Organometallic Nucleoside Analogues: Effect of the Metallocene Metal Atom on Cancer Cell Line Toxicity

A new chiral organometallic nucleoside analogue containing ruthenocene is reported, in which alkylthymine and alkylhydroxyl groups are attached in adjacent positions on one cyclopentadienyl ring. The synthetic procedures for this metallocene derivative and two control compounds are described, along with their characterisation that include cyclic voltammetry and X-ray crystallography. Their biological activities in a human pancreatic cancer cell line (MIA-Pa-Ca-2) were significantly lower than those for three previously reported analogous ferrocene compounds, indicating that the choice of metallocene metal atom (Fe or Ru) plays an pivotal role in determining the anticancer properties of these nucleoside analogues, which in turn suggests a different mode of action from that of a conventional nucleoside analogue.


Introduction
A topical area within the field of metal-based anticancer drug research involves examining the effect of incorporating organometallic moieties into known organic drugs and related biological molecules. 1,2 Ferrocene is a popular choice in this respect as a so-called bioisosteric group 2 because of its stability and wellunderstood reactivity and electrochemistry. Its incorporation into the breast cancer drug tamoxifen to form the ferrocifen family of compounds 3,4 has revealed potent activities in different cell lines to the parent compound. This suggests novel modes of action related to the redox properties of the ferrocene unit, which may help combat drug resistance in the clinic.
Nucleoside and nucleobase analogues are an important class of chemotherapeutic agent, with 5-fluorouracil (5-FU) 5 and gemcitabine (Gem) 6 two examples of leading drugs on the market. This presents a similar opportunity to decorate and derivatise the components of DNA/RNA with organometallic groups to give a range of new biologically active and medicinally relevant compounds. 7,8 As part of our work in this area, we previously reported the thymidine analogue 1-(S,R p )-Fe in which the five-membered Cp ring of ferrocene replaced the five-membered sugar ring of the nucleoside (Fig. 1). 9 This compound demonstrated excellent anticancer activities in a range of human cancer cell lines, with both the hydroxyl linker and the nucleobase moiety required for optimal cytotoxicity. A subsequent structure activity relationship (SAR) study found a correlation between the IC 50 values in cancer cells and the length of the hydroxyalkyl linker in these so-called ferronucleosides. 10 In continuing with this line of enquiry, we next decided to consider the role played by the metal atom in the lead compound 1-(S,R p )-Fe, the subject of this report. In metallocene-based drug discovery, changing the metal from iron to ruthenium is a worthwhile endeavour, given the stability of ruthenocene and its amenability to functionalisation. Furthermore, and of particular relevance from an SAR point of view, ruthenocenes have different redox properties to ferrocenes, having more positive oxidation potentials and less reversible electrochemistry. 11 It follows that any difference in biological activity between the two metallocenes could indicate a role for redox processes in the mode of action. Indeed, work on ruthenocifen derivatives has indicated different anticancer activities to the ferrocifens, with their biological behaviour more similar to the parent organic compound tamoxifen. 3,12 However despite these findings, other reports on the biological activities of ruthenocene compounds 13 or organoruthenium nucleobase derivatives 8a are relatively rare. Here we report on the synthesis and anticancer properties of 1-(S,R p )-Ru, the direct ruthenocene analogue of 1-(S,R p )-Fe, and two related control compounds (Fig. 2). Our findings do indeed suggest an important role for the metal atom in controlling the anticancer activities of these metallocene-containing nucleoside analogues.

