Arene Ruthenium Metalla-Assemblies with Anthracene Moieties for PDT Applications

The synthesis and characterization of three metalla-rectangles of the general formula [Ru4(η-p-cymene)4(μ4-clip)2(μ2-Lanthr)2][CF3SO3]4 (Lanthr: 9,10-bis(3,3’-ethynylpyridyl) anthracene; clip = oxa: oxalato; dobq: 2,5-dioxido-1,4-benzoquinonato; donq: 5,8-dioxido-1, 4-naphthoquinonato) are presented. The molecular structure of the metalla-rectangle [Ru4(η-p-cymene)4(μ4-oxa)2(μ2-Lanthr)2] has been confirmed by the single-crystal X-ray structure analysis of [Ru4(η-p-cymene)4(μ4-oxa)2(μ2-Lanthr)2][CF3SO3]4 · 4 acetone (A2 · 4 acetone), thus showing the anthracene moieties to be available for reaction with oxygen. While the formation of the endoperoxide form of Lanthr was observed in solution upon white light irradiation, the same reaction does not occur when Lanthr is part of the metalla-assemblies.


Introduction
Nowadays, one of the challenges in treating cancers is increasing the drug efficacy while limiting or even annihilating side effects. Photodynamic therapy (PDT) can possibly achieve these goals [1,2]. Although still emerging and under development, this method is already used in the clinic or under clinical trials for skin disorders [3][4][5], infections [6,7], superficial cancers [3,[8][9][10] and for deeper cancers [3,[11][12][13][14] (see Table 1). PDT involves three main elements which are individually harmless: a light-absorbing molecule called photosensitizer (PS), oxygen and light. [15] For medical treatment, the PS is administered either systemically or topically and often intravenously. After a period of systemic distribution, the PS accumulates in the tumor, thanks to both selective accumulation and selective retention [16]. Finally, the tumor is irradiated by light, which activates the PS (Figure 1, part A). At that point, the combination of the PDT elements becomes dangerous for the cells. The PS, at its singlet ground state, is excited and reaches an unstable singlet state S 1 * . Then, the PS reemits this energy and undergoes an intersystem crossing process leading to a triplet state T 1 , which is lower in energy (i.e., a longer-living state) than S 1 . The PS subsequently interacts with the biological environment through two kinds of photochemical reactions ( Figure 1, part B) [17]. Both pathways (type I and type II) involve oxidative stress, mainly due to the formation of reactive oxygen species (ROS), and accordingly need oxygen to be effective [18].   [14].   (µ 4 -dobq) 2 (µ 2 -L anthr ) 2 ][CF 3 SO 3 ] 4 (A 2 ), and [Ru 4 (η 6 -p-cymene) 4 (µ 4 -donq) 2 (µ 2 -L anthr ) 2 ][CF 3 SO 3 ] 4 (A 3 ), respectively. Then, the potential of the linker L anthr and the arene ruthenium metalla-rectangles to act as oxygen carriers was evaluated by various spectroscopic methods.

Results and Discussion
The synthesis of the metalla-rectangles is straightforward. First, the bis-pyridyl linker, 9,10-bis(3,3'-ethynylpyridyl) (L anthr ), is prepared according to a modified version of the previously published method [67]. It starts with a palladium-catalyzed Sonogashira reaction between 1 equivalent of 9,10-dibromoanthracene and 2.2 equivalents of 3-ethynylpyridine (63% yield). Crystals of L anthr are obtained by the slow vapor diffusion of toluene into a solution of L anthr in dichloromethane at room temperature and they have permitted to confirm the molecular structure of L anthr . The compound crystallized in the non-centrosymmetric space group P2 1 , a monoclinic crystal system. In the solid state, the anthracene unit is nearly planar with the root mean square deviation of the 12 carbon atoms of the plane being 0.024 Å. The angles between the anthracene plane and the pyridyl rings are 1.6 • and 2.9 • , respectively. As illustrated in Figure 2, the occupancy of the nitrogen atoms of the pyridyl rings is poorly defined, showing the rotating flexibility around the ethynyl axes despite the conjugated aromatic system. The formation of the metalla-rectangles A1, A2 and A3 was first confirmed by electrospray ionization mass spectrometry (ESI-MS). All spectra show the typical pattern of arene ruthenium metalla-assemblies with trifluoromethanesulfonate as counterions: a monocationic peak corresponding to [M-CF3SO3 − ] + (respectively at m/z = 2323.2 and m/z = 2423.0; not observed for A3) and a dicationic peak for [M-2(CF3SO3 − )] 2+ (respectively at m/z = 1088.9, m/z = 1139.2 and m/z = 1188.1). They also present a peak which corresponds to [M-2Lanthr-2(CF3SO3 − )] + at respectively m/z = 708.0, m/z = 759.3 and m/z = 809.0. The experimental results perfectly correlate with the calculated isotopic distributions of the different species.
