Photoactivation of the Cytotoxic Properties of Platinum(II) Complexes through Ligand Photoswitching

The development of photoactivatable metal complexes with potential anticancer properties is a topical area of current investigation. Photoactivated chemotherapy (PACT) using coordination compounds is typically based on photochemical processes occurring at the metal center. In the present study, an innovative approach is applied that takes advantage of the remarkable photochemical properties of diarylethenes. Following a proof-of-concept study with two complexes, namely C1 and C2,1 a series of additional platinum(II) complexes from dithienylcyclopentene-based ligands have been designed and prepared. Like C1 and C2, these new coordination compounds exhibit two thermally stable, interconvertible photoisomers that display distinct properties. The photochemical behavior of ligands L3─L7 has been analyzed by 1H NMR and UV-vis spectroscopy. Subsequently, the corresponding platinum(II) complexes C3─C7 have been synthesized and fully characterized, including by X-ray diffraction for some of them. Next, the interaction of each photoisomer (i.e., containing the open or closed ligand) of the metal complexes with DNA has been examined thoroughly using various techniques, revealing their distinct DNA-binding modes and affinities, as observed for the earlier compounds C1 and C2. The antiproliferative activity of the two forms of the complexes has then been assessed with five cancer


INTRODUCTION
An elegant drug-design approach aimed at enhancing tumor selectivity while reducing systemic toxicity is the use of light activation. 2 Recent advances in laser and fiber-optic technologies have driven the development of various medical applications based on light, 3 including the design of metal-containing prodrugs that can be activated through irradiation. 2,4 Photoactivated chemotherapy (PACT) is a prominent field of investigation, which provides both temporal and spatial control over drug activity, and offers tremendous potential for the treatment of different types of cancer. 5 Photoactivatable prodrugs based on coordination compounds typically involve activation at the metal center, with concomitant generation of pharmacologically active species. 6 Metalmediated PACT can be achieved via distinct mechanisms of action, namely photodissociation 7 and photosensitisation, 8 which are often associated to the redox properties of the metal center, 2 and photothermal reaction. 9 A number of research groups worldwide are developing elegant systems that are photoactivatable through the action of light on the metal. [10][11][12][13][14][15][16][17] The activation of a metallodrug through the photomodification of a coordinated ligand has not been investigated extensively. Different types of organic molecular photoswitches may be used for this purpose, like azobenzenes, spiropyrans or diarylethenes. 18 Azobenzenes and spiropyrans are thermally unstable and gradually revert back to their initial state in the absence of light. 19 In 4 contrast, diarylethenes are more suitable for such biological application as they exhibit negligible thermal relaxation; as a matter of fact, these unique properties have converted them as one of the most interesting classes of photoresponsive molecular devices, as reflected by their numerous applications. 20 In the present study, the potential use of diarylethene moieties to generate potential lightactivatable metallodrugs has been investigated further. 1 Indeed, it has recently been shown that the two thermally stable states of two molecular switches (complexes C1 and C2 in Figure 1) exhibit distinct biological activities, one of the two photoisomers being cytotoxic whereas the other one not. 1 Hence, based on these results, a series of new simple diarylethene-based ligands were designed and prepared, and the corresponding platinum(II) complexes were synthesized.
Subsequently, the DNA-interacting properties of their open and closed forms were examined using various techniques, and their cytotoxic behavior against various cancer cell lines was then evaluated and compared with those of C1 and C2. Promising results were achieved, further indicating that this innovative mechanism of photochemical control may be applied to produce a new class of photoactivatable metallodrugs. It can indeed be stressed that the photoactivation of the cytotoxic properties of the platinum(II) compounds reported herein is based on the photomodification of the coordinated ligand, and not on the metal center (as commonly described in the literature).