Synthesis
We considered that the synthesis of 1-(S,R p )-Ru and the two control compounds 2-(S) and 3 would allow a direct comparison with the three analogous ferrocene compounds previously reported 9 and also enable similar synthetic routes to be followed. In the case of the main target 1-(S,R p )-Ru, this meant building the compound up from the known acetoxy derivative 4-(R) 14 (Scheme 1), which itself was prepared via a two-step route from acetylruthenocene (see ESI). This was then treated with NHMe 2 to give the ruthenocene version of Ugi's amine 5-(R), 14 whose chiral purity was found to be greater than 98%, as evidenced by chiral HPLC (see ESI). Its X-ray structure was determined for the first time from crystals grown from a solution of the racemate in DCM layered with hexane (see ESI). † The next step was to introduce planar chirality through the diastereoselective synthesis of 6-(R,S p ) via treatment with n-BuLi in diethylether and then quenching with iodine in THF. This compound was then converted to the acetoxy derivative 7-(R,S p ) by heating at 50 ᵒC for two hours in acetic anhydride. A short reaction time and a relatively low temperature were used to avoid elimination of the amine group to give the alkene. The arm was then extended to three carbon atoms by reacting with freshly prepared 1-ethoxyvinyloxy trimethylsilane to give the ethyl ester product 8-(S,S p ). The ester was reduced to the corresponding alcohol 9-(S,S p ) using the mild reducing agent DIBAL-H, before being protected with the TBDPS group to give compound 10-(S,S p ). This compound was then formylated in dry ether in two steps in situ by reacting with n-BuLi in a lithium-halogen exchange followed by addition of DMF to give compound 11-(S,S p ). A Wittig reaction on the aldehyde added another carbon atom to give the alkene 12-(S,R p ), which was then converted to the primary alcohol 13-(S,R p ) by hydroboration-oxidation with BH 3 .THF. Finally, a Mitsunobu coupling reaction on the alcohol 13-(S,R p ) with benzoylprotected thymine gave the fully protected product, which was treated first with TBAF to remove the silyl group and then with ammonia in methanol to remove the benzoyl group, giving the target compound 1-(S,R p )-Ru. HPLC analysis on this compound confirmed its formation in high chiral purity (>98%, see ESI). Crystals suitable for X-ray crystallography were grown from an acetonitrile solution of the racemate at 0 ᵒC (Fig. 2). † An internal O-H…O H-bond is formed between the hydroxyl hydrogen atom and one carbonyl oxygen atom on the thymine base. Intramolecular H-bonding has previously been observed within other bioorganometallic compounds. 8d,25 Scheme 1 Synthesis route for the target compound 1-(S,R p )-Ru from synthon 4-(R).

Fig. 2
Crystal structure of one of the two crystallographically-independent molecules of 1-(S,R p )-Ru with ellipsoids drawn at 50% probability level. The structure also contains two independent molecules of acetonitrile, which have been omitted for clarity. Intramolecular hydrogen bonding is shown using a dotted line.
The synthesis of control compounds 2-(S) and 3 also started from compound 4-(R), and proceeded through the routes depicted in Scheme 2. The chiral alcohol was obtained via a linker extension reaction using 1-ethoxyvinyloxy trimethylsilane, followed by reduction of the ester 14-(S) with LiAlH 4 . Achiral 3 was obtained in four steps, first involving elimination of the acetoxy group to give the vinyl ruthenocene 15. Crystals of this compound suitable for X-ray diffraction were successfully grown from a solution of DCM layered with hexane (see ESI). A hydroboration-oxidation reaction then yielded the anti-Markovnikov product 16 containing the desired hydroxyethyl linker. The X-ray structure of this compound was also obtained from crystals grown using the same conditions (see ESI). The route was completed using the same methodology described earlier via a Mitsunobu coupling reaction with the protected thymine base to give the protected product 17, which was then deprotected with ammonia to give the target compound 3. Crystals of the latter were grown by slow evaporation from a solution of ethyl acetate layered with hexane. The resulting X-ray structure, showing the correct bond connectivity, is depicted in Fig. 4.

Electrochemistry
The electrochemistry of ruthenocene is more complicated than that of ferrocene. The 17-electron ruthenocenium cation is considerably more unstable and reactive than its ferrocene counterpart with Ru(VI) products formed from both the chemical 15,16 and electrochemical 11,16,17 oxidation of ruthenocene. The appearance of ruthenocene cyclic voltammograms (CVs) and those of its derivatives are highly dependent on the type of solvent and electrolyte. 18,19,20 In non-coordinating electrolyte systems and in the presence of noncoordinating boron-containing electrolytes that do not form ion pairs, the cation has been reported to form two different dimers in a temperature-dependent ratio. 20,21 In coordinating solvents and in the presence of more conventional electrolytes, 18 All three ruthenocene compounds showed a similar EC (electrochemical-chemical) oxidation process at a positive potential value, with no return wave observed under the conditions used. Voltammograms for 1-(S,R p )-Ru at different scan rates are displayed in Fig. 5, with those for 2-(S) and 3 presented in the ESI. The E pa data for all three compounds are presented in Table 1, along with the corresponding value for the ferrocene analogue 1-(S,R p )-Fe, which is considerably more negative (E pa = 455 mV, E 1/2 = 424 mV 10 ). This difference in value reflects the large difference in the relative stabilities between the oxidized and reduced forms of the two metallocenes, with the ferrocene derivative clearly being thermodynamically much easier to oxidise than its ruthenocene counterpart. The ferrocene derivative also shows reversible electrochemistry. The increase in E pa for compounds 2-(S) and 3 compared with 1-(S,R p )-Ru can be explained by a greater inductive effect (+I) as the number of electron donating groups on the Cp ring increases, giving more stability to the charged ruthenocenium ion. The same trend is observed for the analogous ferrocene control compounds of 2-(S) and 3, which have E pa values of 497 mV and 540 mV respectively vs. dmfc (see ESI).