The 1 H NMR spectra of A1, A2 and A3 in acetone-d6 present a chemical shift of the Lanthr protons, as compared to the non-coordinated material ( Figure 3). The pyridyl protons show a mixed behavior; those adjacent to the Ru-N bond are shifted upfield, while the other two are shifted downfield. However, the protons of the anthracene units (marked by an orange triangle and a red dot) are strongly upfield shifted in all metalla-rectangles because of the shielding effect of the anthraceneanthracene parallel arrangement. In view of trapping singlet oxygen, these protons will be the most affected by the formation of the endoperoxide derivative. Then, the dinuclear arene ruthenium clips were prepared according to published methods: [Ru 2 (η 6 -p-cymene) 2 (µ 4 -oxa)Cl 2 ] [68], [Ru 2 (η 6 -p-cymene) 2 (µ 4 -dobq)Cl 2 ] [69] or [Ru 2 (η 6 -p-cymene) 2 (µ 4 -donq)Cl 2 ] [70]. These dinuclear complexes react with two equivalents of silver trifluoromethanesulfonate in dichloromethane to afford a reactive intermediate (not isolated), and after filtration of AgCl, 1 equivalent of L anthr is added to obtain the corresponding metalla-rectangles (Scheme 1). The resulting cationic p-cymene metalla-assemblies are isolated as their trifluoromethanesulfonate (CF 3 SO 3 − ) salts in good yields (between 68% and 82%). The solubility of the metalla-rectangles depends on the nature of the dinuclear clip: A 1 is soluble in ethanol but not in dichloromethane, while A 2 and A 3 show the opposite, being soluble in dichloromethane but not in ethanol. Moreover, only A 1 is slightly soluble in water, but they are all soluble in acetone. These metalla-rectangles were fully characterized by various spectroscopic techniques (see the Supplementary Materials). compound crystallized in the non-centrosymmetric space group P21, a monoclinic crystal system. In the solid state, the anthracene unit is nearly planar with the root mean square deviation of the 12 carbon atoms of the plane being 0.024 Å. The angles between the anthracene plane and the pyridyl rings are 1.6° and 2.9°, respectively. As illustrated in Figure 2, the occupancy of the nitrogen atoms of the pyridyl rings is poorly defined, showing the rotating flexibility around the ethynyl axes despite the conjugated aromatic system. Then, the dinuclear arene ruthenium clips were prepared according to published methods: [Ru2(η 6 -p-cymene)2(μ4-oxa)Cl2] [68], [Ru2(η 6 -p-cymene)2(μ4-dobq)Cl2] [69] or [Ru2(η 6 -p-cymene)2(μ4donq)Cl2] [70]. These dinuclear complexes react with two equivalents of silver trifluoromethanesulfonate in dichloromethane to afford a reactive intermediate (not isolated), and after filtration of AgCl, 1 equivalent of Lanthr is added to obtain the corresponding metalla-rectangles (Scheme 1). The resulting cationic p-cymene metalla-assemblies are isolated as their trifluoromethanesulfonate (CF3SO3 − ) salts in good yields (between 68% and 82%). The solubility of the metalla-rectangles depends on the nature of the dinuclear clip: A1 is soluble in ethanol but not in dichloromethane, while A2 and A3 show the opposite, being soluble in dichloromethane but not in ethanol. Moreover, only