Functionalized Ruthenium Complexes: Selective “Turn-on” Detection of Biologically Relevant Anionic Species

: To study the effect of the structure of 3,3 ′ -modified bipyridyl ruthenium complexes on their ability to recognize organic anions, various ruthenium complexes have been prepared. The binding functions and large-sized modified bipyridyl ligand turned out to be essential for selectivity in acetonitrile. The selectivity for dicarboxylates or phosphates can be switched by using guanidinium- or ammonium-functionalized probes. One of these probes turned out to be selective towards glutamate over aspartate and γ -aminobutyric acid (GABA). An-other was selective towards adenosine 5 ′ -triphosphate (ATP) in the Supporting Information). These results confirm that both amides and terminal functions are implicated in the recognition process, mainly through the formation of hydrogen bonds and electrostatic interactions with the anions.


Introduction
Anions play fundamental roles in a wide range of chemical, biological, and environmental processes. For instance, phosphates and nucleotides such as adenosine 5′-triphosphate (ATP), adenosine 5′-diphosphate (ADP), and adenosine 5′monophosphate (AMP) are involved in energy storage, ionchannel regulation, [1] and phosphorylation. [2] During phosphorylation, kinases transfer the terminal phosphate from ATP onto a protein, a sugar, or a lipid. In parallel, ADP is released. The phosphorylation of proteins, by inducing structural modifications, switches on or off their enzymatic activity and in this way regulates many essential biological processes. [3] A way to measure the activity of enzymes that catalyze metabolic processes leading to the production of ADP would be to detect ADP selectively.
The development of luminescent chemosensors for halides, carboxylates, and inorganic and organic phosphates, such as nucleotides, has received substantial attention in the last few decades. [4] Many examples of fluorescent probes with metallic receptors, mainly zinc, [5] copper, [6] and more recently aluminium, [7] have been successfully used to detect selectively pyro- [ over adenosine 5′-diphosphate (ADP), pyrophosphate (PPi), adenosine 5′-monophosphate (AMP), and orthophosphate (Pi). In both cases, the binding was attributed to coulombic interactions and hydrogen bonding. π-stacking interactions also occurred with nucleotides. Replacing ammonium by zinc-dipicolylamine units made possible the recognition of phosphorylated species in buffered aqueous systems. To our delight, this probe showed selectivity for ADP over ATP and we proved that the selectivity was partly due to substitution at the 3-and 3′-positions of the ligand.
phosphate (PPi). Luminescent probes able to recognize preferentially ATP over PPi and other adenosine phosphates have also been abundantly described. [8] In general, the observed selectivity is ascribed to stronger electrostatic interactions between the probe and the triphosphate. Conversely, examples of selective chemical probes for AMP or ADP are scarce. [9] In the case of ADP, binding experiments with these sensors were conducted either in organic solvent [9g,9k] or in hydroorganic medium. [9b-9d] The bindings of ADP and PPI were otherwise compared in only one case. [9d] Thus, the elaboration of new probes able to interact preferentially with organic anions, in particular with ADP, is still a challenge. Ideally, the luminescence intensity of such probes should increase upon the binding of guests.
Having this in mind, we developed a straightforward procedure for the synthesis of flexible probes for anion recognition (Figure 1). We explored the structure-binding relationship in acetonitrile to determine both the interactions involved and the parameters that influence the binding and selectivity. Then, we introduced the zinc dipicolylamine group for the recognition of phosphorylated species in aqueous systems.
The chosen luminophore was a tris-bipyridyl ruthenium complex, a luminescent compound with absorption and emission spectra in the visible region. [4c,10] Many bipyridyl ruthenium(II) luminescent probes for the detection of small inorganic anions have been described. [4d-4f ] Usually, the recognition is achieved by the functionalization of one of the bipyridyl ligands, most of the time at the 4,4′-positions. [4e,f,11] In this work, one of the bipyridines was modified at the 3,3′-positions through amide subunits. Indeed, in a preliminary study we saw that even if the quantum yield of the 3,3′-bipyridyl-modified ruthenium(II) complex is lower than that of its 4,4′-analogue, the increase in luminescence, observed upon the addition of different species, is often greater. [12] Three different binding functions were selected: Guanidinium and ammonium groups, both well-known to interact strongly with oxo anions through hydrogen bonds and charge interactions, [4b,4f,13] and a zinc(II) dipicolylamine complex, able to bind anions in aqueous medium. Various chain lengths were introduced between the amides and terminal binding sites to study the impact of the size of the molecular probe on both the recognition and selectivity. Finally, an analogous compound without a terminal binding site was also prepared to determine the role of amides in the recognition process.