Biological studies
The three ruthenocene compounds were next tested for cytotoxic activity in the pancreatic ductal adenocarcinoma cell line MIA-Pa-Ca-2 and compared with the ferrocene counterpart 1-(S,R p )-Fe as well as with cisplatin. Assays were performed after 4 days incubation time using crystal violet staining. Cell viabilities, expressed as a percentage of a negative control, were plotted against concentration (µM) as shown in Fig. 6, with the resulting IC 50 values presented in Table 1. As found previously for other cancer cell lines, 9 the IC 50 value for the ferrocene derivative 1-(S,R p )-Fe sits in the low micromolar range, with a value similar to that of cisplatin. However, the five-fold reduction in the toxicity for the ruthenocene analogue clearly shows that the identity of the metal ion has a significant impact on cytotoxicity. ‡ It is worth noting that the control compounds 2-(S) and 3 were even less toxic than 1-(S,R p )-Ru, with IC 50 values of >80 µM. This agrees with our previous findings on analogous and related ferrocene compounds, 9,10 in that those metallocenes that are more electron rich, for example by having two groups attached to one cyclopentadienyl ring, are more cytotoxic. Indeed the previously published ferrocene analogues of 2-(S) and 3, which display more positive E pa values, are less toxic than 1-(S,R p )-Fe. 9 Overall the trend in the biological data supports the hypothesis that there is a significant relationship between the redox properties of the metallocene units in this series and cancer cell line toxicity.

Conclusion
A ruthenocene-containing nucleoside analogue and two control compounds have been synthesised and fully characterized by a combination of spectroscopic, X-ray crystallography and electrochemical measurements. Their oxidation potentials were affected by the type and number of the linker groups attached to the ruthenocene unit. All three compounds gave very low biological activities in MIA-Pa-Ca-2 pancreatic ductal adenoma carcinoma cells, with IC 50 values for the two mono-functionalised controls higher than that for the bis-functionalised target compound. The main finding of this study is the five-fold difference in cytotoxicity between 1-(S,R p )-Ru and its ferrocene counterpart 1-(S,R p )-Fe. Given their otherwise identical chemical structures and stereochemistries, this difference can confidently be attributed to the change in metal atom from iron  to ruthenium. While such a change would make little difference to a metallocene's size or lipophilicity, it clearly does affect its redox properties. The ferrocenes in this series demonstrate more reversible electrochemistry than their ruthenocene counterparts, with their oxidised forms accessible at significantly lower potentials. These differences in electrochemical behaviour signify an important role for the iron atom in determining the anticancer activity of the lead compound 1-(S,R p )-Fe. This in turn suggests a mode of action different from that of a conventional nucleoside analogue, one that points more towards intra-cellular redox-triggered and ROSmediated pathways leading to cell death. This line of enquiry is currently under investigation in our laboratory. The Ugi amine 5-(R) (0.1 g, 0.33 mmol) was dissolved in Et 2 O (5 mL) at room temperature. n-BuLi (0.3 mL, 1.7 M, 2 eq) was added and the mixture stirred overnight. The mixture was cooled to −78 °C, and iodine (0.23 g, 0.91 mmol), in THF (10 mL), was added over 10 mins. The mixture was stirred at −78 °C for 90 mins before being warmed to room temperature and stirred for an additional 90 mins. The reaction was then quenched at 0 °C with sodium thiosulfate (10 mL, 25% w/v

(S)-1-[α-Methyl-(3-(hydroxy)propyl)]ruthenocene, 2-(S)
Compound 14-(S) (0.48 g, 1.39 mmol, 1 eq) was dissolved in dry THF and stirred at 0 o C for 5 mins under an atmosphere of argon. LiAlH 4 (0.11 g, 2.78 mmol, 2 eq) was added slowly, and the solution stirred for two hrs at RT. The reaction was quenched with a saturated solution of sodium potassium-tartrate (10 mL), extracted with diethylether and dried over MgSO 4 . The solvent was removed in vacuo and the crude product purified by flash column chromatography on silica gel using an eluent system of 15% Et 2 O in hexane. The solvent was removed in vacuo to give the titled compound as a yellow oily product (0.35 g, 83%). δH

