A structural remedy toward bright dipolar fluorophores in aqueous media

Structural factors governing the poor emission of dipolar dyes in aqueous media are identified, leading to new acedan derivatives with brighter fluorescence and enhanced two-photon properties.


S2
One-photon spectroscopic analysis UV/Vis absorption spectra were obtained using a HP 8453 UV/Vis spectrophotometer. Fluorescence spectra were recorded on a Photon Technology International fluorimeter with a 10 mm cuvette. The excitation and emission wavelength band paths were both set at 2 nm. Stock solutions of each dye were prepared by dissolving them separately in each solvent used (each 1.0 mM). In case of an aqueous solution, a stock solution was made in DMSO (1 mM) and it was added to the aqueous solvent by keeping the concentration of DMSO within 1% of total volume. Final titrant volume is the same for all measurement (3 mL).   S4  (1) and its derivatives 2-9 in different solvents (HEPES buffer, water, ethanol, acetonitrile, DMF, dichloromethane, and cyclohexane). Emission spectra were obtained by dissolving each compound at a concentration of 10 −5 M in the corresponding spectroscopic grade solvent by excitation at the maximum absorbance wavelength (λ abs ) of each compound.

Fig. S3
Lippert-Mataga plot for acedan and its derivatives 1-9 in aprotic solvents (Toluene, dioxane, chloroform, diethyl ether, dichloromethane, dimethylsulfoxide and acetonitrile), where ν a and ν e mean the maximum absorbance wavenumber and the maximum emission wavenumber, respectively, and ∆f means the solvent polarity factor.
The plots show significant deviation both in diethyl ether and in dichloromethane, but apparently give an increment of the slope as the medium polarity increase. Compounds with N,N-dialkyl-substituted-amine donors are differentiated from those with N-monoalkyl-substituted ones; the former derivatives (1, 2, 8, and 9) have larger slopes than the latter compounds (3-7).

Fig. S4
Absorption and emission spectra of naphthalimide derivatives 10a-10c in different solvents (water, acetonitrile, and dichloromethane). Spectra were obtained by dissolving each compound at a concentration of 10 −5 M in the corresponding spectroscopic grade solvent. The emission spectra were obtained by exciting each molecule at the maximum absorption wavelength ( abs ) in the corresponding solvent.

Fluorescence quantum yield measurement
The fluorescence quantum yields were measured by using rhodamine B, rhodamine 6G, and coumarine 47 as references (1), because two-photon absorption cross sections have to be measured as well. The sample solutions were excited by a laser light with the wavelengths tuned to 370, 380, and 390 nm, and fluorescence was detected by a spectrograph equipped with a CCD detector. The full emission spectra were measured and integrated to give the quantum yields. Rhodamine B was employed as a primary reference because it is widely used as a reference for the two-photon absorption cross section measurement (2). Quantum yields of coumarin 47 (also called coumarin 460 or Coumarin 1 which have absorptions and emissions in the similar range that of acedan derivatives) and rhodamine 6G were cross-checked using rhodamine B as the reference to avoid any uncertainty that may arise due to the weak absorption of rhodamine B at those wavelengths. After confirming no uncertainty S8 in the calculated quantum yield of coumarin 47 and rhodamine 6G compared with literature values, we measured the quantum yields of all the acedan, naphthalimide, coumarin and NBD derivatives using rhodamine B as reference.

Two-photon spectroscopic analysis
The measurement of two-photon emission spectra was performed by using home-built cavity-dumped values were determined using rhodamine B as a standard (2).
The measurement of two-photon emission spectra was performed for selected compounds 1, 5-7 and 10a-c.

Exclusion of a possibility of nanoparticle formation in this study
To check whether nanoparticles can be formed under our experimental conditions of preparing aqueous solutions of those acedan derivatives (a DMSO stock solution were mixed with water at 25 °C to make final compound concentration of 10 µM by keeping volume of DMSO at 1%), we have carried out DLS (dynamic light scattering) analyses for aqueous solutions prepared with acedan and the representitive acedan derivative 5 under the same experiemental conditions used in the above: In no cases, we were able to observe nanoparticles from the DLS analyses, confirming that we have dealt with molecular solutions, not nanoparticle solutions.   Output from the source first passed through a combination of a half wave plate and a polarizer for power control.

