Organometallic Half-Sandwich Dichloridoruthenium(II) Complexes with 7-Azaindoles: Synthesis, Characterization and Elucidation of Their Anticancer Inactivity against A2780 Cell Line

A series of organometallic half-sandwich dichloridoruthenium(II) complexes of the general formula [Ru(η 6-p-cym)(naza)Cl2] (1–8; p-cym = p-cymene; naza = 7-azaindole or its derivatives) was synthesised and fully characterized by elemental analysis, mass spectrometry, and infrared and multinuclear NMR spectroscopy. A single-crystal X-ray structural analysis of [Ru(η 6-p-cym)(2Me4Claza)Cl2] (6) revealed a typical piano-stool geometry with an N7-coordination mode of 2-methyl-4-chloro-7-azaindole (2Me4Claza). The complexes have been found to be inactive against human ovarian cancer cell line A2780 up to the highest applied concentration (IC50 > 50.0 μM). An inactivity of the complexes is caused by their instability in water-containing solvents connected with a release of the naza N-donor ligand, as proved by the detailed 1H NMR, mass spectrometry and fluorescence experiments.

Herein we report the synthesis and characterization of the complexes of the general composition [Ru(η 6 -p-cym)(naza)Cl 2 ] (1-8; p-cym = p-cymene) containing 7-azaindole or its derivatives (naza). The reported complexes 1-8 (Fig 1) were, due to known antitumor effect of their structural analogues containing different N-donor ligands, investigated for their in vitro cytotoxic effect against human ovarian cancer cell lines A2780. Because all the complexes were identified as cytotoxic inactive (IC 50 > 50.0 μM), we strived to investigate and explain the reasons of their inactivity by means of 1 H NMR, ESI+ mass spectrometry and fluorescence studies.
X-ray crystallographic data for 6 have been deposited in the Cambridge Structural Database under the accession Cambridge Crystallographic Data Centre number CCDC 1416250. The crystal data and structure refinements are given in S1 Table. The molecular graphics were drawn and additional structural calculations were interpreted using DIAMOND [26] and Mercury [27].

Cell culture and in vitro cytotoxicity testing
The in vitro cytotoxicity towards human ovarian carcinoma A2780 (ECACC No. 93112519) was tested by an MTT assay evaluated spectrophotometrically at 540 nm (TECAN, Schoeller Instruments LLC). The cancer cell lines were cultured according to the ECACC instructions and were maintained at 37°C and 5% CO 2 in humidified incubator. 1-8, cisplatin and naza (0.01-50.0 μM concentrations) interacted with the cancer cells for 24 h, using 96-well culture plates. The cells was tested in parallel with vehicle (DMF; 0.1%, v/v), and Triton X-100 (1%, v/ v) to assess the minimal (100% viability) and maximal (0% viability) cell damage, respectively. The cytotoxicity data were received from three independent experiments (each conducted in triplicate) using cells from three consecutive passages.

Studies of solvolysis and interactions with glutathione (GSH)
The Fluorescence quenching experiments ctDNA (154 μM) and EtBr (3 mM) were mixed together in TRIS/NaCl buffer (pH = 7.2) and incubated for 30 min at ambient temperature. 0, 100, 200, 400, 600 and 1000 μL of the representative complexes 2, 5 and 8 (stock solutions of the 150 μM concentration) dissolved in 10% methanol solution in TRIS/NaCl buffer were added to the EtBr/ctDNA system and the volume was refilled to 3 mL with TRIS/NaCl buffer. The mixtures were incubated at ambient temperature for next 15 min. The emitted fluorescence (546 nm excitation wavelength) was recorded on a fluorescence spectrometer AvaSpec HS1024x122TE using a 1 cm quartz cell.
Molecular and crystal structure of 6. The crystals of [Ru(η 6 -p-cym)(2Me4Claza)Cl 2 ] (6) suitable for a single-crystal X-ray analysis were prepared by a diffusion of diethyl ether into the saturated chloroform solution of 6. The molecular and crystal structures are depicted in Fig 3,  and S3 Fig, respectively, while the selected bond lengths and angles can be found in Table 1.
