A Deoxyuridine-Based Far-Red Emitting Viscosity Sensor

A novel deoxyuridine (dU) benzothiazolium (BZ) derivative, referred to as dU-BZ, is reported that was synthesized via Sonogashira coupling reaction methodology. The deoxyuridine building block was introduced to enhance hydrophilicity, while an alkynylated benzothiazolium dye was incorporated for long wavelength absorption to reduce potential phototoxicity that is characteristic of using UV light to excite common fluorphores, better discriminate from native autofluorescence, and potentially facilitate deep tissue imaging. An impressive 30-fold enhancement of fluorescence intensity of dU-BZ was achieved upon increasing viscosity. Fluorescence quantum yields in 99% glycerol/1% methanol (v/v) solution as a function of temperature (293–343 K), together with viscosity-dependent fluorescence lifetimes and radiative and non-radiative rate constants in glycerol/methanol solutions (ranging from 4.8 to 950 cP) were determined. Both fluorescence quantum yields and lifetimes increased with increased viscosity, consistent with results predicted by theory. This suggests that the newly-designed compound, dU-BZ, is capable of functioning as a probe of local microviscosity, an aspect examined by in vitro bioimaging experiments.


Introduction
Information, such as physiological composition, can be reflected by microenvironments in cellular compartments [1]. For example, an increase in mitochondrial membrane viscosity was discovered after exposure to β-amyloid, which is essentially involved in Alzheimer 1 s disease [2]. Mechanical methods [3,4] have been universally applied to measure the viscosity of bulk liquids. However, viscosity on the microscopic scale may differ largely. It is a significant challenge to use techniques to measure microviscosity on the order of micrometers so that intracellular viscosity can be probed.
One method to monitor viscosity changes at the single cell level is the use of fluorescence imaging with molecular rotors [5,6]. Molecular rotors are fluorophores whose fluorescence intensity is affected by intramolecular rotation that can be greatly affected by the viscosity of its surrounding environment. This can be accomplished via an intramolecular charge transfer (ICT) mechanism by molecular twisting in the excited state. Two excited states, a local excited (LE) state and twisted intramolecular charge

Synthesis of dU-BZ
The synthesis of molecular rotor dU-BZ is illustrated in Scheme 1. Intermediate compound 3 was synthesized according to the literature [25], and the resulting NMR matched the reported data. Next, compound 4 was synthesized by condensation of intermediate 3 with dimethylaminobenzaldehyde via a Knoevenagel reaction; acetic anhydride was employed as both base and solvent.
The following procedure was used for the conjugation of a deoxyuridine analog and alkynylbenzothiazolium 4 through a triple bond. Rather than directly using deoxyuridine, a modified form, idoxuridine, was exploited not only due to the iodo group provided for conjugation, but also its structure may be incorporated into DNA/RNA strands for future study. Hydroxyl groups of idoxuridine are all unprotected in order to avoid possible low overall yield. Although protected hydroxyl groups possess enhanced water solubility when compared to unprotected ones, in consideration of the overall yield of the reaction, protected nucleosides were not pursued.
dU-BZ was obtained via Sonogashira coupling between (+)-5-iodo-2 1 -deoxyuridine and alkyne 4 in 21% yield after purification by column chromatography. After conjugation with the deoxyuridine analog, dU-BZ exhibited enhanced water solubility when compared to 4. The 1 H-NMR, 13 C-NMR, and HR-MS spectra were in good accordance with the chemical structure as expected.

Synthesis of dU-BZ
The synthesis of molecular rotor dU-BZ is illustrated in Scheme 1. Intermediate compound 3 was synthesized according to the literature [25], and the resulting NMR matched the reported data. Next, compound 4 was synthesized by condensation of intermediate 3 with dimethylaminobenzaldehyde via a Knoevenagel reaction; acetic anhydride was employed as both base and solvent.
The following procedure was used for the conjugation of a deoxyuridine analog and alkynylbenzothiazolium 4 through a triple bond. Rather than directly using deoxyuridine, a modified form, idoxuridine, was exploited not only due to the iodo group provided for conjugation, but also its structure may be incorporated into DNA/RNA strands for future study. Hydroxyl groups of idoxuridine are all unprotected in order to avoid possible low overall yield. Although protected hydroxyl groups possess enhanced water solubility when compared to unprotected ones, in consideration of the overall yield of the reaction, protected nucleosides were not pursued.
dU-BZ was obtained via Sonogashira coupling between (+)-5-iodo-2′-deoxyuridine and alkyne 4 in 21% yield after purification by column chromatography. After conjugation with the deoxyuridine analog, dU-BZ exhibited enhanced water solubility when compared to 4. The 1 H-NMR, 13 C-NMR, and HR-MS spectra were in good accordance with the chemical structure as expected. Scheme 1. Synthetic route of molecular rotor dU-BZ. The arrow on the dU-BZ structure indicates the major molecular rotation that contributes to the fluorescence on-off feature.