Preparation of cell lines for imaging
HeLa human cervical carcinoma cells were maintained in Dulbecco's Modified Eagles Medium (DMEM, Hyclone) containing 10% fetal bovine serum (Hyclone) and penicillin-streptomycin (Hyclone) at 37 ºC in a 5% (v/v) CO 2 humid incubator. Approximately, 20,000 cells/cm 2 were seed on a cover glass in each well of 24-well plate. To stain HeLa cells with fluorophores 1, 5, 10a and 10c, the cells were incubated with the corresponding fluorophores (100 μM) for 30 min. Following the incubation, the cells were washed three times with phosphatebuffer saline (PBS) and fixed with 4% paraformaldehyde (PFA) for 10 min.
For imaging of thiols in HeLa cells with probes P1 and P2, the cells were incubated with the corresponding probe (5 µM and 10 μM) for 60 min. Following the incubation, the cells were washed with PBS buffer and fixed with 4% paraformaldehyde (PFA) for 10 min, and then subjected to fluorescence imaging by two-photon microscopy.

Preparation of mouse tissue samples for imaging
C57BL6 type mice (5 weeks old, male, SAMTAKO corp.) were used for this experiment. The mice were dissected to separate three organs: brain, liver and kidney. These organs were washed several times with PBS buffer. After being washed, each organ was freezed by dipping inside liquid nitrogen for 5 minutes. Then, the freezed organs were crushed into small pieces with a hammer. The small organ pieces were fixed by treatment with OCT (optimal cutting temperature medium), and then sliced to make the tissue samples using a section machine (Cryostat machine, Leica, CM3000 model) at a thickness of 16 μm. The tissue slice samples were placed on specimen blocks (Paul Marienfeld GMbH & Co.). These specimen blocks were immersed in 4% paraformaldehyde for 10 min, and then the specimen blocks were washed several times with PBS buffer. After being washed, the tissues were fixed properly on the specimen blocks using a mount solution (Gel Mount, BIOMEDA). The prepared tissue sample slides were incubated by dipping them into a solution of the acedan fluorophore (100 μM in PBS buffer) for 10 min, and then the samples were washed three times with PBS buffer and then fixed with 4% paraformaldehyde (PFA) for 10 min. When the mount solution became hard, the samples were subjected to one-and two-photon microscopic analysis to obtain the fluorescent images.  To inspect the correlation between hydrogen bonding and the fluorescence characteristics from the MD trajectory snapshots, we need to judge whether a given solvent-solute conformation indeed has the hydrogen bonding character. For this purpose, we have adopted a simple approach of using geometrical parameters of the given conformation. Any water molecule was considered to be a candidate for hydrogen bonding when the distance between its hydrogen atom and the nitrogen in the inspected acedan derivative is below 2.27 Å and the OHN angle is larger than 140 deg (14,15).