With respect to the aforementioned findings regarding the mutually different in vitro cytotoxicity of 1-8 against A2780, as compared with their analogues containing different monodentate N-donor ligands [14][15][16][17], we decided to perform several relevant experiments ( 1 H NMR, ESI+ mass spectrometry), designed to shed a light on the reasons of inactivity of 1-8 (studied for the representative complexes 2 and 8). The experiments were designed to prove whether: 1/ the complexes are stable under the used experimental condition and inactivity is due to low sensitivity of the A2780 cells towards 1-8 (in this case, different cancer cell line could be used); and 2/ the complexes are unstable under the experimental condition used and thus unsuitable for further biological studies.
1 H NMR spectroscopy of complexes 2 and 8 in DMF-d 7 The reason why a pure DMF-d 7 was utilized for the stability study is that this solvent is typically used for the pre-dissolution (and consequently, the DMF-d 7 solution is diluted by medium to a maximal DMF concentration of 0.1%) of the corresponding complex during the cytotoxicity testing. Logically, it was of a great interestto find out if the composition of the com-  to the naza and p-cymene ligands, which revealed at 7.75/5.78/5.49 ppm (for 2) and 7.48/5.99/ 5.87 ppm (for 8), may be associated with the formation of decomposition products, probably connected with the solvolysis (substitution of the chlorido ligand(s) with DMF). This finding is, in the case of 8, supported by an increasing portion of this impurity from < 5% in the fresh DMF-d 7 solution to ca 20% after 48 h (Fig 4). 1 H NMR spectroscopy and ESI mass spectrometry of complexes 2 and 8 in 10% DMFd 7 /D 2 O After the addition of D 2 O into the DMF-d 7 solution of 2, four pairs of signals belonging to C13,15-H 2 /C12,16-H 2 were detected at 6.10/5.86, 6.23/5.97, 5.95/5.72 and 5.50/5.30 ppm, with the integral intensities being 1.00, 0.25, 0.36, and 0.30, respectively (S4 Fig, S5 Fig) Addition of D 2 O into the DMF-d 7 solution of 8 led to the formation of white precipitate which was centrifuged, and the isolated solid was dissolved in DMF-d 7 and proved to be free 3I5Braza by 1 H NMR experiment. After removing of the precipitate, the 1 H NMR spectrum of 8 showed, in total, four pairs of doublets belonging to C13,15-H 2 /C12,16-H 2 detected at 5.51/ 5.31, 5.64/5.44, 5.95/5.75 and 6.10/5.87 ppm, with the integral intensities being 0.18, 0.15, 1.00, and 1.00, respectively (Fig 4). No 3I5Braza signals were found in the appropriate proton spectra. In other words, addition of D 2 O into the DMF-d 7 solution of 8 led to its complete decomposition connected with a release of 3I5Braza (Fig 4). Similarly to 2, the mass spectrum of 8 dissolved in 10% DMF/H 2 O revealed [Ru 2 (η 6 -p-cym) 2 Cl 3 ] + , [Ru(η 6 -p-cym)Cl] + , the overlapped peaks of [Ru 2 (η 6 -p-cym) 2 (OH)Cl 2 ] + and [Ru 2 (η 6 -p-cym) 2 (OH) 3 Cl] + , {[Ru 2 (η 6 -pcym) 2 Cl]+O} + , [Ru(η 6 -p-cym)(DMF)Cl] + and [Ru(η 6 -p-cym)(H 2 O)Cl] + (288.9 m/z), indicating analogical behaviour of both the studied complexes 2 and 8 in the used mixture of solvents.
Generally said, dissolving of the studied complexes in 10% DMF-d 7 /D 2 O (or 10% DMF/ H 2 O) led to the release of naza and the formation of low active or most probably non-potent ruthenium-containing species. Since the used mixture of solvents was similar to that one used for in vitro cytotoxicity testing (0.1% DMF in RPMI-1640 medium), it can be anticipated that similar processes proceeded within the performed biological testing, altogether resulting in inactivity of the studied complexes against the used human cancer cell line.
