Solvatochromic fluorene-linked nucleoside and DNA as color-changing fluorescent probes for sensing interactions

Color-changing fluorescent nucleotide and oligonucleotide probes for studying interactions with other biomolecules were designed and prepared, and perform better than currently known environment-sensitive fluorophores.


Introduction
Fluorescence spectroscopy in combination with environmentsensitive uorescent dyes is the method of choice for studying biomolecular interactions. 1 Fluorophores containing push-pull electron-donating and electron-withdrawing groups attached to a polarizable p-electronic system typically exert solvatochromic properties. They undergo intramolecular charge transfer (ICT) in the excited state resulting in a highly increased dipole moment of the molecule. Fluorescence characteristics of solvatochromic dyes (emission maxima, uorescence quantum yield, and uorescence lifetime) strongly depend on the microenvironment mainly due to the dipole-dipole interactions occurring in the excited state. 1 Due to that, they can sense the local variations of polarity in proteins, lipid membranes and cellular organelles, 1-4 which are oen associated with certain biological events. However, to fully reveal the potential of solvatochromic dyes as biophysical probes, they must be siteselectively incorporated into the biomolecules of interest.
Diverse environment-sensitive uorescent nucleoside analogues (FNAs) and their corresponding oligonucleotide probes have been used to study a variety of molecular processes in which nucleic acids (DNA or RNA) are involved. Apart from 2-aminopurine and other FNAs which respond to changes in the microenvironment by changing uorescence intensity, 5,6 solvatochromic uorophores provide changes in emission maxima (emission color) as an additional source of information. Particularly, nucleosides with a tethered 6-propionyl-2-(dimethylamino)-naphthalene (PRODAN, Chart 1) moiety 7 have been used to monitor microenvironmental changes in the major and minor grooves of B and Z-DNA 8 and for the detection of single nucleotide polymorphism (SNP). 9 Nucleoside analogs with the nucleobase replaced by solvatochromic uorophores, such as 4-aminophthalimide 10 or Nile red, 11 have been used to probe the polarity within the DNA double helix. A series of isosteric solvatochromic uorescent analogs of purines and pyrimidines which report on changes in local polarity while not perturbing the stability of nucleic acids have been developed by the Tor group. 12 Saito and co-workers have developed a wide variety of solvatochromic FNAs based on substituted purines and diverse aza-and/or deaza-analogues of purine bases which probe DNA hybridization. 13 Ultra-sensitive environment-sensitive nucleosides based on the 3-hydroxychromone uorophore exhibiting a two-band ratiometric uorescence response have been reported to probe DNA hybridization and DNA-protein interactions. 14 Finally, our group has reported solvatochromic nucleoside analogues and their corresponding nucleoside triphosphates (dNTPs) as building blocks for the enzymatic synthesis of DNA probes for hybridization and DNA-protein interactions. 15 An ideal solvatochromic uorescence label for DNA should meet the following criteria: (i) it must show high sensitivity to the microenvironment, resulting in a large uorescence red shi in highly polar media; (ii) it must retain high quantum yield in polar protic solvents, especially water; (iii) it must be sufficiently stable under physiological conditions; (iv) it must be localized at a site on DNA where the changes in local polarity are the most significant; (v) the corresponding 2 0 -deoxyribonucleoside 5 0 -O-triphosphate (dNTP) must be a substrate for DNA polymerases to allow for the enzymatic preparation of long labelled DNA and to be potentially suitable for labeling DNA in vivo. To the best of our knowledge, none of the existing solvatochromic nucleoside analogs meet these criteria. The majority of them were incorporated into DNA only by the phosphoramidite chemical synthesis, 7-14 which signicantly limits the size of the oligonucleotide probes which can be prepared. Most of the known solvatochromic nucleosides show only moderate solvatochromic shi, low-to-moderate uorescence quantum yields in water. 15,16 Some other solvatochromic FNAs are chemically unstable, e.g. solvatochromic aminophtalimide nucleosides easily hydrolyze leading to the decomposition of the chromophore. 10,15b With the aim to overcome all these limitations, herein we report a new uorescent 2 0 -deoxycytidine nucleoside linked to a highly solvatochromic uorene derivative through a nonconjugated tether, its conversion to dNTP and enzymatic synthesis of color-changing DNA probes for interactions with lipids or proteins. The experimental data are complemented by quantum chemical calculations and molecular dynamics (MD) simulations of the labeled DNA molecule complexed with the p53 protein.