EXPERIMENTAL DETAILS
Materials and methods. All reagents and HPLC-grade solvents for the synthesis of the ligands and complexes were purchased from commercial sources and used as received. Ethidium bromide, sodium cacodylate, TAE (Tris Acetate-EDTA) and calf thymus DNA (ct-DNA) were purchased from Sigma-Aldrich. Plasmid pBR322 DNA was purchased from Roche. All reagents used for the in vitro DNA-interaction studies were obtained from Sigma-Aldrich and Invitrogen.
Anhydrous solvents were distilled under an inert atmosphere using a PureSolv™ solvent purification system from Innovative Technologies.
When required, the reactions were performed under an atmosphere of dinitrogen using standard procedures. Column chromatography purifications were carried out in air, using ultrapure silica gel (60-200 μm, 60 Å) from Acros Organics, and monitored by analytical thin-layer chromatography (TLC) using pre-coated aluminum plates.
ESI mass spectrometry was carried out using a LC/MSD-TOF spectrometer from Agilent Technologies equipped with an electrospray ionization (ESI) source, at the Centres Cientifics i Tecnologics de la Universitat de Barcelona. Samples were eluted with a H2O/CH3CN 1:1 mixture and measured in the positive mode. C, H, N and S elemental analyses were performed at the 7 Centres Cientifics i Tecnologics de la Universitat de Barcelona, using a Thermo EA 1108 CHNS/O analyzer from Carlo Erba Instruments.
Spectroscopic measurements in buffered aqueous media were performed in cacodylate buffer solution (1 mM sodium cacodylate, 20 mM NaCl, pH = 7.2), which was prepared with ultrapure water and whose pH was adjusted with HClaq. The UV and visible irradiation of the samples was carried out with a MAX-303 light source from Asahi Spectra, using appropriate bandpass filters.
The UV-Vis absorption spectra were recorded on a Varian Cary 100 Bio spectrophotometer.
Fluorescence and circular dichroism spectra were collected in 1 cm-pathlength quartz cuvettes, using cacodylate buffered solutions.
The stock solutions for the DNA-interaction studies were prepared as follows: the commercial plasmid pBR322 stock solution (250 μg mL ─1 , ca. 385 μMbp) was used as received. A 150 μM stock solution of ct-DNA was prepared by dissolving the highly-polymerized sodium salt of the biomolecule in cacodylate buffer, and the exact concentration was determined by absorbance at 260 nm (εbp = 13200 M ─1 cm ─1 ). 5 mM stock solutions of the studied complexes were freshly prepared in DMSO before each experiment, and further diluted in DMSO/buffer mixtures as required. Stock solutions of closed complexes were obtained by irradiation of the corresponding solutions of the open complexes, at λ < 365 nm, until reaching the photostationary state (PSS). Synthesis 1,5-Bis(5-chloro-2-methyl-3-thienyl)pentane-1,5-dione (1). 21 Aluminum(III) trichloride (29.0 g, 217 mmol) was slowly added to a mixture of 2-chloro-5-methylthiophene (25.0 g, 189 mmol) and glutaryl dichloride (12.0 mL, 94 mmol) dissolved in 200 mL of ice-cold nitromethane. After the addition of AlCl3, the resulting dark red solution was stirred for 3 h at room temperature and 150 mL of ice-cold water were subsequently added in small portions. The reaction mixture was 8 then placed in an ice bath and vigorously stirred for 1 h, until an abundant precipitate formed. This precipitate was then poured over a glass filter, washed with cold n-pentane and finally dried under reduced pressure to yield the crude final product 1 as a pale brown solid (27.6 g, 81%). 1

1,2-Bis(5-chloro-2-methyl-3-thienyl)cyclopentene (2)
. Titanium(IV) tetrachloride (4.4 mL, 40 mmol) was carefully added under a nitrogen atmosphere using a syringe to an ice-cooled suspension of zinc dust (5.2 g, 80 mmol) in 150 mL of anhydrous THF. The resulting grey-blue reaction mixture was slowly heated to 50 °C and stirred for 2 h, after which pyridine (3.2 mL, 40 mmol) was added dropwise, producing a brown solution. After 10 min, 1 (7.2 g, 20 mmol) was added and the resulting mixture was stirred in the dark overnight. Next, the solution was cooled in an ice bath and quenched with a 20% aqueous K2CO3 solution (25 mL), yielding an abundant black precipitate. After vigorous stirring of the precipitate with diethyl ether (40 mL), the remaining solid was isolated by filtration over a glass filter and washed with additional diethyl ether (2 × 25 mL). The combined organic phase was placed in an ice bath for 30 min and filtered again to eliminate the resulting undesired white precipitate. The solution was washed with acidified water (2 × 20 mL), dried over Na2SO4, filtered and concentrated under reduced pressure to give an orange oil. This oil was purified by column chromatography on silica gel with cyclohexane as the eluent, yielding an oil that slowly solidified to colorless product 2 (4.75 g, 72%).  The resulting two-phase system was heated to 50 °C, stirred in the dark overnight and then cooled to room temperature before dichloromethane (25 mL) and water (15 mL) were added. The organic layer was separated, washed with brine (25 mL), dried over Na2SO4, filtered and concentrated under reduced pressure. The crude product was purified by column chromatography on silica gel with dichloromethane as the eluent, yielding the final product 4 as an off-white solid (0.94 g, 82%).