A1 is slightly soluble in water, but they are all soluble in acetone. These metalla-rectangles were fully characterized by various spectroscopic techniques (see the Supplementary Materials). The formation of the metalla-rectangles A 1 , A 2 and A 3 was first confirmed by electrospray ionization mass spectrometry (ESI-MS). All spectra show the typical pattern of arene ruthenium metalla-assemblies with trifluoromethanesulfonate as counterions: a monocationic peak corresponding to [M-CF 3   The 1 H NMR spectra of A1, A2 and A3 in acetone-d6 present a chemical shift of the Lanthr protons, as compared to the non-coordinated material ( Figure 3). The pyridyl protons show a mixed behavior; those adjacent to the Ru-N bond are shifted upfield, while the other two are shifted downfield. However, the protons of the anthracene units (marked by an orange triangle and a red dot) are strongly upfield shifted in all metalla-rectangles because of the shielding effect of the anthraceneanthracene parallel arrangement. In view of trapping singlet oxygen, these protons will be the most affected by the formation of the endoperoxide derivative. In addition, typical peak multiplicities and chemical shifts of the arene ruthenium units are observed in the 1 H NMR spectra of A1, A2 and A3. The aryl protons of the p-cymene ligands are observed between ~6.85 and ~6.35 ppm, the isopropyl protons at ~1.40 ppm, the septuplet of the -CH from the isopropyl groups at ~3.00 ppm and the methyl groups at ~2.30 ppm (see the Experimental section). Interestingly, the spectrum of A1 suggests the formation of two isomers due to the presence  In addition, typical peak multiplicities and chemical shifts of the arene ruthenium units are observed in the 1 H NMR spectra of A 1 , A 2 and A 3 . The aryl protons of the p-cymene ligands are observed between~6.85 and~6.35 ppm, the isopropyl protons at~1.40 ppm, the septuplet of the -CH from the isopropyl groups at~3.00 ppm and the methyl groups at~2.30 ppm (see the Experimental section). Interestingly, the spectrum of A 1 suggests the formation of two isomers due to the presence of broad signals in the aromatic region and due to additional splitting of some protons of the p-cymene ligands. On the other hand, the 1 H NMR spectra of A 2 and A 3 do not show this phenomenon and their overall signals are rather well-defined as compared to those of A 1 . These elements support the presence of cis and trans isomers, in which the pyridyl units are pointing to the same or opposite sides of the metalla-rectangle: The free rotation of the pyridyl units being restricted upon formation of the metalla-rectangles. In fact, in the oxalato derivative A 1 , the short distance between the ruthenium atoms within the dinuclear clip (5.4 Å) forces the two L anthr panels to be in close proximity [71]. Therefore, the chemical environment of the protons associated with the cis and trans isomers is more different. Accordingly, π-π stacking interactions between the two L anthr units of the metalla-rectangles are stronger in A 1 than in A 2 and A 3 , where the Ru-Ru distances are estimated to be at 7.9 and 8.4 Å, respectively [71,72].