Synthesis
Our interest in designing a solvatochromic nucleoside analogue with improved photophysical properties and chemical stability was stimulated by the fact that a series of new PRODAN-inspired environment-sensitive dyes with superior properties were reported during the last decade. 4,17 We speculated that the uorene core-based uorophore FR0 (Chart 1) reported by Klymchenko and co-workers 17a would retain its highly favorable environment-sensing properties aer tethering to DNA via a exible linker. We proposed nucleoside dC FL (Chart 1) as the target molecule. Comparing with FR0, the uorophore features a propargyl group graed to the nitrogen atom to serve as a linker; moreover the reactive formyl group was replaced with acetyl functionality which has less electron-withdrawing properties but is far less reactive. We also note that, in the course of our research, this uorophore was independently reported as a uorescent stain for lysosomes. 4b The synthesis of the uorene-labeled nucleoside dC FL is shown in Scheme 1. At rst, we developed a robust and efficient method for the preparation of the alkyne-linked uorophore 8 building block. It was synthesized from 2-aminouorene following a convergent strategy with key issues of mono-methylation of the amino group and dimethylation of the uorene central ring in order to mask undesired reactivity and to red-shi the absorption maximum. 18 Starting from Boc-protected 2-aminouorene 2 (ref. 19), all three methyl groups were installed in a one-pot reaction using an excess of methyl iodide and potassium tert-butoxide as a base in THF, to give trimethylated compound 3 in 83% yield.
The acid-labile Boc group was then exchanged in two steps to give the more stable acetyl derivative 5 which was used in a Friedel-Cras acetylation to install the acetyl moiety onto the opposite aromatic ring. The presence of the N-acetyl group suppressed possible ortho-substitution to give exclusively the 7-acetylated uorene 6. Finally, the amide protection on the amine functionality was cleaved by hydrolysis using NaOH in a water-methanol mixture providing secondary amine 7, which was subsequently propargylated yielding the desired uorophore-linked terminal alkyne 8. In this way, the uorophore building block 8 was prepared from 2-aminouorene (1) in seven synthetic steps with four chromatographic purications in roughly 42% overall yield, which is signicantly easier and more efficient than the alternative route starting from 2-nitro-uorene published recently. 4b The uorene-linked acetylene (8) was then attached to a nucleoside through the Sonogashira cross-coupling with 5-iodo-2 0 -deoxycytidine (dC I ). The reaction was performed in the presence of PdCl 2 (PPh 3 ) 2 and CuI in DMF and Et 3 N at 45 C to give the desired uorescent nucleoside 9 (dC FL ) in 85% yield. Finally, to obtain the substrate for polymerase incorporation into DNA, the nucleoside was phosphorylated under Ludwig's conditions 20 giving the corresponding solvatochromic uorenetethered dCTP analogue 10 (dC FL TP) in 16% yield.