Supporting results and discussions
Allylic strain. Quantum chemically calculated transition dipoles of acedan (1) and its N-methyl analogue 4 are respectively. One should expect that the higher solvation structure will induce better hydrogen bonding, which will subsequently shift the electron distribution to affect the fluorescence characteristics as well as provide a solventassociated, nonradiative decay route in general.
For more detailed analysis, we have counted the number of water molecules that may be forming hydrogen bond with the nitrogen atom from the total of 51 snapshots along the MD trajectories. Indeed, many snapshots from acedan (1) and compound 4 show one or two nearby water molecules, whereas majority of the snapshots from compounds 5 and 7 are accompanied by no potentially H-bonding water (Table S13). Thus, the H-bonding is less feasible with compound 5, likely due to the high steric constraint and for compound 7 where the nitrogen atom is intramolecularly hydrogen-bonded. Averages from the corresponding snapshots within the given group.
Computing the average oscillator strengths of acedan and compounds 4, 5 and 7 with the surrounding water molecules indeed shows that there is a good correlation between the solvation and the fluorescence strength.
Namely, when there are more surrounding water molecules, the oscillator strength decreases substantially. Therefore, compound 5, with less H-bonding probability due to high steric constraint, will exhibit enhanced fluorescence in the excited state, and the data shown in the Table S13 supports that the H-bonding between the S17 nitrogen in acedan (and its derivatives) and water indeed suppresses the fluorescence. Suppressing the Hbonding around the amine nitrogen thus reduces the water-associated nonradiative decay including the initial solvent reorganization process required for attaining the ICT excited state, plausibly the TICT state.
Another aspect that strongly affects the fluorescence is the relative prevalence of the competing nonradiative decay paths. The torsional twisting around the nitrogen atom is likely to be related to the nonradiative decay. Comparing the torsional angle () distributions obtained with MD simulations for the two acedan compounds (Fig. S18), we can see that the dihedral angle distribution become broader in acedan (1) compared with that in compound 5. In addition, the distributions suggest that the rotational free energy barrier will be much lower with acedan, which, in turn, suggests that the internal rotation and the associated nonradiative decay are more feasible with acedan than with compound 5. Even though this is a rough analysis as our MD simulations only involved ground state parameters and did not incorporate the actual nonadiabatic and nonradiative decay processes, we can at least infer that the fluorescence from acedan will be further suppressed due to the more favorable twisting around the nitrogen atom than that of compound 5. This absorbance increase is likely due to the reduced 1,3-allylic strain, as pointed out above, which promotes the electron delocalization from the nitrogen to the naphthalene moiety.

Estimation of the relative contribution of the three factors (allylic strain, hydrogen bonding and rotational freedom) to emission properties in aqueous media
It is difficult to separate the effect of one factor from others and also the emissive sate is still under debate. By a rough analysis below, it seems that the allylic strain plays the major role in water.
Comparison #1. Let's compare two acedan derivatives 4 and 5, which belong to N-monoalkyl derivatives; the latter shows approximately 1.5-fold enhanced fluorescence in water compared with the former (Fig. 3b). In both cases, we may assume that there is a negligible difference in the allylic strain factor. Hence the remaining factors, the rotational freedom and "N-hydrogen bonding," are expected to play more favorably in the case of compound 5 that has a bulky and hydrophobic amino substituent. Here, we noted as N-hydrogen bonding to specifically indicate the water hydrogen bonding to the nitrogen lone pair, which would reduce the intramolecular charge-transfer (ICT).
Thus, we may roughly say that two factors (rotational freedom + hydrogen bonding) cause 1.5-fold fluorescence enhancement.
Comparison #2. Let's compare acedan (1)   Preparation of compound 2c. In an oven dried round bottom flask containing a stirrer bar was charged with compound 2b (384 mg, 1.48 mmol) and anhydrous DMF (20 mL) under nitrogen at room temperature. The flask was placed in an oil bath, and the mixture was stirred and heated at 135 °C for 4 h until most of the starting material disappeared (TLC). After being cooled to room temperature, the reaction mixture was diluted with ethyl acetate (300 mL). The organic layer was washed with water (3 × 50 mL) and brine, dried over anhydrous   being cooled to room temperature, the mixture was treated with 6 N HCl (4 mL) and heated at 60 °C for 4 h.
The reaction mixture was diluted with ethyl acetate (100 mL), and washed sequentially with water (50 mL), 5% aqueous NaHCO 3 solution (50 mL), and brine. The organic layer was dried over anhydrous Na 2 SO 4 , filtered and concentrated in vacuo. The crude product was then purified by silica gel column chromatography (Hexane/EtOAc = 19/1) to give 2 (100 mg, 62%) as yellow solid, which was further purified by crystallization