1 H NMR spectroscopy of complexes 2 and 8 in MeOD-d 4 Since it was observed that the studied complexes are stable in CDCl 3 (see section Chemistry, Fig 2, with a coordination ability index of -2.2) and unstable in DMF-d 7 (see above, with a coordination ability index of -0.2), as judged by 1 H NMR spectra, we strived to investigate their solution behaviour also in another relevant solvent (i.e. methanol), with a coordination ability of -0.4 [34].  (Fig 4). These signals pointed out the fact that both the p-cymene and naza ligands are present in both the complex species. Moreover, another set of the naza signals was also found in the spectra of both 2 and 8. The position of the C2-H signal is consistent with the free 3Claza, as proved for 2 (7.42 ppm). However, the same could not be proved in the case of complex 8 because it contains 3I5Braza which is insoluble in methanol. This indicated that the mentioned set of the 3I5Braza signals detected in the MeOD-d 4 solution of 8 most probably did not belong to a released 3I5Braza but to a complex containing the mentioned ligand. Moreover, in the case of 8, another pair of p-cymene aromatic hydrogen atom signals (i.e. C13,15-H 2 /C12,16-H 2 ) was observed at 5.79/5.54 ppm, but it was not accompanied by the appropriate signals (C2-H) at the 3I5Braza region (Fig 4).
Overall, the individual signals of the spectra as well as their integral intensities did not change in time up to 48 h of standing at ambient temperature for both 2 and 8 (Fig 4, S4 Fig). Thus, it could be concluded that at least a part of the complexes 2 and 8 decomposed in MeOD-d 4 together with the release of naza, which correlates well with the mass spectrometry results obtained for MeOH solutions of the studied complexes, as discussed above. It can be concluded that, as in the case of 10% DMF-d 7 /D 2 O, 2 and 8 rapidly decompose/ solvolyse in the used 10% MeOD-d 4 /D 2 O solution, which is in the case of 8 provably connected with a release of 3I5Braza. The solvolysis (most probably hydrolysis) of 2 and 8 was indicated also by ESI+ mass spectrometry utilizing the samples dissolved in 10% MeOH/H 2 O solution. As can be seen from S2 Fig and S6 Fig, the [35]. With respect to the above mentioned findings, it is evident that most of the studied complex 2 decomposed leading to the formation of an Ru-containing adduct with the deprotonated glutathione coordinated through the S-atom (Ru-SG). An integral intensity of the signals of the Ru-SG adduct was not consistent with the above mentioned p-cymene signals, proving that the Ru-SG adduct does not contain neither p-cymene nor 3Claza ligand. Interestingly, a ratio of Cys-α and Cys-β signals of free and coordinated glutathione (1: 9 for the fresh solution, as mentioned above) changed to ca 1: 1 after 48 h of standing at the ambient temperature, meaning that the formed Ru-SG adduct is unstable in the used mixture of solvents, which is connected with a release of GSH. 1  MeOD-d 4 /D 2 O. However, an integral intensity ratio (9: 1) was found to be inverse as compared with 2, and did not change in time, which showed that equilibrium between free GSH and Ru-SG adduct is reached rapidly. Again, the Ru-SG adduct does not contain neither 3Claza nor pcymene ligands. The results of 1 H NMR studies proved that interactions of 2 and 8 with GSH led to the formation of Ru-SG adduct and to a complete release of naza from the structures of 2 and 8. These results are consistent with a detection of various adducts with GSH in the ESI+ mass spectra of the studied complexes dissolved in 10% MeOH/H 2 O. Concretely, the peaks, whose mass and isotopic pattern correspond to [Ru(η 6 -p-cym)(GS)] + (542.1 m/z), {[Ru 2 (η 6 -pcym) 2 (GS) 2 ]-H} + (1082.7 m/z) and [Ru 2 (η 6 -p-cym) 2 (GS) 3 ] + (1390.2 m/z) formed by an interaction of GSH with 2 (S7 Fig), were detected. Surprisingly, the analogical ESI+ mass spectrometry experiments performed for 8 led to the finding that this technique is not suitable for the solution of this issue because the spectra revealed no peaks associated that the adduct containing Ru-SG species.