Photophysical properties
Absorption and steady-state uorescence spectra of uorene uorophore 8 were studied in a set of solvents of varying polarity and compared with those of nucleoside dC FL and triphosphate dC FL TP; the results are summarized in Fig. 1 and Table 1. The uorophore exhibited relatively high solvent-dependent absorption coefficients (up to 30 700 M À1 cm À1 ). The absorption spectra of uorene 8 centered within a narrow range of 357-366 nm in these solvents. The hypsochromic shi in the absorption of 8 compared to the parent compound FR0 (390-399 nm) 17a can be explained by the weaker electron-withdrawing properties of the acetyl group over the formyl group, which rendered a lower excited state dipole moment of the molecule. A strong solvent-dependent variation of emission maxima was observed for 8. The positions of the emission maxima ranged from 421 nm in dioxane to 544 nm in methanol. Hence, emission colors of the solvatochromic uorene are similar to those reported for FR0 17a and cover a signicant part of the visible spectrum, from deep-blue to reddish orange ( Fig. S1a †). More importantly, high uorescence quantum yields were observed for the whole range of studied solvents, including protic methanol which readily quenches the emission of solvatochromic dyes. 16b,21 Nucleoside dC FL exhibited similar spectroscopic properties (Table 1, Fig. 1B). The positions of absorption and emission bands were nearly identical to 8, indicating that the tethering of the nucleoside via the nonconjugate propargyl linker did not change the electronic properties of the uorophore. At the same time, uorescence quantum yields were 0.06-0.10 lower than those of 8, which can   be explained by the higher number of rotatable bonds in the nucleoside and therefore its higher conformational mobility. Nevertheless, the nucleoside retained relatively high quantum yields of uorescence even in polar protic solvents (0.64 and 0.57 for ethanol and methanol, respectively). The nucleoside triphosphate dC FL TP dissolved in aqueous buffer exhibited an absorption band of the uorophore centered at 352 nm and the reddest emission spectrum centered at 584 nm, resulting in a large Stokes shi of 232 nm (Fig. 1C). To our satisfaction, despite this large solvatochromic shi the uorophore did not suffer from excessive quenching by water. The nucleoside triphosphate exhibited high uorescence quantum yield (0.27), which to the best of our knowledge is one of the highest values ever exhibited by solvatochromic nucleoside analogs. 22 The Stokes shis of all the measured compounds ranged from 4400 cm À1 for 8 in 1,4-dioxane to 11 400 cm À1 for dC FL TP in aqueous buffer. Interestingly, the Stokes shi of the uorophore perfectly correlates with the empirical solvent polarity scale Et(30) 23 ( Fig. S1b †). Altogether, these experiments show that the solvatochromic uorene uorophore has good potential for DNA labeling and for probing DNA interactions.

Quantum chemical calculations of the emission spectra
Further insight into the origin of the color-changing behaviour of the studied nucleoside can be obtained from quantum chemical calculations. An excellent starting point for our computational study is the theoretical description of the emitting state of the structurally related PRODAN molecule (Chart 1) which identied two distinct S 1 equilibrium geometriesplanar or twisted 24, 25 and calibrated a computational TD-DFT (B3-LYP) and resolution-of-identity algebraic diagrammatic construction through second order RI-ADC(2) methods (see ESI † for Computational details). The results of the calculations of the vertical absorption spectra at the ground-state equilibrium geometry of 8 are presented in Table 2, both in the gas-phase and in the solvent (modeled by polarized continuum). Both TD-DFT and ADC(2) methods clearly elucidated the nature of the S 1 and S 2 excited states. One of the two excitations is of p / p* character, where the two orbitals involved in the excitation are delocalized over the polycyclic aromatic system (Fig. 2). This is responsible for an intense absorption peak at $365 nm observed experimentally in alcohols and computed at 375 nm employing the ADC(2) method and the 3 r ¼ 80 polarized continuum (c.f. Tables 1 and  2). The second excited state (mostly S 2 ) is of the n / p* character where the lone pair orbital is localized on the oxygen of the CO group (Fig. 2), and appears as a low-intensity peak at 310 nm Concerning the uorescence properties of 8, the TD-DFT (B3-LYP)/def2-TZVP gas-phase S 1 geometry optimizations resulted in two distinct excited state minima (Fig. 2). Their emissive properties are listed in Table 3. The two structures differ mainly by the rotation of the substituted amino-group with respect