Probe Design and Synthesis
To access the ruthenium probes we developed a rapid and easy synthesis starting from binicotinic acid, prepared on a gram scale from 1,10-phenanthroline. [14] The reaction of binicotinic acid with 2-(n-aminoalkyl)-1,3-diBoc-guanidines in the presence of EDCI and DMAP, followed by acidic deprotection of the guanidinium functions (n = 2, 4, and 6) led to guanidinium-functionalized ligands L 1 /L 3 in yields varying from 57 to 71 % (Scheme 1). Other ligands were obtained by using the same strategy; L 4 and L 5 , from commercially available (aminoalkyl)carbamates, in yields of 48 and 84 %, respectively, and L 6 , from hexylamine, in 77 % yield. Heating 1 equivalent of the appropriate ligand L i at reflux with cis-[(bipy) 2 RuCl 2 ]·2H 2 O led to the ruthenium(II) complexes. After ion exchange, Ru 1 -Ru 6 were recovered in good yields (52-79 %) as their hexafluorophosphate salts. They were obtained pure, except Ru 4 , recovered with 90 % purity. In fact, the high solubility of Ru 4 in water prevented us from removing the excess of ammonium hexafluorophosphate used for the anion exchange. The amount of remaining NH 4 PF 6 was estimated by 1 H and 19 F NMR experiments using (R)-(-)-1-(9-anthryl)-2,2,2-trifluoroethanol as internal reference. The elemental analysis of ruthenium complexes Ru 1 -Ru 5 showed the presence of four hexafluorophosphate anions, proving the protonation of both guanidines and amines in the solid state. Access to zinc(II) dipicolylamine analogue Ru 7 was achieved by classical reductive amination of Ru 5 in the presence of 2-pyridinecarbaldehyde, followed by the addition of 2 equivalents of zinc dichloride. The water-soluble probe Ru 7 was obtained pure in 94 % yield.

Photoluminescence Characterization of the Ruthenium Complexes
As shown in Table 1, in acetonitrile, all the synthesized trisbipyridyl ruthenium complexes show two characteristic absorption bands at about 287 and 440 nm corresponding to ligandcenter (LC) and metal-to-ligand charge-transfer (MLCT) transitions, respectively. Owing to the electron-withdrawing nature  [15] [h] From ref. [19] of the amide substituents, which leads to lower MLCT transition energies, the MLCT absorption band is blueshifted with respect to [Ru(bipy) 3 ] 2+ (λ max = 450 nm). [15] The Ru i complexes display emission spectra with a single red-orange MLCT emission band at around 710 nm, 100 nm redshifted with respect to the band of [Ru(bipy) 3 ](PF 6 ) 2 (λ max = 612 nm). In Hepes buffered solution, this shift is even more important (for Ru 7 , 734 nm in aqueous medium). Such a phenomenon has already been observed in ruthenium complexes functionalized with amide or ester groups. [14][15][16] The quantum yields of all the 3,3′-functionalized complexes (1.9-2.4 %) are much lower than that of the [Ru(bipy) 3 ] 2+ standard (Φ = 6.2 %). This can be explained by the ligand distortion in the 3,3′-modified ruthenium complexes that disfavors charge transfer between the metal center and the ligands. It is noteworthy that the 3,3′-modified ruthenium complexes exhibit a biexponential decay of the luminescence intensity. A short decay time (29-93 ns) and a longer one (140-208 ns) were determined. It is clear by centesimal analysis that these biexponential decay times do not derive from the presence of a luminescent impurity in the synthesized ruthenium complexes. According to the literature, biexponential decay times have already been found in the case of tris-bipyridyl ruthenium complexes. [17] The authors reported that the observed biexponential decay may arise from the existence of two blocked conformers. Indeed, the relative rigidity of the amide moieties may prevent the conformers from interchanging within the lifetime of the excited state. In the present case, cis/trans amide isomers coexist in solution, as proved by the 1 H NMR spectra of these ruthenium complexes (the presence of two multiplets corresponding to the methylene at the α position with respect to the amide functions). The nonradiative decay rates increase with the linker length and, as expected, they are especially low for the most rigid probes, in other words, those with the shortest linker (Ru 1 and Ru 4 ). [18] 3