4-Bromo-5-methyl-2-(3-quinolyl)thiophene (5).
Compound 5 was prepared following the procedure described for compound 4, but using 3-bromoquinoline (0.9 mL, 6.7 mmol) instead of 4-bromopyridine. The crude product was purified by column chromatography on silica gel with dichloromethane as the eluent, to yield the final product as an off-white solid (0.81 g, 59%).  bromobenzene (0.4 mL, 3.75 mmol) were dissolved in a solvent mixture containing anhydrous THF (20 mL) and 20% aqueous K2CO3 solution (20 mL), under a dinitrogen atmosphere. This two-phase system was stirred at 50 °C for 15 min and the previous freshly prepared boronic derivative solution was added dropwise with a syringe. The resulting reaction mixture was stirred in the dark overnight, after which it was cooled to room temperature; dichloromethane (25 mL) and water (15 mL) were subsequently added. The organic layer was separated, washed with brine 11 (25 mL), dried over Na2SO4, filtered and concentrated under reduced pressure. The crude product was purified by column chromatography on silica gel with a 19:1 mixture of cyclohexane and ethyl acetate as the eluent, yielding 6 as a slightly colored oil (0.74 g, 66%). 1

1,2-Bis(2-methyl-5-(4-pyridyl)-3-thienyl)cyclopentene (L1).
Ligand L1, whose synthetic pathway was described earlier, 1 was prepared as described below. This procedure was applied for all ligands including a central cyclopentene ring (Ligands L3, L5, L6 and L7). then stirred in the dark overnight, after which it was cooled to room temperature and treated with dichloromethane (25 mL) and water (15 mL). The organic layer was separated, washed with brine 12 (25 mL), dried over Na2SO4, filtered and concentrated under reduced pressure. The crude product was purified by column chromatography on silica gel with ethyl acetate (containing 1% NH3) as the eluent, yielding the final product as a white solid (0.60 g, 48%). 1
Afterwards, the cooling bath was removed and the stirring continued at room temperature for 3 h, after which dichloromethane (25 mL) and water (25 mL) were added. The organic layer was separated, washed with brine, dried over Na2SO4, filtered and concentrated under reduced pressure. The crude product was purified by column chromatography on silica gel using ethyl acetate as the eluent, yielding L4 as a pale blue solid (0.18 g, 14%).
After the incubation, all samples were treated with 4 μL of a xylene cyanol 1x aqueous solution (containing 30% glycerol) and subsequently electrophoretized in agarose gel (1% in TAE buffer, Tris-Acetate-EDTA) for 1 h at 6.25 V cm ─1 , using a Bio-Rad horizontal tank connected to a Consort EV231 variable potential power supply. Next, the gel was treated with SYBR Safe DNA gel stain and subsequently photographed with a Bio-Rad Gel Doc EZ imager.