Linear Photophsical Characterization of dU-BZ by Varying Temperature
The fluorescence quantum yields (Φf) of dU-BZ were measured in 99% glycerol/1% methanol (v/v) solution at various temperatures, ranging from 343 to 293 K, with viscosity ranging from 50.6 to 1412 cP (Table 1). According to Figure 1a,b, no significant shifts were observed in absorption and emission spectra, but an increase in the fluorescence intensity was obtained with decreasing temperature when excited at 551 nm, and Φf increased from 0.04 to 0.34. Scheme 1. Synthetic route of molecular rotor dU-BZ. The arrow on the dU-BZ structure indicates the major molecular rotation that contributes to the fluorescence on-off feature.

Linear Photophsical Characterization of dU-BZ by Varying Temperature
The fluorescence quantum yields (Φ f ) of dU-BZ were measured in 99% glycerol/1% methanol (v/v) solution at various temperatures, ranging from 343 to 293 K, with viscosity ranging from 50.6 to 1412 cP (Table 1). According to Figure 1a,b, no significant shifts were observed in absorption and emission spectra, but an increase in the fluorescence intensity was obtained with decreasing temperature when excited at 551 nm, and Φ f increased from 0.04 to 0.34.

Linear Photophysical Characterization of dU-BZ as A Function of Viscosity
At the same concentration, there was no obvious change observed in the absorption spectra for dU-BZ by varying the ratio of glycerol and methanol in solution. However, without any changes in shape of the emission spectrum or the peak emission wavelength, a 30-fold increase in fluorescence intensity appeared at 608 nm by increasing the viscosity from 1.8 cP to 950 cP ( Figure 2).  Figure 3 shows the fluorescence lifetime decay of dU-BZ with decreasing viscosity in glycerol/methanol solutions. As a function of viscosity, the fluorescence lifetime varied markedly from 0.19 ns at 58 cP to 1.07 ns at 950 cP.

Linear Photophysical Characterization of dU-BZ as A Function of Viscosity
At the same concentration, there was no obvious change observed in the absorption spectra for dU-BZ by varying the ratio of glycerol and methanol in solution. However, without any changes in shape of the emission spectrum or the peak emission wavelength, a 30-fold increase in fluorescence intensity appeared at 608 nm by increasing the viscosity from 1.8 cP to 950 cP ( Figure 2).

Linear Photophysical Characterization of dU-BZ as A Function of Viscosity
At the same concentration, there was no obvious change observed in the absorption spectra for dU-BZ by varying the ratio of glycerol and methanol in solution. However, without any changes in shape of the emission spectrum or the peak emission wavelength, a 30-fold increase in fluorescence intensity appeared at 608 nm by increasing the viscosity from 1.8 cP to 950 cP ( Figure 2).  Figure 3 shows the fluorescence lifetime decay of dU-BZ with decreasing viscosity in glycerol/methanol solutions. As a function of viscosity, the fluorescence lifetime varied markedly from 0.19 ns at 58 cP to 1.07 ns at 950 cP.

In Vitro Bioimaging of dU-BZ
Highly viscous, up to 400 cP, intra-and intercellular environments [15] have been reported. The in vitro fluorescence enhancement using dU-BZ was explored, and, indeed, after incubation with 3T3 cells (mouse embryonic fibroblast cells) for 30 min, dU-BZ appeared to readily enter the cells, and remarkably clear fluorescence images were obtained ( Figure 4, Hoechst stained cell nuclei as reference). These results suggest that dU-BZ can, potentially, be utilized to visualize viscous regions at the cellular level, providing motivation for further studies with this promising new probe.