Fluorescence quenching experiments. It is well-known that intercalative binding of EtBr to DNA leads to fluorescence emission, which could be quenched by replacement of EtBr from the mentioned EtBr/DNA adducts by various agents (e.g. transition metal complexes) [36,37]. On the other hand, 7-azaindole and its derivatives are known fluorescence emitting agents themselves [38][39][40]. Being aware of these facts and the above discussed results of 1 H NMR spectroscopy and ESI+ mass spectrometry indicating a release of the naza ligands from the structures of the studied complexes, it was of great interest to perform the fluorescence quenching experiments on EtBr/ctDNA in the presence of the selected representatives of 1-8.
After the addition of the representative Ru(II) complexes 2, 5 or 8, the concentrationdependent fluorescence quenching of EtBr was observed, showing on the ability of the studied complexes or their decomposition/solvolysis product/s to interact with the used ctDNA (Fig 5  for 5); note: 5 was used together with the representative complexes 2 and 8, because 5Faza contained in its structure showed markedly higher fluorescence as compared with 3Claza (contained in 2; lower fluorescence) and 3I5Braza (contained in 8; no fluorescence because of limited solubility in the medium used). Except the fluorescence maximum of EtBr/ctDNA adduct detected at 615 nm, another peak showed in the obtained fluorescence spectra of 2 (at 415 nm) and 5 (at 405 nm; Fig 5). The position of this peak corresponds to that of free naza (inset in Fig 5), which proved, as anticipated, a release of the naza ligands.

Conclusions
The organometallic [Ru(η 6 -p-cym)(naza)Cl 2 ] complexes (1-8; p-cym = p-cymene; naza = 7-azaindole or its derivatives) were prepared and thoroughly characterized by relevant techniques including a crystallographic study of [Ru(η 6 -p-cym)(2Me4Claza)Cl 2 ] (6) showing a piano-stool arrangement with N-donor ligand coordination through its N7 atom. The complexes were studied for their anticancer activity against the A2780 human cancer cell line, however, no cytotoxicity was found up to the tested concentration of 50 μM. That is why we were looking for the reason of the inactivity. Thus, the complexes were studied for their stability in various solvents (MeOH, DMF, and their mixtures with water). These studies revealed the complexes to be highly unstable, because besides a rapid solvolysis they decompose to the starting ruthenium(II) compound, [Ru(μ-Cl)(η 6 -p-cym)Cl] 2 and/or its solvolysis products, which is connected with a release of naza. With respect to instability of the studied complexes in the solvents used, it may be concluded that the anticancer inactivity of the compounds is associated with this property and formation of inactive species (starting ruthenium(II) dimer, released naza, solvolysis products). It has to be noted that such findings are rather unexpected because the literature data clearly reveal that complexes of a general formula [Ru(η 6 -p-cym)(L)Cl 2 ] should show anticancer activity, as can be seen from references [14][15][16][17] reporting the complexes of the mentioned general formula potent against A2780 ovarian carcinoma cells used also in this work.  (6). The drawing shows the formation of supramolecular 3D structure through the selected C12-H12ÁÁÁC6, C13-H13ÁÁÁC19, C13ÁÁÁCl1, C16-H16ÁÁÁCl2, C18-H18AÁÁÁC5 and C20-H20AÁÁÁC16 non-covalent contacts (red dashed lines); symmetry codes: ii) 1.5-x, y-0.5, 0.5-z; iii) x, y-1, z; iv) 1-x, 1-y, -z; v) 1-x, -y, -z; vi) 1.5-x, y+0.5, 0.5-z; vii) x, y+1, z. The hydrogen atoms not involved in the depicted non-covalent contacts were omitted for clarity.  Table. Crystal data and structure refinement for [Ru(η 6 -p-cym)(2Me4Claza)Cl 2 ] (6) (PDF) S2 Table. Selected bond lengths (Å) and angles (°) of non-covalent contacts detected in the crystal structure of [Ru(η 6 -p-cym)(2Me4Claza)Cl 2 ] (6). (PDF) S1 Text. The results of FTIR spectroscopy and ESI+ mass spectrometry of 1-8. (PDF)