2-[(Thyminyl) ethyl]-ruthenocene, 3
Triphenylphosphine (0.149 g, 0.567 mmol), N3-benzoylthymine (0.11 g, 0.49 mmol), and 16 (53 mg, 0.189 mmol) were dissolved in THF (10 mL) and the mixture was stirred for 10 mins at room temperature. The flask was then covered with foil, and DIAD (0.11 mL, 0.57 mmol) was added. The mixture was heated to 65 °C for 2 hrs. The solvent was evaporated and the residue redissolved in EtOAc (15 mL), washed with brine (10 mL) and water (5 mL), and dried over Na 2 SO 4 . The solvent was removed in vacuo to yield the protected product. Deprotection of the benzoyl group was achieved by treating the crude with ammonia solution (7 N in methanol, 5 mL) for 1 h. The solvent was then removed in vacuo and the residue purified via flash column chromatography (30% EtOAc in hexane) to give the product (43 mg, 59% Cleaning of all glassware was achieved by soaking overnight in 1:1 ammonia (35%) and hydrogen peroxide (30%), followed by multiple rinsing with ultrapure water (from a Millipore tandem Elix-A10 system, resistivity > 18 MΩ cm, TOC < 5 ppb). The glassware was then left overnight in ultrapure water, then rinsed again and dried in an oven prior to use. The electrodes were cleaned as follows before their use: the R.E was cleaned with dry acetonitrile and the C.E. was flame annealed. The W.E was cleaned by polishing with aqueous slurries of successively finer grades of alumina (1.0 μm, 0.3 μm and 0.05 μm) and then rinsed with ultrapure water and MeCN, dried with a flow of argon, then kept in analyte solution. used in all calculations. The final R 1 was 0.0369 (I > 2σ(I)) and wR 2 was 0.0967 (all data). The crystal was a non-merohedral twin with the two domains related by 180 about the reciprocal direction [0 0 1] with the refined percentage ratio 59.4 (1) : 40.6 (1). The hydrogen atom bonded to N(3) was located in the electron density and its position refined. All remaining hydrogen atoms were fixed as riding models and the isotropic thermal parameters (U iso ) of all hydrogen atoms were based on the U eq of the parent atom. CCDC 1953303 -CCDC 1953308 contain the supplementary crystallographic data for this paper. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.

Biological studies
MIAPaCa2 (85062806) pancreatic ductal adenoma cancer cells were purchased from the European Collection of Authenticated Cell Cultures. Cell culture media and supplements were purchased from Gibco (Thermo Scientific), all plasticware was purchased from Greiner Bio-One. Cells were maintained at 37 °C in a 5% CO 2 humidified incubator and grown in T 75 tissue culture flasks in DMEM supplemented with 10% (v/v) foetal bovine serum, 100 U/mL penicillin, 100 µg/mL streptomycin, and 2 mM l-glutamine. Cells were sub-cultured twice weekly before confluency using a standard trypsin-EDTA protocol. Cell cultures were confirmed free from Please do not adjust margins Please do not adjust margins Mycoplasma sp. contamination using the EZ-PCR mycoplasma detection kit according to the manufacturer's instructions.
Crystal violet assay: Cells were sub-cultured into 96-well plates at a density of 6250 cells per well in 100 µL of complete DMEM and left overnight to allow the cells to attach. The next day culture media was removed and replaced with fresh media containing test compounds (0-80 µM) dissolved in DMEM with a final concentration of 0.5% v/v DMSO prepared from 50 mM stock solutions, except cisplatin which was prepared as a 2 mM stock solution in phosphate buffered saline (PBS). All cultures were incubated for 72 hrs prior to commencement of the crystal violet assay as described below. The old media was removed, cells washed with 100 μL of PBS, then 100 μL 4% v/v paraformaldehyde was added. After 15 min this was removed and 100 μL of crystal violet solution (0.5% w/v in 10% v/v ethanol) added and plates incubated for 20 min. Next the crystal violet solution was removed, and plates washed with PBS (4 x 100 μL) before being allowed to air dry for 20 min at room temperature. Finally, samples were solubilized using 10 % v/v acetic acid before before measuring the absorbance at 590 nm using a well-plate reader. Changing the metal atom within a metallocene nucleoside analogue from iron to ruthenium results in a five-fold reduction in biological activity in a pancreatic cancer cell line.