to the plane of the aromatic carbon rings, resulting in semi-planar (S 1 p) and twisted (S 1 t) structures, respectively. In the semi-planar (S 1 p) structure, the S 1 excited state is of n / p* character (vide supra and Fig. 2), while the twisted structure (S 1 t) is characterized by the p / p* S 1 transition with a charge transfer character (Fig. 2). The singlepoint polarized continuum (3 r ¼ 2.2, 38.6 and 80 to model 1,4dioxane, acetonitrile and water environments) ADC(2) calculations at the respective TD-DFT (B3-LYP/def2-TZVP) gas-phase optimized geometries were carried out to elucidate the origin of the observed Stokes shis of 8. With the increasing polarity of the environment the relative stability of the less polar S 1 p state, characterized by dipole moment m ¼ 2.1 D, decreases with respect to the S 1 t (m ¼ 22.5 D). While the S 1 p state is stabilized by 0.67 eV over the S 1 t state in dioxane, this difference is only 0.19 eV in water. The calculated emission energies of both S 1 p and S 1 t are shown in Table 3. The increasing polarity of the solvent results in a blue shi of the less polar S 1 p state, from 467 nm (2.63 eV) in 3 r ¼ 2 to 450 nm (2.73 eV) in 3 r ¼ 80. The opposite trend has been observed for the more polar S 1 t state, i.e. a red shi from 427 nm (2.90 eV) in 3 r ¼ 2 to 583 nm (2.13 eV) 3 r ¼ 80, in accordance with the experimental observations. Thus, the computed values provide us with a consistent picture of the nature of the solvatochromism observed in the emission spectra. The results indicate that both states may play a role in the emission depending on the polarity of the solvent. In less polar solvent, the emission from the S 1 p geometry might occur whereas in polar solvents the twisted conformation S 1 t is the likely representation of the emitting state.  Finally, it can be mentioned that an attempt has been made to use the qualitative model of the protein-DNA FL complex described in this study (vide infra) and calculate emission spectra in the context of DNA-bound uorene interacting with the protein. However, the potentially quantitatively accurate ADC(2) calculations turned out to be beyond current computational feasibility and TD-DFT calculations, with several explicit protein residues included which did not provide data of sufficient accuracy that would go signicantly beyond the predictions based on the polarity of the environment modelled by the implicit solvation model (vide supra).

Polymerase incorporation of uorene-linked nucleotide
Subsequently, we investigated whether the modied triphosphate dC FL TP can be used for the polymerase synthesis 26 of DNA with a graed solvatochromic uorene. We tested dC FL TP in a primer extension (PEX) reaction using a 50 nt template and 25 nt primer designed so that the product would contain two deoxycytidines (Fig. 3a). When the PEX was performed in the presence of KOD XL DNA polymerase or Bst DNA polymerase (large fragment), polyacrylamide gel electrophoresis (PAGE) (Fig. 3b) showed a clean formation of the full-length primer extension products when dC FL TP was used in combination with the three remaining natural dNTPs (lanes 3, 6). These products had similar mobility as those obtained by PEX with natural dNTPs (positive control, lanes 2, 5). Other popular DNA polymerases, namely Vent(exo-), Pwo, Phusion, 9 N m , DyNAzyme II, were unable to efficiently use dC FL TP as a substrate (data not shown). Correct incorporation of dC FL TP into DNA by KOD XL polymerase was also conrmed by PEX with a biotinylated template followed by magnetic separation with streptavidin magnetic beads followed by MALDI-TOF analysis (Fig. S2 †).
We also found out that KOD XL is able to use dC FL TP in PCR amplication yielding the labeled 98 bp-long double-stranded DNA containing as many as 34 uorene-modied deoxycytidine residues ( Fig. 3c and d). Finally, we showed that the uorenelabeled DNA can be puried from the unreacted uorenelabeled triphosphate using a widely used method based on absorption of DNA on a silica-gel membrane in the presence of chaotropic salts. 27 A primed DNA template was incubated with the modied and three natural dNTPs either in the presence or in the absence of the polymerase and then DNA was isolated from both samples using one of the commercially available DNA purication kits. The incorporation of the uorene moiety into DNA was unambiguously demonstrated for the polymerase-positive sample by UV-vis absorption and steady-state uorescence spectroscopy (Fig. S3 †). On the contrary, the uorophore was not detected in the polymerase-negative sample, indicating that the uorene-modied nucleoside triphosphate do not interact nonspecically with DNA via intercalation and can by efficiently removed using standard techniques.

