Modulating photostability and mitochondria selectivity in far-red/NIR emitting coumarin fluorophores through replacement of pyridinium by pyrimidinium.

Mitochondrial dysfunction has been associated with a wide range of human pathologies, including cancer, aging and neurodegenerative diseases. Thus, the availability of selective fluorescent probes for mitochondria could play a key role in the future for monitoring cellular function and disease progression. In this work, we have investigated how the photophysical properties and subcellular accumulation of nonconventional coumarin-based COUPY fluorophores can be fine-tuned through replacement of the para-pyridinium moiety with several heterocycles. Among them, ortho,para-pyrimidinium substitution provided novel fluorophores with suitable photophysical properties for bioimaging applications, including far-red to NIR emission, large Stokes's shifts and high photostability. Furthermore, the compounds exhibited excellent cell permeability in living HeLa cells and a higher preference for mitochondria compared with the parent COUPY dyes. Overall, these results provide valuable insights for the design and optimization of mitochondria-targeted fluorescent probes based on low molecular weight scaffolds since higher selectivity for this organelle can be achieved through the replacement of conventional N-alkylated pyridinium moieties by the corresponding N-alkylated-ortho,para-pyrimidinium counterparts.


INTRODUCTION
Fluorescence imaging in combination with organic fluorophores has become a powerful tool for understanding of biological events at a molecular level. In this context, the use of fluorescent probes with operability in the less energetic far-red and near-infrared (NIR) region of the electromagnetic spectrum offers several appealing features for in vivo imaging applications, such as increased tissue penetration depth and minimal autofluorescence interference from natively occurring biomolecules. 1 For this reason, recent efforts have been devoted to the development of novel organic chromophores operating in the phototherapeutic window with optimal photophysical (e.g., long-wavelength absorption and emission, brightness, large Stokes' shifts and photostability) and physicochemical (e.g., aqueous solubility, good cell permeability and target specificity) properties. Ideally, such fluorescent compounds should be based on chemically stable, low molecular weight scaffolds amenable to smart and simple structural modifications to fine-tune the abovementioned properties as well as to facilitate conjugation to targeting ligands. 2 Besides biocompatibility, an ideal fluorophore should permit the interrogation of intracellular architectures and dynamics without disturbing and compromising the integrity of the cellular target. 3 Mitochondria are involved in many key cellular processes, including ATP synthesis, calcium signaling and redox homeostasis. 4 Mitochondrial dysfunction has been associated with a wide range of human pathologies, such as cancer disease, aging and metabolic and neurodegenerative diseases. 5 Thus, the availability of fluorescent dyes that can stain selectively mitochondria opens the door to monitoring cellular function and disease progression by studying mitochondrial morphology and mitophagy. 6 Among fluorescent mitochondrial probes described to date, some of them intrinsically target this organelle such as some cyanine derivatives (e.g., IR-780 and MHI-148) and some rhodamines (Rhodamine 123). 7 Another approach to confer mitochondria selectivity consists of incorporating lipophilic positively charged moieties such as triphenylphosphonium or pyridinium groups, which exploit the negative potential across outer and inner mitochondria membrane. 8 In this context, anticancer mitochondria-targeted fluorescent molecules are considered attractive theranostic agents. 9 Similarly, mitochondria-targeted photocages based on organic chromophores provide a powerful method for releasing bioactive compounds within this organelle. 10 In our group, we have recently described a new class of coumarin-based chromophores in which the carbonyl group of the electron-withdrawing lactone in conventional coumarin 1
It is worth noting that 1 H-1 H NOESY experiments (Figures 1 and S2-S5) account for the existence of two species in equilibrium in solution because of the rotation around the exocyclic C=C bond, which reproduces the behavior previously found in the parent COUPY scaffolds (compounds 2 and 3, Scheme 1). 11 As shown in Figure 1, chemical exchange crosspeaks between the resonances of E and Z rotamers were observed in the 2D NOESY spectrum of coumarin 8. In the case of coumarins 6, 8 and 9, the two rotamers were present in a ∼60:40 7 ratio according to the integration of the 1 H NMR spectrum, being the E rotamer the major species. This is in contrast with para-pyridine-containing coumarins 2 and 3 in which the E/Z ratio was ∼90:10. Surprinsingly, the Z rotamer was the major species in solution (95%) in the COUPY scaffold containing the ortho,ortho pyrimidine heterocycle (7), which was confirmed by the existence of a NOE cross-peak between the proton at position 3 of the coumarin moiety and the proton at the ortho position of the pyrimidine ring ( Figure S3). It is worth noting that in all of the COUPY scaffolds the chemical shift of H3 appears at higher δ in the Z isomer (e.g., 8.18 ppm in 7 and 8.26 ppm in 8) than in the E isomer (e.g., 6.70 ppm in 7 and 6.75 ppm in 8). Next, we synthesized the corresponding N-methylated pyridinium (10) and pyrimidinium (11)(12)(13) dyes by reaction with methyl trifluoromethanesulfonate in DCM at room temperature. 11 The four new coumarin derivatives were isolated after silica column chromatography as reddish-orange (10-11), purple (12) and dark blue (13) solids (yields 64-80 %), and their purity was assessed by reversed-phase HPLC ( Figure S1). Characterization was carried out by HR ESI-MS and 1D and 2D NMR spectroscopy. As expected, methylation occurred at the less sterically-hindered nitrogen in the ortho,para-pyrimidine derivatives (12 and 13) according to NOESY NMR characterization, and in all cases the E rotamer was the major species in solution ( Figures S6-S9).