The diffusion coefficient (D) of A 1 , A 2 and A 3 was determined by Diffusion-Ordered Spectroscopy (DOSY) NMR experiments, recorded in acetone-d 6 at 23 • C. The presence of a single diffusion line suggests, for each metalla-rectangle, the formation of one discrete species (isomers being undistinguishable). Interestingly, A 1 and A 3 have a similar value, at, respectively, 6.67 × 10 −10 m 2 ·s −1 and 6.80 × 10 −10 m 2 ·s −1 , while A 2 is slightly different (6.04 × 10 −10 m 2 ·s −1 ). Generally, the bigger the assembly is, the smaller its diffusion coefficient. Herein, the size of the metalla-rectangles is increasing from A 1 to A 3 . However, the results are not consistent with their theoretical molecular size since A 2 has the smallest value of D. However, they might possess different arrangement in solution. Then, according to these D values and by applying the Stockes-Einstein equation, the hydrodynamic radius r H or the metalla-rectangles was calculated (viscosity of 0.31 mPa·s for acetone and a temperature of 298 K): the r H values are 10.5 Å for A 1 and 10.4 Å for A 3 , while for A 2 it is 11.7 Å. These values are consistent with the formation of the expected metalla-assemblies [59,60,[71][72][73][74][75][76].
Indeed, the molecular structure of A 1 was further confirmed by single-crystal X-ray structure analysis ( Figure 4). Crystals were obtained by slow diffusion of a mixture of ether-benzene (99:1) into an acetone solution of the corresponding metalla-rectangle. The salt crystallizes in the triclinic space group P-1. The asymmetric unit includes the tetra-cationic metalla-rectangle (trans isomer), four CF 3 SO 3 − anions, and four acetone molecules. The size of the rectangle defined by the four Ru-Ru edges is 5.4 × 17.7 × 5.4 × 17.9 Å, which correlates well with the r H determined by the DOSY experiment.
As mentioned before, π-π stacking interactions play an important role in the formation of these metalla-rectangles [77]. In the crystal structure of A 1 · 4 acetone, both anthracene groups are positioned in a "parallel fashion" that maximize π-π stacking interactions. The distance between the two centroids of the anthracene moieties is only 3.8 Å. However, as compared to L anthr , the pyridyl groups are far from co-planarity with the anthracene unit ( Figure 4). Two pyridyl rings are observed at an angle of approximately 7 • , while the two others are rotated by 34 • , from the idealized plane of their anthracene group. This distortion is probably imposed by the arene ruthenium oxalato clips as well as by an optimization of the anthracene-anthracene π-π interactions.  The electronic absorption spectra have been measured at room temperature in dichloromethane for Lanthr, A2 and A3 and in ethanol for A1 ( Figure 5). The intense high energy band centered at 270 nm is assigned to ligand-localized or ligand π⟶π* transition and the broad low-energy band corresponds to metal-to-ligand charge transfer (MLCT). Typical anthracene prints in the UV-visible spectra are observed around 400-500 nm. These bands are different in the three metalla-assemblies. In fact, A1 displays a quite similar pattern as Lanthr, while A2 and A3 show broader band and a bathochromic shift of the bands associated with the anthracene located between ~400-550 nm. Moreover, A2 presents a wider band, from which it is not possible to distinguish in the visible region of the spectrum the two bands of the anthracene moiety. In order to evaluate the ability of the ligand and the metalla-rectangles to form endoperoxide derivatives, several experiments were carried out in solution (NMR, UV-visible and fluorescence). Molecular structure of A 1 · 4 acetone at 50% probability level ellipsoids. Trifluoromethanesulfonate and acetone molecules omitted for clarity.
The electronic absorption spectra have been measured at room temperature in dichloromethane for L anthr , A 2 and A 3 and in ethanol for A 1 (Figure 5). The intense high energy band centered at 270 nm is assigned to ligand-localized or ligand π ¡π* transition and the broad low-energy band corresponds to metal-to-ligand charge transfer (MLCT). Typical anthracene prints in the UV-visible spectra are observed around 400-500 nm. These bands are different in the three metalla-assemblies. In fact, A 1 displays a quite similar pattern as L anthr , while A 2 and A 3 show broader band and a bathochromic shift of the bands associated with the anthracene located between~400-550 nm. Moreover, A 2 presents a wider band, from which it is not possible to distinguish in the visible region of the spectrum the two bands of the anthracene moiety.  The electronic absorption spectra have been measured at room temperature in dichloromethane for Lanthr, A2 and A3 and in ethanol for A1 ( Figure 5). The intense high energy band centered at 270 nm is assigned to ligand-localized or ligand π⟶π* transition and the broad low-energy band corresponds to metal-to-ligand charge transfer (MLCT). Typical anthracene prints in the UV-visible spectra are observed around 400-500 nm. These bands are different in the three metalla-assemblies. In fact, A1 displays a quite similar pattern as Lanthr, while A2 and A3 show broader band and a bathochromic shift of the bands associated with the anthracene located between ~400-550 nm. Moreover, A2 presents a wider band, from which it is not possible to distinguish in the visible region of the spectrum the two bands of the anthracene moiety. In order to evaluate the ability of the ligand and the metalla-rectangles to form endoperoxide derivatives, several experiments were carried out in solution (NMR, UV-visible and fluorescence).  Comparisons were made between the results obtained under an inert-atmosphere to those obtained with an O 2 -saturated atmosphere, as well as before or after light irradiation.
