An ultrasensitive near-infrared ratiometric fluorescent probe for imaging mitochondrial polarity in live cells and in vivo† †Electronic supplementary information (ESI) available: Synthetic procedure for 1–4; NMR and mass spectra of 1–4; spectroscopic properties and confocal imaging. See DOI: 10.1039/c5sc04099j

We describe a new mitochondria-targeting fluorescent probe MCY-BF2 that is singularly sensitive and specifically responsive to mitochondrial polarity.


Introduction
The polarity of a cell is the feedback of a series of complex mechanisms that establish and maintain functionally of particular domains. 1 Many cellular processes involved in spatial arrangement and protein composition such as differentiation, localized membrane growth, activation of the immune response, directional cell migration, and vectorial transport of molecules across cell layers may lead to changes and development of polarity. Therefore, abnormal changes in polarity are closely linked with disorders and diseases. [2][3][4] As the powerhouses of cells, mitochondria are vital intracellular organelles and play critical roles in cellular metabolism, including providing metabolic energy by oxidative phosphorylation, a cell signaling platform via reactive oxygen species (ROS) production, regulation of Ca 2+ homeostasis, and triggering of cell apoptosis. [5][6][7][8][9] It is noteworthy that the unique functions of mitochondria are closely related to maintaining homeostasis of their parameters and microenvironments, such as pH, [10][11][12] viscosity, 13 polarity, 14 temperature 15 and so on. In particular, mitochondrial polarity is a crucial characteristic of these indispensable organelles. Numerous events may be affected by the mitochondrial polarity, including transportation or interaction of proteins, specic activity and stability of proteins or enzymes, and the maintaining of cell function and homeostasis. Thus, to accurately track mitochondrial polarity is of great importance.
Fluorescence imaging is a promising and powerful method for monitoring various bioactive molecules in living systems. Consequently, an increasing number of uorescent probes suitable for real-time imaging have been developed in recent years, which has facilitated progress in cell biology and therapeutics imaging. [16][17][18] Nevertheless, ideal uorescent probes for monitoring mitochondrial polarity should possess the following unique merits: (1) high sensitivity and selectivity to polarity without being affected by other complicated mitochondrial microenvironments; (2) maximum uorescence excitation and emission wavelengths in the NIR region to reduce interference from background uorescence; (3) accurate mitochondria-targeting ability, and (4) a ratiometric uorescence response for a quantitative assay. Although some polaritysensitive uorescent probes for solvents, the surface of a protein, and even live cells have been exploited in recent years, 14,19-27 a versatile polarity probe satisfying all the aforementioned requirements has not been reported.
To solve these problems mentioned above, we herein fabricated a series of new compounds, 1-4 (Scheme 1), composed of a merocyanine moiety with various lipophilic side chains and a diuoroboronate moiety used to sensitively monitor the polarity. In this framework, the long conjugated system and asymmetric structure account for the NIR excitation/emission spectra and large Stokes shi, respectively. Meanwhile, the tertiary amine and diuoroboronate moieties serve as electron donating and accepting groups, respectively, resulting in the formation of an intense push-pull construction. When the environment polarity increases, the excited state energy can be dissipated due to the dipole-dipole interaction between the probe molecules and solvent due to the excited state ICT, 28,29 which is responsible for the polarity sensitivity. Therefore, the designed compounds should exhibit weak uorescence and a longer emission wavelength in polar media, and in contrast, a strong uorescence and shorter emission wavelength in nonpolar media. 23 As a result of red-shiing of the emission wavelength with increasing polarity, a plot of the uorescence intensity ratios at two different wavelengths versus the dielectric constant may be achieved, which can then be utilized to estimate the polarity of certain media and live cells. Considering the characteristics of mitochondria which distinguish them from other subcellular compartments, cardiolipin (CL, a diphosphatidylglycerol lipid) is exclusively found in the inner membrane of mitochondria. 30 Previous studies revealed that some molecules bearing lipophilic cations can effectively bind mitochondrial CL through electrostatic attraction and lipophilic interactions. 31 Based on this idea, we introduced various substituents, including ethyl, benzyl, n-decane and n-hexadecyl groups attached to the nitrogen atom to give four compounds. In one resonance structure of these molecules, there is one positive charge on the nitrogen atom (Scheme S1 †). Therefore we speculate that they can grasp CL to target mitochondria specically.