General Behavior of Probes Upon Addition of Anions
The binding of anions of interest, such as dicarboxylates and phosphate derivatives, by these probes was studied by UV/Vis and luminescence spectroscopy. The complexation of more simple anions, such as chloride or monocarboxylate anions, was also tested ( Figure 2). The addition of the anions to solutions of Ru i in acetonitrile did not induce any change in the absorption spectra. In the presence of chloride or acetate anions, the Ru i emission spectra were not or only very slightly affected. In contrast, the addition of other species to solutions of Ru i probes resulted in an increase in the luminescence intensity accompanied by a concomitant hypsochromic shift of the maximum emission wavelength λ max .
This increase in luminescence intensity is consistent with the enhancement of the quantum yields of the probes measured in the presence of ions such as glutamate or dihydrogen phosphate (see Table A in the Supporting Information). As expected, regarding the non-modified [Ru(bipy) 3 ] 2+ complex, no changes in the emission spectra were observed upon the addition of anions. The evolution of the Ru 5 emission spectrum upon successive additions of dihydrogen phosphate is shown as an example in Figure 3. The emission enhancement can be ascribed to an increase in receptor rigidity, caused by interactions with the anions, which diminishes the potential pathways of vibrational and rotational nonradiative decay modes. To determine the influence of anions on the decay rates and decay times of the probes, we also measured them in the presence of glutamate and dihydrogen phosphate. The nonradiative decay rates decreased (except for the most rigid probe Ru 1 ), whereas the radiative decay rates showed a two-to three-fold increase in the presence of anions.  Concerning the decay times, as for the complexes Ru i alone, a biexponential decay was found when the probes interact either with glutamate or dihydrogen phosphate (see Table A in the Supporting Information). This observation suggests that the anions interact with the amide groups. Indeed, the MLCT dπ* state has the electron formally residing on the bipyridine ligand, and this configuration is presumably less favorable when a negatively charged guest binds to the amide of the bipyridine. [16,17,20] To verify the existence of interactions between amide subunits and anions, luminescence studies were carried out with the analogous probe Ru 6 , characterized by a 3,3′-bishexylamide modified ligand.
As for Ru 3 and Ru 5 , a hypsochromic shift and an increase in the luminescence intensity were observed when H 2 PO 4 was added to a solution of Ru 6 in acetonitrile. Nevertheless the modification of the luminescence response is not as significant (see Figure S1 in the Supporting Information). These results show that the amide functions are implicated in the recognition event. To better identify the interactions involved we also performed 1 H NMR studies in deuteriated acetonitrile. The spectra of probes Ru 3 and Ru 5 in the presence of increasing amounts of either glutamate or dihydrogen phosphate were recorded. The 1 H NMR spectra of Ru 3 upon sequential additions of glutamate are shown as an example in Figure 4 (aromatic region) and in Figure S2 in the Supporting Information (whole 1 H NMR spectra). Addition of the anion leads to downfield shifts of the amide and guanidinium NH protons. More interestingly, the 4-H and 4′-H aromatic protons of the modified bipyridyl ligand L 3 as well as methylene protons linked to the amide functions are also shifted as the amount of glutamate in the solution increases. Similar results were obtained when dihydrogen phosphate was added to a solution of probe Ru 5 . Nevertheless, in this case, the signals corresponding to the methylene groups were not affected by the presence of the guest (see Figure S3 in the Supporting Information). These results confirm that both amides and terminal functions are implicated in the recognition process, mainly through the formation of hydrogen bonds and electrostatic interactions with the anions.

Alkyl Spacer Length Effect and Selectivity in the Case of Guanidinium-Functionalized Bipyridyl Ruthenium Complexes
Acetate and chloride anions did not induce any change in Ru 3 photoluminescence. The complex Ru 3 recognizes dihydrogen phosphate and dicarboxylates such as glutamate and aspartate, but also γ-aminobutyric acid (GABA; Figure 5a). The titration curve recorded with GABA suggests that the amine function and/or the alkyl chain of the target interact with the binding site of Ru 3 , because acetate is not recognized at all. Titrations with sulfate, succinate, and glutarate were also carried out (see Figure S4 in the Supporting Information). Among all the tested anions, glutamate is the one that induces the most significant changes in luminescence emission intensity ( Figure 5a). For comparison with Ru 3 , luminescence studies were also carried out with the Ru 1 and Ru 2 complexes to study the effect of the alkyl chain length and thus the effect of host size on anion recognition. Ru 1 and Ru 2 gave similar results. As an example, Figure 5b shows the titration curves of Ru 1 upon the addition of various anions: Similar responses were observed for glutamate, aspartate, and dihydrogen phosphate. Moreover, Ru 1 detects very slightly acetate and chloride, which was not 5 observed with Ru 3 . As a consequence, of the complexes studied herein, the use of the hexyl spacer provides the most efficient probe in terms of selectivity. Hence, only the results obtained with probes incorporating a hexyl spacer will be presented below.
The method of continuous variation (Job's plots) was used to characterize the Ru 3 -anion complexes and determine their stoichiometries (see Figure S8a,b in the Supporting Information). The curves indicate that Ru 3 forms complexes with 1:2 and 1:4 stoichiometries in the presence of glutamate and dihydrogen phosphate, respectively. The association constants of Ru 3 with glutamate (log K 1:2 = 13.06) and dihydrogen phosphate (log K 1:4 = 18.02) were determined by using HypSpec®. [21] These stoichiometries are consistent with both the 4+ charge of the probes and the number of hexafluorophosphate anions that can be exchanged during the addition of guests. Most probably, only the replacement of the PF 6 ions close to the binding functions impacts the variation in luminescence intensity. As can be seen in Figure 5a, the first added carboxylate anions did not induce noticeable variation in the luminescence intensity, which indicates that the first PF 6 anions displaced are those interacting with the ruthenium center.