First, UV-visible spectroscopy was used to visualize the trapping of singlet oxygen by the metalla-rectangles. When dealing with anthracene derivatives, a visible-light excitation can be employed to trap singlet oxygen, and upon oxygen addition to the anthracene moiety the intensity of the absorption bands decreases over time [78,79]. Therefore, solutions of all compounds were prepared at a concentration of 10 −5 M (EtOH for A 1 ; CH 2 Cl 2 for A 2 and A 3 ), and kept in the dark before starting the measurements. Then, multiple spectra were recorded: under nitrogen, after 15 min of oxygen bubbling into the solution, and also after one hour of visible light irradiation (cool white light, Hg, 8 W). However, in all cases, no spectroscopic changes were observed under these conditions.
We then turn our attention to fluorescence spectroscopy ( Figure 6). Oxygen is known to be a common quencher of the fluorescence of aromatic compounds, since the formation of endoperoxide disrupts the electron delocalization of aromatic molecules [80][81][82]. Therefore, fluorescence spectroscopy was also used to study the behavior of the metalla-rectangles in the presence of O 2 . Each compound (5 × 10 -8 M concentrations; EtOH for A 1 ; CH 2 Cl 2 for A 2 and A 3 ) was irradiated at a specific wavelength, where the maximal absorbance was detected by the fluorimeter (L anthr , 271 nm; A 1 , 457 nm; A 2 , 273 nm; A 3 , 459 nm). At first sight, the spectra show different bands and intensities before and after placing the compounds under O 2 (30 min of bubbling). In an inert atmosphere, A 2 displays six bands: one at 270 nm, two between 300 and 400 nm, two between 450 and 520 nm and one at~550 nm; while with O 2 , there are seven bands, one additional band at~620 nm ( Figure 6). In the~270 nm and~550 nm regions, the intensity of the bands is decreasing with O 2 , whereas it is increasing in the other parts of the spectrum. Moreover, there is a slight hypsochromic shift of the bands located between 300 and 400 nm. Comparisons were made between the results obtained under an inert-atmosphere to those obtained with an O2-saturated atmosphere, as well as before or after light irradiation. First, UV-visible spectroscopy was used to visualize the trapping of singlet oxygen by the metalla-rectangles. When dealing with anthracene derivatives, a visible-light excitation can be employed to trap singlet oxygen, and upon oxygen addition to the anthracene moiety the intensity of the absorption bands decreases over time [78,79]. Therefore, solutions of all compounds were prepared at a concentration of 10 −5 M (EtOH for A1; CH2Cl2 for A2 and A3), and kept in the dark before starting the measurements. Then, multiple spectra were recorded: under nitrogen, after 15 min of oxygen bubbling into the solution, and also after one hour of visible light irradiation (cool white light, Hg, 8 W). However, in all cases, no spectroscopic changes were observed under these conditions.