Preparation of the ligands and their respective platinum(II) complexes.
The diarylethenebased ligands were prepared using common synthetic procedures ( Figure S1). 1,21,27 Ligands L1, L3 and L5-L7 were obtained in moderate to good yields, ranging from 37% to 72%, by Suzuki cross-coupling reaction between the in situ-generated bis-boronic ester derivative of 1,2-bis(5chloro-2-methyl-3-thienyl)cyclopentane and the corresponding bromo(hetero)arene (see Experimental section). 28 Ligands L2 and L4, containing a perfluorinated cyclopentene ring, were synthesized in low yields (26% and 14%, respectively), by reaction of octafluorocyclopentene with the corresponding bromo heteroarylthiophene (see Experimental details). behavior was investigated theoretically, and was ascribed to the presence of a non-photoreactive HOMO─2→LUMO transition with a higher computed oscillator strength value than that of the HOMO→LUMO one, therefore impeding the photocyclization process. 31 The photochemical transformation of all ligands, except L6, can also be followed by 1  After full characterization, platination of the ligands was carried out, generating complexes C1-C7 with good yields, from 42% to 95% (Figure 1). Hence, reaction of two equivalents of cis- Single crystals, suitable for X-ray diffraction studies, could be obtained for C2, C3, C4 and C6.
The crystal structures of C2 1 and C6 31 were described earlier. The X-ray structures of C3 and C4 are reported herein. Crystallographic and refinement parameters are summarized in Tables S1 and   S3 for C3 and C4, respectively (ESI †), and selected coordination bond lengths and angles for C3 are listed in Table S2 (ESI †), whereas those for C4 are shown in Table S4 Figure 2. It can be mentioned here that one may expect that complexes C1-C7 will exhibit a mechanism of action related to that of cisplatin. Hence, hydrolysis of the chlorides of C1-C7 should lead to the formation of aquated species, 36,37 which will bind to DNA by substitution of the water molecules for purine bases. 38 Due to the trans-disposition of the chlorides, it is expected that each platinum ion will mostly produce 3-intrastand cross-links and monofunctional adducts, which will undergo conversion to interstrand cross-links; 39   Closed C5 exhibits the same behavior as its open form (see lanes 3-6 and lanes 7-10 in Figure   2). Thus, closed C5 does not seem to show the intercalative properties observed with the closed isomers of the previous complexes (see above). This feature may arise from (i) steric hindrance caused by the methyl substituents of the pyridine ring of ligand L5 and/or (ii) the relative disposition of the metal centers (meta position instead of para position for the previous complexes). This drastic alteration of the DNA-interacting properties of closed C5 (compared for instance to those of closed C1 and C2), triggered by a subtle modification of the metalcoordinating unit, again indicates that the binding activity of the two photoisomers proceeds via two different mechanisms.
As reported earlier, complex C6 does not display photoswitching properties. 31   It can be noticed that the quenching effect is clearly more pronounced for closed C1 than for open C1 (Figure 3), hence suggesting that the closed photoisomer has a higher capacity to expel DNA-bound EB, and thus confirming its better intercalating properties (see gel electrophoresis results). 45 To properly compare the DNA affinity of all studied species, the fluorescence emission data obtained were used to determine the corresponding Stern-Volmer quenching constants (KSV) for intermolecular deactivations, by applying the Stern-Volmer equation (1): 46 In this expression, I0 is the initial fluorescence intensity of the DNA/EB system, I represents the fluorescence emission intensity after the addition of a quencher, and Q is the quenching molecule, which is the metal complex in the present study. A plot of I0/I versus [complex] at the maximum emission wavelength (i.e. λmax = 610 nm) gives a straight line, whose slope is equal to KSV.
Similar trends were observed for the open and closed photoisomers of all compounds investigated, whose respective Stern-Volmer constants are listed in Table 1.  In all cases, a significantly higher KSV value was obtained for the closed isomer, hence corroborating their higher intercalative behavior. Thus, one may expect that the closed forms of the platinum(II) complexes will induce higher growth inhibitory effects than their open counterparts. It can also be noticed that lower quenching constants were systematically achieved for the fluorinated closed-ring species, compared with their corresponding non-halogenated analogues (see C1 and C2, and C3 and C4 in Table 1). These features agree with the data achieved by gel electrophoresis (see above) and confirm the negative effect of the fluorinated cyclopentane ring on the intercalation of the corresponding coordination compounds. These differences may result from a slight destabilization of the DNA/complex adducts through electrostatic repulsion between the fluorine atoms and the negatively charged phosphate backbone of the double helix. 47 A noticeably reduced dye-displacement ability is exhibited by the mononuclear complex C7 (in its both forms, see Table 1). For instance, closed C7 shows a KSV value that is twice lower than The results achieved with photoinert C6 requires some attention. As indicated by its KSV value (   Figure 5 for complex C2, Figure 6 for C6 and Figure S6 for C1, C3-C5 and C7. Surprisingly  Subsequently, half-maximal inhibitory concentrations (IC50) were determined for the two most interesting platinum(II) compounds, namely C2 and C6 ( Figure S7), using the two cell lines A375 and SW620, carefully selected from the single-point results (see Figures 5 and 6) and specific characteristics. The melanoma A375 cells constitute the most suitable target for PACT drug candidates such as those herein reported, owing to the inherent accessibility to the treatment area. 48 The choice of the colorectal cancer line SW620 arises from its bad prognosis and current shortfall of efficient therapies. 49 The IC50 for cisplatin with these two cell lines were also determined, for comparison purposes. (melanoma), MCF7 (breast carcinoma), SW620 (colorectal adenocarcinoma) and SKOV3 (ovarian adenocarcinoma). The data shown are means ± SD of three independent experiments.
The IC50 data listed in Table 2 corroborate the single-point cell viability assays (see above) and the cytotoxic potential of the selected compounds, which are more efficient than cisplatin for the two cell lines. The IC50 values for closed C2 and open C6 are in the low micromolar range, the compounds being up to 8 times more active than cisplatin (Table 2).    Table S5.
Open C6 induces a drastically distinct alteration of the conformation of the double helix ( Figure   7 and Table S5). The progressive binding of this imidazolic species leads to a severe decrease of the whole CD signal of B-DNA, particularly for the negative band. These spectral changes suggest that C6 has a distinct effect on the secondary structure of the biomolecule, most likely generating local disruptions of the helical turn, and destabilization of the alignment of adjacent base pairs. This effect is consistent with the efficient expulsion of ethidium bromide noted by fluorescence assays (see above).
Since C6 is the sole metal complex of the series exhibiting this behavior, it may be attributed to the methylimidazole units; however, the exact mechanism through which C6 induces such a distinct structural alteration of B-DNA remains to be elucidated. Probably, the specific electronic structure of ligand H6 31 has a direct effect on the ligand-exchange kinetics and/or the binding affinity of the platinum(II) centers, which may thus become more prone to mediate DNA cross-links. These different DNA-interacting properties may explain the higher cytotoxicity of C6, compared with that of the other open complexes (Table 2 and Figures 5, 6 and S7).