Effect of the Terminal Binding Functions
Guanidinium functions have planar structures more appropriate for interacting with carboxylates than tetrahedral ammonium ions. [4f,13,16] Thus, our idea was to modify the selectivity of such probes by replacing the terminal guanidinium moieties by ammoniums (probes Ru 4 and Ru 5 ).
As already mentioned, these ammonium-functionalized complexes present the same photophysical characteristics as the guanidinium-functionalized analogues (emission spectra with a single red-orange MLCT emission band at ca. λ max = 712 nm, 100 nm redshifted with respect to the band for [Ru(bipy) 3 ]PF 6 and also an attenuated luminescence efficiency). Chloride and acetate anions were not detected by the ammonium-functionalized probe Ru 5 , whereas responses were obtained in the presence of dihydrogen phosphate and dicarboxylates ( Figure 6). A 1:2 stoichiometry was found for the Ru 5 -glutamate complex (see Figure S8c in the Supporting Information) as well as a lower binding constant (log K 1:2 = 11 59), which indicates that glutamate interacts less strongly with ammonium functions than with guanidinium. Ru 5 recognizes dihydrogen phosphate in the same way as Ru 3 (1:4 stoichiometry and log K 1:4 = 18.08; see Figure S8d in the Supporting Information). Nevertheless, for equivalent stability constants, the quantum yield of Ru 5 in the presence of H 2 PO 4 -(Φ = 9.3) is higher than that measured in the case of Ru 3 (Φ = 7.8; see Table A in the Supporting Information). As a consequence, the ammonium-functionalized probe is more sensitive towards phosphate ions than the guanidinium. This sensitivity encouraged us to study the binding ability of Ru 5 with phosphate anions of biological interest, namely ATP, ADP, and AMP.

Application to the Detection of Important Anions
The luminescence intensity of Ru 5 was recorded upon the addition of adenosine tri-, di-, and monophosphate, PPi, and dihydrogen phosphate anions (Figure 7). Although AMP and H 2 PO 4 have the same charge, their titration curves are different. This difference can be ascribed to possible π-stacking interactions between the adenine base and the Ru 5 bipyridine ligands. To verify this hypothesis, titration curves with AMP and H 2 PO 4 were recorded with [Ru(bipy) 3 ] 2+ as the probe. In the presence of dihydrogen phosphate, the luminescence of [Ru(bipy) 3 ] 2+ remained unchanged (no possible π-stacking associations), whereas it increased 1.6-fold upon the addition of AMP (see Figure S5 in the Supporting Information). These results are consistent with π-stacking interactions contributing to the recognition of adenosine phosphate derivatives. Nevertheless, compared with ionic interactions, π-stacking interactions are weak (small increase in luminescence intensity), which explains why there is no difference in the recognition of ADP and PPi (Figure 7). The stoichiometries of the complexes as well as the binding constants were determined for the adenosine phosphates (Table 2) 6 whereas a 1:1 complex is formed with ATP (see Figure S8e-g in the Supporting Information). The binding constants can be used to compare the affinity of a probe towards various species provided the complexes formed have the same stoichiometry. As this is not the case here, we chose to compare the slopes of the titration curves. Indeed, the greater the slope of the curve, the higher the affinity of the probe. According to the curves recorded here, the strongest affinity of the probe Ru 5 is towards ATP, with ADP and PPi showing similar affinities, and the weakest interaction being observed with AMP.
This was confirmed by the binding constants measured for complexes of identical stoichiometry, that is, AMP (log K 1:2 = 7.24) and ADP (log K 1:2 = 9.36). Thus, the affinity of Ru 5 for phosphate derivatives increases with the charge of the guest, in agreement with the literature. [4b] Nevertheless, the ionic interactions of ammonium phosphate are too weak in water and Ru 5 is not suitable for binding such species in aqueous media.
These results prompted us to introduce stronger complexing groups in water. Thus, we turned our attention to the zinc(II) dipicolylamine probe Ru 7 . Preliminary studies conducted in acetonitrile showed that the binding of AMP, H 2 PO 4 -, and glutamate by Ru 7 led to an increase in luminescence intensity. Just as with Ru 5 , selectivity for the nucleotide was observed (see Figure S6 in the Supporting Information). Next, binding studies in pure 25 mM Hepes buffered solution (pH 7.4) were conducted (Figure 8). Under such conditions, chloride, acetate, glutamate, and AMP did not induce any change in the luminescence intensity of the probe, whereas the addition of ADP, PPi, or ATP resulted in a significant increase in luminescence intensity as well as a 10 nm hypsochromic shift of the emission wavelength. The response of the probe upon addition of 20 μM of anions is shown in Figure 8. The variation in luminescence intensity upon addition of ADP is shown in the inset. Contrary to what we observed in acetonitrile with Ru 5 , selectivity for ADP over PPi was observed with Ru 5 in water due to π-stacking interactions. The selectivity for polyphosphates over AMP and H 2 PO 4 is clearly a result of their different charges, but ionic interactions are not the only interaction involved in the binding, because selectivity of ADP over ATP is observed. The binding constants confirm this selectivity ( Table 3). The higher response for ADP over ATP is ascribed to the positions of functionalization. Indeed, the 4,4′-modified probe Ru 8 , an analogue of the 3,3′-functionalized probe Ru 7 , has also been synthesized (see the Supporting Information). As for Ru 7 , the 4,4′-modified probe detects phosphorylated species in pure Hepes buffered solution. However it is not able to discriminate between ATP and ADP (see Figure S7 in the Supporting Information).