We then turn our attention to fluorescence spectroscopy ( Figure 6). Oxygen is known to be a common quencher of the fluorescence of aromatic compounds, since the formation of endoperoxide disrupts the electron delocalization of aromatic molecules [80][81][82]. Therefore, fluorescence spectroscopy was also used to study the behavior of the metalla-rectangles in the presence of O2. Each compound (5 × 10 -8 M concentrations; EtOH for A1; CH2Cl2 for A2 and A3) was irradiated at a specific wavelength, where the maximal absorbance was detected by the fluorimeter (Lanthr, 271 nm; A1, 457 nm; A2, 273 nm; A3, 459 nm). At first sight, the spectra show different bands and intensities before and after placing the compounds under O2 (30 min of bubbling). In an inert atmosphere, A2 displays six bands: one at ~270 nm, two between 300 and 400 nm, two between 450 and 520 nm and one at ~550 nm; while with O2, there are seven bands, one additional band at ~620 nm ( Figure 6). In the ~270 nm and ~550 nm regions, the intensity of the bands is decreasing with O2, whereas it is increasing in the other parts of the spectrum. Moreover, there is a slight hypsochromic shift of the bands located between 300 and 400 nm.  In comparison with Lanthr, the behavior of A2 follows an opposite trend: when the intensity is increasing for Lanthr, it is decreasing for A2. The only different variation is the intensity of its bands: in the UV part and at the end of the visible domain, A2 has a more intense fluorescence, and a less intense one between 450 and 550-570 nm. All together, these results show no evidence for an endoperoxide formation on the metalla-rectangles A1-A3. Therefore, we focused our attention to Lanthr alone, in order to determine if the bis-pyridyl anthracene ligand interacts with O2.
A series of UV-visible spectra have been recorded (Figure 7), in which an air-opened solution of Lanthr was continuously irradiated under white light. After a week, the intensity of the bands associated with the anthracene moiety was significantly reduced, and the solution appeared to have reached an equilibrium. Two isosbestic points were observed, at ~330 nm and ~390 nm, respectively. The formation of a new compound was confirmed by 1 H NMR spectroscopy, in which a new set of signals was observed. Two new doublet of doublets, slightly downfield shifted as compared to the anthracene protons of Lanthr (8.7 and 7.7 ppm), are observed at 8.8 and 7.9 ppm, respectively: thus suggesting the formation of the endoperoxide derivative. For A 1 and A 3 (Figure 6), no band before 450 nm is observed, and all their bands are hyperchromic. Only a very small decrease in the intensity of the band at~470 nm is observed for A 3 . In comparison with L anthr , the behavior of A 2 follows an opposite trend: when the intensity is increasing for L anthr , it is decreasing for A 2. The only different variation is the intensity of its bands: in the UV part and at the end of the visible domain, A 2 has a more intense fluorescence, and a less intense one between 450 and 550-570 nm. All together, these results show no evidence for an endoperoxide formation on the metalla-rectangles A 1 -A 3 . Therefore, we focused our attention to L anthr alone, in order to determine if the bis-pyridyl anthracene ligand interacts with O 2 .
A series of UV-visible spectra have been recorded (Figure 7), in which an air-opened solution of L anthr was continuously irradiated under white light. After a week, the intensity of the bands associated with the anthracene moiety was significantly reduced, and the solution appeared to have reached an equilibrium. Two isosbestic points were observed, at~330 nm and~390 nm, respectively. The formation of a new compound was confirmed by 1 H NMR spectroscopy, in which a new set of signals was observed. Two new doublet of doublets, slightly downfield shifted as compared to the anthracene protons of L anthr (8.7 and 7.7 ppm), are observed at 8.8 and 7.9 ppm, respectively: thus suggesting the formation of the endoperoxide derivative.