CONCLUSIONS
Following earlier promising results, 1 a series of photoswitchable platinum(II) complexes based on photoresponsive dithienylcyclopentene moieties were prepared and fully characterized. In particular, the connectivity of the metal ions was clearly established by X-ray measurements, which revealed a trans disposition of the chlorido ligands and the S-coordination of DMSO. The photochemical properties of the ligands and metal complexes prepared were investigated, which showed that the ligand L6 and the corresponding complex C6 were photoinactive. 31 DNA-interaction studies carried out with all complexes revealed that the activity of these photoisomerisable agents can be efficiently modified through their light-mediated conversion.
Clearly distinct DNA-interacting properties for the open and closed forms of the coordination compounds were observed. For instance, different metal covalent binding and ligand intercalative association were noticed for the two forms. These remarkable results suggest that the activity of each photoisomer may be tailored through appropriate structural modifications, thus opening up design possibilities for the future development of more effective DTC-based metallodrugs (DTC stands for Dithienylcyclopentene).
The antiproliferative activity of the open and closed forms of the complexes were examined, which showed that most of the compounds surprisingly were not cytotoxic; actually, the distinct DNA-interacting ability of each isomer did not translate into interesting results with cells.
Remarkable data were though obtained with complex C2, whose closed form is highly cytotoxic against melanoma and colorectal cancer cells, whereas its open form is non-toxic. These drastically distinct behaviors of open/closed C2 further strengthen the innovative approach proposed in a previous proof of concept study, 1 consisting in using DTC photoswitches to develop novel PACT agents.

Supporting Information
The Supporting Information is available free of charge on the ACS Publications website at DOI:

Accession codes
CCDC 1565000 and 1565001 contain the supplementary crystallographic data for this paper.