Photophysical characterization of COUPY fluorophores.
Having at hand COUPY dyes 10-13, we investigated the effect of replacing the para-pyridine moiety in the parent fluorophores (4-5) by ortho-pyridine or by ortho,ortho-or ortho,parapyrimidine on the spectroscopic and photophysical properties of the compounds with the aim of establishing new SPPR. Although water or phosphate-buffered saline (PBS) are often used to evaluate the usefulness of a new fluorophore within a spectroscopy cuvette, the heterogeneity and complexity of the cellular environment cannot be accurately described by using a simple aqueous buffer. 16 For this reason, we decided to register the UV-Vis absorption and emission spectra in four solvents of different polarity (PBS buffer pH 7.4, EtOH, ACN and DCM; Figures 2 and S10-S17) to get some insights on the photophysical behaviour of the compounds in polar and less-polar environments. As shown in Table 1, the photophysical properties of coumarins 10-13 were compared with those of the parent compounds 4-5 11 to facilitate the establishment of SPPR.
All the new compounds exhibited an intense absorption band in the visible region of the electromagnetic spectrum, with absorption maxima ranging from 502 nm (10) to 600 nm (13) in aqueous solution (PBS pH 7.4), and from 543 nm (10) to 624 nm (13) in DCM. To our surprise, the absorption maximum of coumarin 10 was significantly blue-shifted with respect the reference compound 4 (e.g., compare abs = 515 and 543 nm for 10 and abs = 552 and 569 nm for 4 in EtOH and DCM, respectively). Although not as pronounced as in the case of 10, a similar trend was found in ortho,ortho-pyrimidine-containing coumarin (11) since its absorption maximum was also blue-shifted (18-24 nm depending on the solvent) with respect 4. By contrast, a slight red-shift was found in the absorption maximum of the ortho,parapyrimidine-containing coumarin 12 (12 nm in PBS and 6 nm in DCM). This red-shift was considerably larger in the 4-CF3 analogue (e.g., compare abs = 600 nm for 13 and abs = 568 nm for 5 in PBS). As previously found with the first generation of COUPY dyes, 11,12 10-13 showed negative solvatochromism since the absorption maxima was blue-shifted with increasing polarity of the solvent (e.g., compare the abs of 10-13 in DCM with the corresponding values in non-protic (ACN) and protic (EtOH) polar solvents). Moreover, the molar absorption coefficients of the two derivatives containing the ortho,para-pyrimidine moiety were similar even higher than to those of their respective parent dyes, especially in the 4-CF3 analogue (e.g., compare  = 58 mM -1 cm -1 for 13 and  = 20 mM -1 cm -1 for 5 in EtOH).
The emission maxima of coumarins 10 and 11 were also blue-shifted both in polar and lesspolar solvents with respect coumarin 4 (e.g., em = 564 nm for 11 and em = 603 nm for 4 in EtOH), which reproduces the effect of replacing para-pyridine by ortho-pyridine or ortho,ortho-pyrimidine moieties on the compounds' absorption maxima. By contrast, the emission maxima of COUPY dyes containing the ortho,para-pyrimidine moiety was redshifted with respect their parent compounds (11-24 nm in 12 and 6-19 nm in 13), especially in polar solvents. As a consequence, COUPY dyes 12 and 13 showed emission in the far-red to NIR region (Figures 2 and S10-S17), being the emission maximum in polar media particularly appealing in the case of the 4-CF3 fluorophore (em = 673 nm for 13 in PBS). Interestingly, coumarin 12 exhibited larger Stokes' shifts than its parent dye 4 in all the solvents evaluated while the replacement of para-pyridine by ortho,para-pyrimidine in 5 led to slightly smaller values. Nevertheless, the Stokes' shifts of 12 and 13 in polar solvents are sufficiently large (e.g., 71 and 73 nm in PBS, respectively) to avoid the light reabsorption problems typically found in bioimaging applications. 1 As shown in Table 1, compounds 10 and 11 exhibited very weak fluorescence in all the solvents investigated. By contrast, fluorescence quantum yields for coumarin 13 were much higher than those of the parent compound 5, especially in less-polar solvents (e.g., 0.41 and 0.05 in DCM, respectively). This tendency was reversed in the case of the 4-CH3 analogue since the ΦF for 12 was smaller compared with 4 (e.g., 0.11 and 0.70 in DCM, respectively).
Both ortho,para-pyrimidine-containing compounds exhibited moderate fluorescence quantum yields in polar protic solvents (ΦF = 0.13 for 12 and ΦF = 0.08 for 13 in EtOH).  Finally, the photostability of the most promising coumarins (12 and 13) was investigated in PBS buffer (pH 7.4) under green LED irradiation (Figures 3 and S18). As shown in Figure 3, the replacement of pyridine in coumarin 4 with ortho,para-pyrimidine had a positive effect on the photostability of the resulting fluorophore (12). By contrast, coumarin 13 was found less photostable than its parent compound 5 and coumarin 12, which indicates that replacement of the CH3 group at the 4-position with CF3 in ortho,para-pyrimidine-containing COUPY dyes does not lead to an improvement of the overall photostability of the compounds. Nevertheless, it is worth noting that the two new pyrimidine-containing COUPY dyes were found photostable up to light fluences larger than 1000 J/cm 2 (12) and 200 J/cm 2 (13), which are more than 50-and 10-fold, respectively, higher than those typically used in imaging experiments with living cells. In summary, all these observations allowed us to stablish some structure-photophysical property relationships. First, replacement of para-pyridine in COUPY dyes with orthopyridine or ortho,ortho-pyrimidine moieties had a negative effect on the spectroscopic properties of the fluorophores since both absorption and emission maxima were blue-shifted.
Moreover, compounds 10 and 11 were found weakly fluorescent and their extinction coefficients were smaller than those of the parent coumarin 4. By contrast, the photophysical properties of COUPY dyes were clearly improved when para-pyridine was replaced with ortho,para-pyrimidine. On the one hand, both absorption and emission maxima were redshifted with respect the parent compounds, especially in polar protic solvents. Indeed, the incorporation of the second nitrogen atom in the pyridine moiety in 5 caused a red-shift of the absorption (32 nm) and emission (12 nm