Discussion
Free-volume concepts [27] can be described by the fluorescence quantum yield, Φf, viscosity, η, and temperature,T [9]: where B = (kr /knr0)·(T/A) x , knr0 is the free-rotor reorientation rate, A is a constant, and x is a mediumdependent constant ranging between 0 and 1. When Φf is linearly related to η/T (x = 1), the bulk viscosity of solvent can accurately indicate the friction experienced by the molecular rotor. Normally, Log Intensity (AU)

In Vitro Bioimaging of dU-BZ
Highly viscous, up to 400 cP, intra-and intercellular environments [15] have been reported. The in vitro fluorescence enhancement using dU-BZ was explored, and, indeed, after incubation with 3T3 cells (mouse embryonic fibroblast cells) for 30 min, dU-BZ appeared to readily enter the cells, and remarkably clear fluorescence images were obtained (Figure 4, Hoechst stained cell nuclei as reference). These results suggest that dU-BZ can, potentially, be utilized to visualize viscous regions at the cellular level, providing motivation for further studies with this promising new probe.

In Vitro Bioimaging of dU-BZ
Highly viscous, up to 400 cP, intra-and intercellular environments [15] have been reported. The in vitro fluorescence enhancement using dU-BZ was explored, and, indeed, after incubation with 3T3 cells (mouse embryonic fibroblast cells) for 30 min, dU-BZ appeared to readily enter the cells, and remarkably clear fluorescence images were obtained (Figure 4, Hoechst stained cell nuclei as reference). These results suggest that dU-BZ can, potentially, be utilized to visualize viscous regions at the cellular level, providing motivation for further studies with this promising new probe.

Discussion
Free-volume concepts [27] can be described by the fluorescence quantum yield, Φf, viscosity, η, and temperature,T [9]: where B = (kr /knr0)·(T/A) x , knr0 is the free-rotor reorientation rate, A is a constant, and x is a mediumdependent constant ranging between 0 and 1. When Φf is linearly related to η/T (x = 1), the bulk viscosity of solvent can accurately indicate the friction experienced by the molecular rotor. Normally, Log Intensity (AU)

Discussion
Free-volume concepts [27] can be described by the fluorescence quantum yield, Φ f , viscosity, η, and temperature, T [9]: where B = (k r /k nr0 )¨(T/A) x , k nr0 is the free-rotor reorientation rate, A is a constant, and x is a medium-dependent constant ranging between 0 and 1. When Φ f is linearly related to η/T (x = 1), the bulk viscosity of solvent can accurately indicate the friction experienced by the molecular rotor. Normally, faster rotational diffusion is expected because the fluorophore can occupy a certain free volume within the solvent, in which case x < 1. A plot of log Φ f verses log (η/T) yields a straight line with the exponent x as its slope via Equation (2): As shown in Figure 5, linear behavior was observed when plotting log Φ f vs. log (η/T). The slope of this plot provided the exponent x, 0.57˘0.04, in the range from 0 to 1, with a R 2 value of 0.98. Due to increased viscosity and decreased free volume, a decreased non-radiative rate constant is expected, and this prediction will be described in the following experiments.
The Förster-Hoffmann equation [28] can be used to describe Φ f , and fluorescent lifetime, τ f , of molecular rotors as a function of η: where: then: where z and α are constants, the value of 2/3 for α is predicted by Förster and Hoffmann, and k r and knr are radiative and non-radiative rate constants. [29] According to Equation (6), a straight line with a slope of α will be yielded after plotting log τ f verses log η, since: log τ f " α log η`log pz{k r q, Molecules 2016, 21, 709 6 of 12 faster rotational diffusion is expected because the fluorophore can occupy a certain free volume within the solvent, in which case x < 1. A plot of log Φf verses log (η/T) yields a straight line with the exponent x as its slope via Equation (2): As shown in Figure 5, linear behavior was observed when plotting log Φf vs. log (η/T). The slope of this plot provided the exponent x, 0.57 ± 0.04, in the range from 0 to 1, with a R 2 value of 0.98. Due to increased viscosity and decreased free volume, a decreased non-radiative rate constant is expected, and this prediction will be described in the following experiments.
The Förster-Hoffmann equation [28] can be used to describe Φf, and fluorescent lifetime, τf, of molecular rotors as a function of η: where: τf =1/(kr + knr), then: where z and α are constants, the value of 2/3 for α is predicted by Förster and Hoffmann, and kr and knr are radiative and non-radiative rate constants. [29] According to Equation (6), a straight line with a slope of α will be yielded after plotting log τf verses log η, since: One should note that Equation (3) can only be applied over a limited range of viscosities. According to the Förster-Hoffmann theory, Φf is solvent-independent at low viscosities, whereas at relatively high viscosities, a strong dependence on viscosity of Φf is expected, since radiative processes predominate over non-radiative relaxation. This very range of viscosities is determined by the properties of the particular molecular rotor and the mechanism of viscosity-dependent photophysical behavior.
Measured values, Φf and τf, were used to calculate the rate constants via Equations (2) and (3). Plotted in Figure 6, with viscosity increasing from 58 cP to 950 cP (Table 2), Φf shows significantly increased values as expected. It is worth noting that kr remained constant but knr decreased largely as a function of viscosity. These data suggest that the main contribution to the increase of Φf is via One should note that Equation (3) can only be applied over a limited range of viscosities. According to the Förster-Hoffmann theory, Φ f is solvent-independent at low viscosities, whereas at relatively high viscosities, a strong dependence on viscosity of Φ f is expected, since radiative processes predominate over non-radiative relaxation. This very range of viscosities is determined by the properties of the particular molecular rotor and the mechanism of viscosity-dependent photophysical behavior.
Measured values, Φ f and τ f , were used to calculate the rate constants via Equations (2) and (3). Plotted in Figure 6, with viscosity increasing from 58 cP to 950 cP (Table 2), Φ f shows significantly increased values as expected. It is worth noting that k r remained constant but k nr decreased largely as a function of viscosity. These data suggest that the main contribution to the increase of Φ f is via suppression of the non-radiative process. In a highly viscous environment, because of the intramolecular rotation hindrance, the torsion angle between the benzothiazole and aminobenzene rings is close to zero, which yields the most stable conformation of the molecule in the LE state. At the same time, non-radiative relaxation to the TICT state, which has a conformation angle value close to 90˝, is deactivated, and radiative decay from LE state to ground state starts to take place instead of de-excitation from the TICT state. Table 2. Fluorescence quantum yield (φ f ), fluorescence lifetime (τ f ) *, radiative (k r ) and non-radiative (k nr ) rate constants of dU-BZ as a function of viscosities (η) in glycerol/methanol solutions.