DNA-protein interactions
Having a robust and facile method for the preparation of labelled oligonucleotides and DNA, we proceeded to the study of their properties. To produce 50-bp uorene-labelled dsDNA, we used semi-preparative PEX in the presence of KOD XL DNA Table 3 Emission energies (E emiss ) and relative energies (E rel ) calculated a for the S 1 state in the semi-planar (S 1 p) and twisted (S 1 t) geometries of 8 in different environments polymerase (Scheme S1 †). This DNA (referred to here as DNA FL ) bears two tethered solvatochromic uorene moieties and contains the consensus recognition site of human transcription factor p53 from our previous studies. 28 Circular dichroism (CD) spectra showed that the attachment of the uorophores did not signicantly alter the B-DNA conformation of the DNA (Fig. S4 †). The uorescence spectrum of DNA FL was centred at 581 nm, which was only a 3 nm blue-shi compared to the triphosphate in water. Such a small blue shi between the dNTP in buffer and DNA indicates only a minor shielding effect of DNA or small dielectric differences between water in the bulk and the hydration shell of DNA in the vicinity of the major groove, where the probe is located. From the calibration curve on Fig. S1b † one can estimate that the media surrounding the uorophore can be characterized by Et(30) $ 60 kcal mol À1 . This value is higher than that which was concluded in previous studies using other solvatochromic nucleosides 12c,29 and is in line with the observation that the spectroscopically measured values of polarity of the major groove depend on the nucleoside probe and linker used. 12c To explore the applicability of the solvatochromic uorene nucleoside as a colour-changing reporter of biomolecular interactions, we studied the inuence of protein binding on the uorescence properties of DNA FL . In this study we used the core domain of the human transcription factor p53 with a GST tag (p53CD_GST) expressed in bacteria as a model DNA-binding protein. 30 First, we examined whether the tethering of the bulk hydrophobic uorophore within the recognition site would not disrupt the ability of the DNA to be recognized by the protein.
An electrophoretic mobility shi assay (EMSA) with a radioactively labeled DNA FL on PAGE under native conditions showed no signicant difference in the binding of p53CD_GST between the labeled DNA FL and non-modied DNA control (Fig. S5 †). When uorescence spectra of DNA FL were measured either with or without p53CD_GST, the protein-bound DNA FL showed a blue shi in the uorescence spectrum. The emission maximum shied from 581 nm to 567 nm (Fig. 4a) indicating partial screening of the uorophore within the DNA-protein complex. Nevertheless, the value of the emission maximum was far from the values obtained for 8 in non-polar solvents, indicating that the media surrounding the uorophore remained fully hydrated. Notably, the hypsochromic shi obtained in this experiment was ca. 1.8-fold larger than that reported by Saito and co-workers who used another PRODAN-like nucleoside to probe DNA binding to the Klenow fragment of DNA polymerase I from Escherichia coli. 29 To the best of our knowledge, this is the rst example where a change of color in a labelled DNA probe upon binding to a protein was distinguishable by the naked eye (Fig. 4b).
Color-changing uorescent probes in combination with digital RGB cameras and image-processing soware are becoming important tools for practical implementation of the uorescence sensing technology. 31 We estimated the applicability of our probes in this context. Images of uorescent solutions were taken with a digital colour camera and analysed using the image-processing package ImageJ as described in the literature. 31c To our satisfaction, the changes of emission colour were suitable for quantication (Fig. S6-S8 and ESI † for details). Further, image analysis of the DNA FL and DNA FL -p53CD_GST solutions shown in Fig. 4b was performed in terms of ratiometric and hue quantication. 31c The changes in hue (H) and red/green (R/G) ratio (Fig. 4c) reecting that different microenvironment of the probe can be distinguished. Notably, the change of ratiometric signal (R/G) were superior compared to the difference in hue.