Fluorescence imaging of COUPY dyes in living cells
Taking into account the large photostability and the photophysical properties of the two ortho,para-pyrimidine-containing COUPY dyes (12 and 13), we decided to evaluate their usefulness as fluorescent probes in a more realistic situation (e.g., in living cells). As previously stated, the heterogeneous environment of an organic fluorophore within a cell or within a specific cellular organelle might be considerably different than the homogeneity of a solution in a spectroscopy cuvette. In fact, the presence of biomolecules such as proteins and lipids or the interaction with the components of cellular membranes might provide the fluorescent probe with an environment less hydrophilic than expected, thereby modifying key parameters for bioimaging applications such as brightness.
The cellular uptake of 12 and 13 was first investigated in living HeLa cells (2 M, 30 min incubation) by using confocal microscopy and compared with that of the para-pyridinecontaining coumarins (compounds 4 and 5, respectively). Irradiation was carried out with a yellow light laser (λex = 561 nm) in the case of 4-CH3 coumarins (4 and 12), while the higher red-shift absorption of the 4-CF3 compounds (5 and 13) allowed the use of a red one (λex = 633 nm). As shown in Figure 4, in all cases the fluorescence signal was clearly observed inside the cell, which confirms that the excellent cellular uptake of the parent COUPY dyes was retained after replacement of pyridine with pyrimidine. In addition, it is worth noting that no cell toxicity was observed during these studies. The overall pattern of staining (but not the relative fluorescence intensity between organelles; see below for a discussion) of pyrimidinecontaining coumarins (12 and 13) was similar to that found with the parent fluorophores (  All images are at the same scale as A and colour coded using the Fire lookup table from Fiji (intensity calibration bar is showed in B).
As shown in Figure 5, the distribution of the fluorescence emission of the compounds was similar to that of Mitotracker Green FM, which confirmed accumulation into the mitochondria. Pearson's and Manders' (M1 and M2) coefficients were used to measure the degree of co-localization. [11][12]18 On the one hand, Pearson's coefficients of 0.86 (12) and 0.87 (13) confirmed a clear correlation between the coumarins' signals and MTG (Pearson's coefficients range from −1 to +1, being +1 the indicator of a perfect match). Such coefficients were higher than those obtained for the parent COUPY dyes (0.65 for 4 and 0.73 for 5), indicating a better correlation between pyrimidine-containing coumarins and MTG. On the other hand, the Manders' coefficients (which range from 0 to 1 and determine the intensities of one channel co-localizing with the other) also confirmed that 12 and 13 were mainly placed in the mitochondria. The degree of co-localization of 12 over MTG (M1 coefficient) was 0.40, whereas that of MTG over 12 (M2 coefficient) was 0.73. These values indicate that there is more MTG signal co-localizing with 12 than 12 that co-localizing with MTG. The localization of the coumarin probe in other organelles such as nucleoli and intracellular vesicles accounts for the differences in both Manders coefficients. It is worth noting that smaller values for M1 (0.28) and M2 (0.68) were obtained in the case of the parent coumarin 4, which indicates that the signal from the fluorophore that co-localizes with MTG is higher in the case of pyrimidine-containing coumarin (12) than in the pyridine analogue (4)  Very interestingly, the nucleoli inside the nuclei was much less intensely stained in compounds 12 and 13 than in the parent coumarins: compare in Figure 4 panels C (12) and D (13) with panels A (4) and B (5), respectively. As shown in Figure 7, measurement of the mean fluorescence intensity both in mitochondria and nucleoli confirmed this observation.