In this article, we present the synthesis and characterization of four new polarity sensitive compounds. By analyzing their optical properties, we found these compounds could accurately analyse the dielectric constant of the microenvironment. Further comparison with their localization effect shows that a selected compound termed MCY-BF 2 bearing an n-hexadecyl group displays a better mitochondria-targeting capability than the other three compounds. As the rst NIR ratiometric uorescent probe, it was utilized to discriminate the polarity variance between normal and cancer cells. By means of the uorescence intensity ratio of 795 and 765 nm, the precise mitochondria polarity in different cells and in C. elegans at different developmental stages was also measured. In addition, the polarity difference in normal and tumor s of a mouse was visualized in vivo.

Results and discussion
Compounds 1-4, as green solids, were synthesized in moderate yield under relatively mild conditions (Scheme S2 †). In brief, quaternary ammonium salts containing various lipophilic side chains were reacted with 2-chloro-1-formyl-3-hydroxymethylenecyclohexene to obtain intermediates 5-8. Subsequently, intermediates 5-8 underwent a condensation reaction with compound c to form compounds 1-4. Full synthetic details and characterization of the new compounds can be found in the ESI. † With these compounds in hand, we rstly tested their photophysical properties in detail. The absorption and emission proles of 1-4 in eight common solvents i.e. water (dielectric Fig. 1, Fig. S1 and Table S1. † As expected, their maximum emission peaks are all in the NIR range (>750 nm). All the emission spectra of 1-4 show dramatic polarity-dependence and the absorption intensity in all the solvents slightly changes ( Fig. 1 and S1 †). For example, there is a 120-fold increase in the uorescence intensity in dioxane (3 ¼ 2.21) compared with that in DMSO (3 ¼ 48.9) for MCY-BF 2 at 800 nm. These phenomena are in accord with the polarity responsive principle. 28 Accompanied by the increase in solvent polarity, all four compounds exhibit Scheme 1 Chemical structures of compounds 1-4 and the corresponding functional units. a red-shi in the maximum emission wavelength, from about 780 nm in dioxane to about 830 nm in acetone. Meanwhile, the uorescence quantum yield decreases noticeably with increasing solvent polarity from about 11% in dioxane to about 1.0% in DMSO (Table S1 †).
Next we evaluated the mitochondrial localization ability of the four compounds. Firstly, co-localization imaging experiments were performed in living mouse mammary carcinoma 4T1 cells simultaneously loaded with these compounds and Mito Tracker Green, a commercial mitochondrial dye. As shown in Fig. S2, † all four compounds can readily penetrate into 4T1 cells. By comparing the co-localization effects, MCY-BF 2 showed apparently superior mitochondria-targeting ability to compounds 1, 2 and 4. In addition, a similar targeting ability of MCY-BF 2 was testied in human hepatoma cells (HepG2) by means of co-localization experiments using Mito Green Tracker ( Fig. 2A) and 5,5 0 ,6,6 0 -tetrachloro-1,1 0 ,3,3 0 -tetraethylbenzimidazolcarbocyanine iodide (JC-1, Fig. S3 †), with a co-localization coefficient of 0.91 and 0.92, respectively. Other subcellular compartment stain experiments using MCY-BF 2 and the corresponding commercial organelle-specic dye were then carried out (Fig. S4 †). Inside lysosomes (Lyso), endoplasmic reticulum (ER), and Golgi apparatus (Golgi), signicantly smaller colocalization coefficient values were found. These results demonstrate that MCY-BF 2 can preferentially accumulate in mitochondria. To further verify the mitochondria-targeting mechanism, 4T1 cells treated with the membrane-potential uncoupler 3-chlorophenylhydrazone (CCCP) that can collapse the mitochondrial membrane potential were simultaneously incubated with MCY-BF 2 and Mito-Tracker Green. In this experiment, we found that the uorescence brightness of MCY-BF 2 was consistent in the absence or presence of CCCP ( Fig. 2B3  and B4). Moreover, the mitochondria-targeting effect of MCY-BF 2 is almost unchanged upon treatment with CCCP ( Fig. 2B5 and B6). The above results implied that MCY-BF 2 might locate in mitochondria in live cells via anchoring CL due to the lipophilic cation group bearing long aliphatic chains, and is independent of the mitochondrial membrane potential. Also, Fig. S5 † suggests that MCY-BF 2 has low cytotoxicity. On the basis of the above selectivity results, we draw the conclusion that the selected compound MCY-BF 2 is a new fascinating probe endowed with polarity sensitivity and excellent mitochondriatargeting ability.