9,10-bis(3,3'-ethynylpyridyl)anthracene (Lanthr):
In a Schlenk flask, a mixture of 9,10dibromoanthracene (400 mg, 1.19 mmol) and 3-ethynylpyridine (270 mg, 2.62 mmol) was dissolved in a solution of toluene:trimethylamine (1:1, 25 mL) and let under a nitrogen atmosphere during 15 min. Then, a mixture of palladium(II)acetate (5 mg, 0.024 mmol), copper(I)iodide (6 mg, 0.030 mmol) and triphenylphosphine (17 mg, 0.065 mmol) was added to this solution. The reaction mixture was stirred at reflux for 24 h. The solvent was removed under vacuum. The residue was dissolved in water and stirred for 2 h at RT, to eliminate the triethylammonium salt. The solid was filtered off and dried under vacuum. Recrystallization was done in toluene and the product was obtain as orange needles, which were dried under vacuum (285 mg, 63%). 1

9,10-bis(3,3'-ethynylpyridyl)anthracene (L anthr ):
In a Schlenk flask, a mixture of 9,10-dibromoanthracene (400 mg, 1.19 mmol) and 3-ethynylpyridine (270 mg, 2.62 mmol) was dissolved in a solution of toluene:trimethylamine (1:1, 25 mL) and let under a nitrogen atmosphere during 15 min. Then, a mixture of palladium(II)acetate (5 mg, 0.024 mmol), copper(I)iodide (6 mg, 0.030 mmol) and triphenylphosphine (17 mg, 0.065 mmol) was added to this solution. The reaction mixture was stirred at reflux for 24 h. The solvent was removed under vacuum. The residue was dissolved in water and stirred for 2 h at RT, to eliminate the triethylammonium salt. The solid was filtered off and dried under vacuum. Recrystallization was done in toluene and the product was obtain as orange needles, which were dried under vacuum (285 mg, 63%). 1 13  General procedure for the synthesis of metalla-rectangles A 1 , A 2 and A 3 : A mixture of metalla-clip (oxa: 200 mg, 0.32 mmol; dobq: 200 mg, 0.29 mmol; donq: 200 mg, 0.27 mmol) and silver trifluoromethanesulfonate (2 eq.) was dissolved in dichloromethane and was stirred for 3 h at RT. The mixture was filtrated in order to eliminate silver chloride. The resulting solution was added to a dichloromethane solution of L anthr (1 eq.). Then, the mixture was refluxed overnight and consequently concentrated under vacuum. The concentrated solution was slowly poured into cold diethyl ether to induce precipitation. After filtration, the metalla-rectangles were filtered off and dried under vacuum.
[Ru 4 (η 6 -p-cymene) 4    X-ray crystallography: Crystals were mounted on a Stoe Image Plate Diffraction system equipped with a Φ circle goniometer, using Mo Kα graphite monochromated radiation (λ = 0.71073 Å) with Φ range 0-200 • . The structures were solved by direct methods using the program SHELXS-97 [83], while the refinement and all further calculations were carried out using SHELXL-97. The H-atoms were included in calculated positions and treated as riding atoms using SHELXL-97 default parameters. The non-H atoms were refined anisotropically using weighted full-matrix least-square on F 2 . In both structures, relatively high R factors are observed due to disorder within the crystals. Crystallographic details are summarized in Table 2, and Figures 2 and 4 were drawn with ORTEP-32 [84]. CCDC-1853676 (L anthr ) and 1853677 (A 1 · 4 acetone) contain the supplementary crystallographic data for this paper. This can be obtained free of charge at www.ccdc.cam.ac.uk/conts/retrieving.html (or from the Cambridge Crystallographic Data Center, 12, Union Road, Cambridge CB2 1EZ, UK; fax: (internat.) +44-1223/336-033; E-mail: deposit@ccdc.cam.ac.uk).

Conclusions
Three metalla-rectangles containing anthracene-derived linkers have been synthesized and characterized. Upon formation of the metalla-assembly, the propensity of the anthracene moiety to react with oxygen to form endoperoxide derivatives was lost, probably due to electronic or steric constraints. Nevertheless, the introduction of anthracene groups on metalla-assemblies remains an interesting avenue to transport oxygen to cancer cells; however, in view of this study, the anthracene moiety should be anchored elsewhere to allow the formation of metalla-assemblies with endoperoxide groups.