% Glycerol
Viscosity ( suppression of the non-radiative process. In a highly viscous environment, because of the intramolecular rotation hindrance, the torsion angle between the benzothiazole and aminobenzene rings is close to zero, which yields the most stable conformation of the molecule in the LE state. At the same time, non-radiative relaxation to the TICT state, which has a conformation angle value close to 90°, is deactivated, and radiative decay from LE state to ground state starts to take place instead of de-excitation from the TICT state.  From Equation (7), a straight line was obtained in the plot of log τf vs. log η and, as expected, linear behavior was obtained (Figure 7) with a slope, α, of 0.59 ± 0.04, consistent with the value predicted by the Förster-Hoffmann equation, and a R 2 value of 0.96 for dU-BZ. It was also found that plots below 58 cP fit in the same straight line, but lifetime values lower than 0.2 ns were not reliable due to the resolution of the experimental system (Table 2). Only plots from 58 cP to 950 cP are shown in the figure. From Equation (7), a straight line was obtained in the plot of log τ f vs. log η and, as expected, linear behavior was obtained (Figure 7) with a slope, α, of 0.59˘0.04, consistent with the value predicted by the Förster-Hoffmann equation, and a R 2 value of 0.96 for dU-BZ. It was also found that plots below 58 cP fit in the same straight line, but lifetime values lower than 0.2 ns were not reliable due to the resolution of the experimental system (Table 2). Only plots from 58 cP to 950 cP are shown in the figure.

Synthesis
Synthetic reagents and solvents were used as received from commercial suppliers. 5-Bromo-2-methylbenzothiazole was purchased from TCI America (Portland, OR, USA). Iodoethane and (+)-5-iodo-2 1 -deoxyuridine were purchased from Alfa Aesar (Ward Hill, MA, USA). 1 H-and 13 C-NMR spectra were recorded on a Bruker Avance III 400 NMR spectrometer at 400 and 101 MHz, respectively (Billerica, MA, USA). High-resolution mass spectrometry analysis was performed in the Department of Chemistry, University of Florida. Uncorrected melting points were determined using a Laboratory Devices mel-temp.

Viscosity Values
Viscosities of pure glycerol at different temperature were employed to approximate the viscosity of 99% glycerol/1% methanol solution. Reported values [30] were directly used as viscosities of solutions with glycerol percentages ranging from 10% to 95%.