Subsequently, we investigated whether CL could inuence the uorescence of MCY-BF 2 , since it accumulates in mitochondria in combination with CL. As illustrated in Fig. 3A, upon addition of CL from 0 to 20 mM i.e., the physiological range of CL in mitochondrial membranes, 32,33 the uorescence intensity of MCY-BF 2 (10 mM, 3 ¼ 80.4) increased slightly. In contrast, MCY-BF 2 showed intense uorescence in a low polarity environment (10 mM, 3 ¼ 17) in the presence of 20 mM CL. These results validate that MCY-BF 2 was not affected by CL. In addition, interference experiments showed that the uorescence signal of MCY-BF 2 changed very little in the presence of various ROS, nucleophilic thiols, amino acids, different pH buffers and viscosity (Fig. S6-S8 †). Further experiments showed that MCY-BF 2 had high photostability (Fig. S9 †).
To accurately analyze the mitochondrial polarity in living cells and in vivo using MCY-BF 2 , the curve of the quantitative relationship between the uorescence intensity ratio and polarity should be examined. For this purpose, we investigated the emission of MCY-BF 2 in mixtures of dioxane and water with different proportions of water from 0 to 40% to represent the rise in polarity. 34 Fig. S10A † indicates that the absorbance of MCY-BF 2 in mixtures of water from 0 to 40% is almost unchanged, whereas, as illustrated in Fig. 3B, the emission maximum wavelength is red-shied from about 780 nm in dioxane to about 820 nm in 40% water. Notably, there is an approximately 45-fold enhancement in the uorescence intensity at 780 nm when decreasing the polarity of the solvent from a dielectric constant of 28.1 (40% water) to 2.21 (0% water). The relative uorescence quantum yield was determined to be 11% in 0% water and 1.1% in 40% water (Fig. S10B †). All the results suggest that MCY-BF 2 is extremely sensitive to polarity changes. More importantly, to the best of our knowledge, this is the most sensitive near-infrared uorescent probe for polarity up to now. By plotting the uorescence intensity ratio F 795 /F 765 against the dielectric constant (Fig. 3C), a calibration curve was obtained with a relationship coefficient of R 2 ¼ 0.99. Thus, it can be used to measure the polarity of certain media or live cells using confocal uorescence imaging in view of the collection window up to a maximum wavelength of 800 nm. Moreover, Fig. 3C reveals that the F 795 /F 765 ratio increases dramatically from 0.8 to 3.0 with increasing solvent polarity when the dielectric constant is below 15, then the change tends to be gradual when the solvent dielectric constant is above 20, indicating that MCY-BF 2 is extraordinarily sensitive to weak polarity media. Taken together, MCY-BF 2 is an excellent near-infrared uorescent probe that can be used to exclusively quantify polarity uctuations based on changes in its uorescence intensity ratio F 795 / F 765 .
Accumulating evidence suggests that normal and cancer cells display microenvironment differences, [35][36][37] prompting us to exploit the biological feasibility of MCY-BF 2 as a ratiometric uorescence imaging probe. We intend to quantitatively detect mitochondrial polarity in cells using confocal uorescence imaging in different cells. HepG2 cells and normal human liver cells (HL-7702) were incubated with MCY-BF 2 (10 mM), and two uorescence channels, green for 760-770 nm and red for 790-800 nm, were collected. As illustrated in Fig. 4, the red uorescence intensity (Fig. 4B and F) was obviously higher. Fig. 4D and H show the uorescence intensity ratio between the red and green channels, in which a mean ratio was 2.44 AE 0.13 (Fig. 4I) in HepG2 cell mitochondria. The corresponding dielectric constant was 7.72 AE 0.32 according to the uorescence intensity ratio curve in Fig. 3C. The mean ratio in HL-7702 cell mitochondria is 4.40 AE 0.19, which indicates that the corresponding dielectric constant should be more than 30. Similar results are obtained in 4T1 cells and normal human mammary epithelial cells MCF-10A (Fig. S11 †). This demonstrates that polarity in cancer cells is lower than that in normal cells. [38][39][40] All these results substantiated that MCY-BF 2 can serve as a highly sensitive ratiometric uorescence imaging probe to quantitatively indicate the mitochondrial polarity within live cells.