Conclusions
Convinced by the potential of 3,3′-bipyridyl ruthenium complexes, we have developed a general and rapid synthesis that led to a series of novel ruthenium luminescent probes with either guanidinium, ammonium, or zinc(II) dipicolylamine binding sites. These complexes were subjected to anion complexation studies. Upon the addition of anions, the luminescence of the probes increased and a hypsochromic shift was observed. As expected, guanidinium-functionalized probes proved to be more efficient in the recognition of carboxylates. It is noteworthy that the probe Ru 3 detects glutamate preferentially to phosphates or other neurotransmitters (aspartate and GABA). On the other hand, the analogous ammonium probe Ru 5 preferentially interacts with phosphate derivatives and more particularly with nucleotides. Moreover, as Ru 5 is able to discriminate phosphate derivatives as a function of their respective charge, selectivity for ATP was observed within the family of adenosine phosphate derivatives. Nevertheless, these probes are not able to bind such species in aqueous media, clearly because of the solvation of anions. This difficulty has been easily circumvented through metal-anion interactions. In Hepes-buffered solution, Ru 7 binds di-and triphosphates and is able to detect ADP selectively. This study shows that the binding of anions occurs through electrostatic and π-stacking interactions and that functionalization at the 3,3′-positions plays a key role in the observed selectivity.

Experimental Section
General Information and Materials: UV/Vis spectra were recorded with a Varian Cary 100 Scan spectrophotometer. Photolumines-7 cence spectra were collected with a Varian Eclipse fluorescence spectrophotometer. The excitation wavelength was 490 nm and the emission spectra were recorded between 500 and 850 nm. All the emission spectra were corrected. Emission quantum yields were determined by using [Ru(bipy) 3 ] 2+ as standard, which has a known emission quantum yield of 0.062 at 25°C in acetonitrile. [19] Titration experiments were carried out at 25°C with either 5 × 10 -6 M or 5 × 10 -5 M solutions of Ru i in distilled and degassed acetonitrile and increasing anion concentration. Measurements were repeated to verify their reproducibility. Photoluminescence emission (A/A 0 ), based on the titration data, represents the PL emission area of the probe in the presence of the guest (A) normalized to the initial PL emission area (A 0 ) in the absence of the anion.
Time-resolved fluorescence measurements were performed on dilute solutions (ca. 10 -6 M, optical density 0.1) in standard 1 cm quartz cuvettes using an Edinburgh Instruments (FLS920) spectrofluorimeter in photon-counting mode. Fluorescence lifetimes were measured by time-correlated single-photon counting (TCSPC) using the same FLS920 spectrofluorimeter. Excitation was achieved by a hydrogen-filled nanosecond flash lamp (repetition rate 40 kHz). The instrument response (FWHM ca. 1 ns) was determined by measuring the light scattered by a Ludox suspension.