Materials and Methods
Unless otherwise stated, common chemicals and solvents (HPLC grade or reagent grade quality) were purchased from commercial sources and used without further purification.
Aluminium plates coated with a 0.2 mm thick layer of silica gel 60 F254 were used for thinlayer chromatography analyses (TLC), whereas flash column chromatography purification was carried out using silica gel 60 (230-400 mesh). Reversed-phase high-performance liquid chromatography (HPLC) analyses were carried out on a Jupiter Proteo C18 column (250x4.

2-(Pyrimidin-4-yl)acetonitrile (16)
A solution of hydroxylamine-O-sulfonic acid (7.31 g, 57.5 mmol) in Milli-Q water (100 mL) was added to the crude product 17, and the reaction mixture was heated at 50 ºC with a hot plate magnetic stirrer for 45 min. Then, the mixture was cooled at 0 ºC and saturated NaHCO3 was added until basic pH, and the aqueous phase extracted with ethyl acetate (6 x 200 mL).

Photophysical characterization of the compounds.
Absorption spectra were recorded in a Jasco V-730 spectrophotometer at room temperature. (1) where AreaSample and AreaRef are the integrated fluorescence for the sample and the reference and ƞSample and ƞRef are the refractive index of sample and reference solutions respectively.
The uncertainty in the experimental value of F has been estimated to be approximately 10%.  Image processing and analysis was performed using Fiji. 20 Intensity measurement: the nuclei and coumarin channels were processed by median filtering (radius = 2), Gaussian filtering (sigma = 2) and background subtraction (rolling ball radius = 300). Mean intensity in the nucleoli was measured in the maximum intensity projection of the processed coumarin image after manually draw ROIs around each nucleoli.
Intensity measurement in mitochondria needed of further processing. First, the nuclei were segmented using Intermodes algorithm, 21 and the resulting binary image was processed by filling holes and opening operations. Then the binary image of the nuclei was subtracted to the processed coumarin image to get rid of any nuclear signal. After the subtraction, the coumarin staining channel was projected by the maxim intensity and the mitochondria were segmented using the Phansalkar algorithm (radius = 5). 22 The binary image of the mitochondria was used as a mask to obtain the coumarin signal in the mitochondria and then measure its mean intensity.
Bleaching analysis: Images were first processed by median filtering (radius = 1) and background subtraction (rolling ball radius = 50). Then a ROI was manually drawn around each cell and in the background to measure the mean intensity along time. Intensity normalization was performed using the following equation (2): were Cell(t) is the mean intensity in a cell at a t time, Cell(0) is the mean intensity in that cell at the beginning of the experiment, Backg(t) is the mean intensity in the background at a t time and Backg(0) is the mean intensity in the background at the beginning of the experiment.
After intensity normalization, the time at which the intensity dropped to half of the initial one (t50) was obtained. Differences between the different t50 of coumarin 4 and 12 were tested with a T-Student.
Co-localizations coefficients. The mitotracker or lysotracker and coumarin channels were processed by median filtering (radius = 1), Gaussian filtering (sigma = 1) and background subtraction (rolling ball radius = 300). Colocalization coefficients were measured using the JaCoP plugin 17 on the different stacks of images (n = 4) with each stack containing 3−5 cells.
The threshold for the coumarin channel was set to include the signal in the mitochondria, nucleoli and vesicles. The threshold for the mitotracker or lysotracker channels was set to select specifically mitochondria and lysosomes respectively.

Supporting Information
Copies of HPLC traces and UV−vis absorption and fluorescence emission spectra of the compounds; additional fluorescence imaging studies; 1D NMR ( 1 H, 13 C, and 19 F), MS, and selected 2D NMR spectra.
This material is available free of charge via the Internet at http://pubs.acs.org.

AUTHOR INFORMATION
Corresponding Author