Previous studies speculate that the polarity may change continuously in the C. elegans development process. 41,42 To explore whether these inherent polarity differences exist, we used MCY-BF 2 to measure the polarity by in vivo visualization of C. elegans at various development stages. C. elegans at embryonic development stage (EDS) and young adult stage (YAS) were incubated with MCY-BF 2 (20 mM) for 30 min, aer which confocal uorescence imaging was performed immediately without washing the incubation solution. As shown in Fig. 5, MCY-BF 2 can uoresce well in C. elegans both at EDS and YAS. Interestingly, its uorescence at EDS (Fig. 5A and B) was remarkably brighter than that at YAS (Fig. 5E and F). To further quantify the precise polarity of C. elegans at EDS and YAS, we investigated the uorescence ratio based on two groups of  confocal uorescence images captured with l em ¼ 760-770 nm and l em ¼ 790-800 nm. As shown in Fig. 5, the ratio of the two channels at EDS (Fig. 5D) is obviously smaller than that at YAS (Fig. 5H). The mean ratio value of C. elegans at EDS is 2.36 AE 0.05, which corresponds to a dielectric constant of 7.20 AE 0.19. Meanwhile, the mean ratio at YAS is 2.76 AE 0.11, indicating a dielectric constant of 10.07 AE 0.29 (Fig. 5I). This is the rst time it has been conrmed that the polarity of C. elegans at EDS is smaller than that at YAS.
One of major advantages of the new MCY-BF 2 probe over existing polarity-sensitive probes is that it has both absorption and emission in the NIR region. We consider MCY-BF 2 is well suited for biological imaging in live animals because NIR light leads to minimum photodamage to biological samples, deep tissue penetration, and reduced interference from background autouorescence by biomolecules in living animals. [43][44][45] Thus, a further effort was made to distinguish the polarity variance of normal and tumor tissues in live small animals. We constructed a mouse model of mammary carcinoma for in vivo imaging. Specically, mammary carcinoma 4T1 cells were inoculated into the neck of mice, and aer 10 days a tumor mass was obtained. Subsequently, MCY-BF 2 (2 Â 10 À5 M in PBS containing 1% DMSO, 100 mL) was injected into the tumor-bearing and normal abdomen tissue of the mice, followed by in situ imaging using IVIS Lumina III without shaving the mouses' skin. The neck tumor mass exhibited a signicantly stronger uorescence (pseudocolor) than that of the normal tissue in the abdomen. The neck tumor tissue displayed an approximately 9-fold higher uorescence intensity than that of normal tissue (Fig. 6B). We investigated that hypoxia had no effect on the uorescence of MCY-BF 2 (Fig. S12 †), further conrming that lower polarity in tumor tissue indeed causes intense uorescence. These results demonstrate that MCY-BF 2 is a prominent uorescent probe for imaging polarity differences in vivo.

Conclusion
In conclusion, we have developed a new near-infrared ratiometric uorescent probe, MCY-BF 2 , for ultrasensitive sensing of mitochondrial polarity. It has both absorption and emission in the NIR region as well as a modest Stokes shi and is insensitive to various ROS, thiols, amino acids, oxygen, pH values and viscosity. Also, it exhibits good photostability and low cytotoxicity. MCY-BF 2 can effectively stain mitochondria by anchoring CL in the inner membrane of mitochondria independent of the mitochondrial membrane potential. It was successfully utilized to distinguish cancer cell and normal cell polarity differences of mitochondria. Moreover, the uorescence intensity ratio F 795 /F 765 at different developmental stages of C. elegans showed that the embryonic development stage has a less polar microenvironment with a dielectric constant of 7.20 AE 0.19, while the young adult stage has a more polar microenvironment with a dielectric constant of 10.07 AE 0.29. Furthermore, MCY-BF 2 was successfully applied to monitor the polarity distinction in normal and tumor tissues in live animals. To the best of our knowledge, this is the rst near-infrared and most sensitive ratiometric uorescent probe for imaging polarity up to now. Our present work may provide a new method to quantitatively study the polarity both in normal or cancer cells and in vivo. Furthermore, it paves the way to elucidating biological processes and pathological mechanisms of mitochondrial polarity-related diseases.