Linear Photophysical Characterization
The linear absorption spectra were obtained using an Agilent 8453 UV´VIS spectrophotometer (Agilent, Santa Clara, CA, USA) in 10 mm path length quartz cuvettes in solvents with different glycerol/methanol ratios, with molar concentration C = 1ˆ10´5 M. The steady-state fluorescence was measured with a PTI QuantaMaster spectrofluorimeter using 10 mm spectrofluorometric quartz cuvettes with C = 1ˆ10´6 M. The correction for the spectral response of the PTI detection system was performed for all fluorescence spectra. The fluorescence quantum yields, Φ f , were obtained by a standard method [31] relative to cresyl violet in methanol. Fluorescence lifetimes, τ f , were measured using a PicoQuant PicoHarp 300 time-correlated single photon-counting system with time resolution « 80 ps, a Coherent Mira 900 fs laser system was used for excitation, linearly polarized at the magic angle.

In Vitro Bioimaging
3T3 cells (ATCC ® , Manassas, VA, USA) were seeded on poly-D-lysine coated coverslips at a concentration of 5ˆ10 4 cells/mL and incubated for 48 h. A dU-BZ stock solution in DMSO (dimethyl sulfoxide) was then diluted to 15 µM with DMEM medium (Cellgro ® , Mediatech, Menassas, VA, USA) and added to the cells. Cells were co-incubated with diluted dU-BZ for 30 min and then fixed with 4% formaldehyde. NaBH4 was added twice at 1 mg/mL for 5 min to reduce auto-fluorescence. Cells were then permeabilized with 0.1% Triton-X. 1% BSA was applied to block non-specific binding. Hoechst 33258 (Invitrogen™, Carlsbad, CA, USA) was added in to cell for 5 min to visualize cell nuclei. Coverslips were then washed with PBS (phosphate-buffered saline, Cellgro ® ) and mounted on slides with ProLong Gold ® (Invitrogen™) antifade reagent (Invitrogen™).
Cells were imaged with an IX70 DSU microscope (Olympus, New York, NY, USA). A Texas Red filter cube (562/40 ex., 593, 624/40 em.) was employed to excite dU-BZ and collect the fluorescence in the optimized wavelength range.

Conclusions
A new deoxyribonucleoside-modified cyanine dye was prepared and characterized. Far-red absorption and emission of this new dye are potentially favorable for in vitro and in vivo imaging to better discriminate the fluorescence signal from autofluorescence and facilitate deep tissue imaging. Viscosity-dependent studies, including fluorescence quantum yields, fluorescence lifetimes, and non-radiative rate constants were determined, and results were in accordance with that predicted by theory for molecular rotors. An impressive 30-fold enhancement in fluorescence intensity in homogenous glycerol/methanol solutions was obtained in a viscosity-dependent manner. Correspondingly, fluorescence lifetimes increased from 0.19 to 1.07 ns with increasing viscosity from 58 to 950 cP. Subsequent in vitro investigation suggested that dU-BZ may be capable of functioning as a microviscosity sensor at cellular and subcellular levels. Our results qualitatively support that the dye can readily enter cells and exhibit a modulated fluorescence response in an in vitro environment as a function of viscosity, which demonstrates the potential of this new compound as a microviscosity probe at the cellular level. Other factors and quantitative analysis may be considered in future studies.
The new probe, with absorption in the red and emission in the far-red, and viscosity-dependent fluorescence without confounding aggregation or polarity effects, as reported for previous probes [25], is promising for cell culture studies to study the dynamics of cell mitosis where various stages of mitosis are characterized by accompanying changes in viscosity. Red excitation is much less phototoxic than short wavelength visible or UV, thus, should provide valuable information. Additionally, this probe may be useful in the study of mucociliary transport and the dynamics of mucus formation important in various respiratory diseases, such as asthma and cystic fibrosis, in which alterations of the viscoelastic properties of mucus exerts a significant influence on organ function and disease development [29]. Probing the viscosity of mucus not only in cell culture, but also in vivo, via bronchoscopy, is particularly intriguing due to the ready accessibility of the lung surface with a red-absorbing and far-red emitting probe. To translate the use of this probe in vivo, two-photon excitation may be possible with excitation at ca. 1000 nm, an aspect that may be the subject of future studies as deep tissue imaging of up to 1.6 mm has been reported with two-photon fluorescent probes in muscle tissue vasculature [32]. Thus, the newly designed deoxyribonucleoside-modified cyanine dye is a promising candidate as a far-red viscosity sensor for bioimaging.