead hybrid HL 1 is comparable to that of 3-AP (in MDA-MB-231) and doxorubicin (in COLO-205 and COLO-320), while HL 1 is much less toxic to healthy MRC-5 cells than both of these drugs. Regarding mR2 RNR inhibition, the lead hybrid HL 1 was shown to be less efficient than 3-AP in tyrosyl radical reduction, which is most likely attributed to the antioxidant properties of the coumarin moiety. It is anticipated that HL 1 can chelate iron from mR2, but that the formed iron complex is not a potent inhibitor such as [Fe II (3-AP) 2 ]. Even though R2 RNR is the biomolecular target for 3-AP, these results indicate that the high antiproliferative activity of HL 1 is not due to inhibition of this enzyme. Another likely target is Topoisomerase IIα (Topo IIα), which regulates the DNA topology during cell division [73,74]. α-N-Heterocyclic TSCs were found to bind to the enzyme ATP binding pocket, while their cytotoxicity was found to correlate with Topo IIα inhibition [75]. Moreover, 2-formylpyridine thiosemicarbazones show Topo IIα inhibition activity, which is further increased by complex formation with copper(II) [76]. The key role in this enhancement of the Topo IIa inhibition is the adoption of square-planar coordination geometry, which was also established for compounds investigated in this work.