In order to further rationalize the experimentally observed color change upon interaction of the labelled DNA molecule with the protein and provide a missing link between quantum chemical calculations and experiment, an attempt was made to predict the three-dimensional structure of the complex. To this aim, the available crystal structure of the p53 + DNA complex (PDBID 3EXJ) was used as a template and modied using the YASARA modelling package. 32 The uorene molecules were covalently linked to the corresponding nucleobases in the DNA. Next, hydrogen atoms (missing in the crystal structure) were added to the protein into their standard positions with the protonation states corresponding to the neutral pH and their position were subsequently optimized. The whole structure was immersed into the solvent box (periodic boundary conditions, PBC). The system was electro-neutralized by placing the appropriate number of Na + and Cl À ions corresponding to their 0.15 M concentration. The AMBER ff03 force eld 33 and parameter set were used for the protein and DNA. Ligand 8 was optimized in a vacuum and partial charges on its atoms were obtained by a restrained t to the electrostatic potential (RESP) at the AM1BCC level 34 whereas standard gaff parameters were used for bonded and van der Waals terms of 8. Aer initial minimization, the simulated annealing protocol, i.e. repeated cycles of molecular dynamics under the PBC, with YASARA defaults 32 was applied and the resulting model was subjected to nal minimization. The full structure is given in the ESI † whereas the uorene environment is displayed in Fig. 5.
Though such a simplied model does not allow one to quantify the polarity change, the modeled structure (Fig. 5) clearly shows that the uorene labels at least partly interact with the protein. Apparently, this interaction changes the polarity of the environment around the labels from highly polar water to a less polar protein and this change is signicant enough to shi the emission maxima of the uorophores by 14 nm resulting in a naked-eye-visible color change from orange to yellow.

DNA-lipid interactions
Cationic lipids are important nucleic acid-transferring reagents widely used in cell biology and in gene therapy. 35 Therefore, we decided to examine whether the formation of lipoplexes (compact complexes of DNA and lipids) can be monitored by the uorescence of the solvatochromic uorene uorophore in order to study the inuence of cationic lipid DOTAP on the uorescence of DNA FL . We observed a signicant blue shi in the emission maximum accomplished by a change in the shape of the spectrum upon addition of DOTAP small unilamellar vesicles to DNA. The observed 33 nm blue shi was easily observable by the naked eye at sub-micromolar concentrations of DNA (Fig. 6a, c) and it was even more signicant than in the case of protein binding (p53CD_GST).
The plot of the spectral center of mass (SCM) versus DOTAP/ DNA FL charge ratio had a sigmoidal shape (Fig. 6b) with a transition at charge ratio $1, in line with our previous observations. 28b Further, we analyzed the RGB images of UV-illuminated DNA FL and DOTAP-DNA FL samples in the same way as described above for DNA-protein interactions ( Fig. 6c and d). Again, one can notice apparent changes in H and R/G ratio upon addition of DOTAP. Altogether, these experiments prove the utility of dC FL as a uorescent color-changing probe for studies of DNA interactions in vitro.

Time-dependent uorescence shi (TDFS) assay
In contradiction to the above shown steady-state data, the timedependent uorescence shi assay provides direct and simultaneous information on hydration and mobility of the intimate dye's solvent shell. 2c To that end, we performed time-resolved emission spectroscopy (TRES) of dC FL and thus monitored the so-called solvent relaxation process at the corresponding position within the DNA molecule. Previous investigations of DNA solvation dynamics were performed with probes placed inside the duplex-DNA either by covalent attachment or by non-covalent minor groove binding. 36 The characterized dynamics in DNA were found to be nonexponential and extended into subpicosecond to nanoseconds timescales, best described by a power-law relaxation. 37 There is no unied explanation of such dispersed dynamics primarily because of the complicated coupling between motions of components localized in the closest (<1 nm) vicinity of the probe: water molecules, segments of the DNA molecule as well as cations. However, for an interpretation one should keep in mind that the solvent relaxation response of bulk water occurs on a sub-picosecond time scale 38 and slower responses (i.e. on the ps and ns time-scale) have to be attributed to DNA dynamics. 39 In this work, we measured temporal evolution of the emission maximum of the labelled DNA FL with dC FL exposed to the major groove of DNA in several microenvironments. The experimentally observed transient Stokes for DNA FL in buffer, buffer-glycerol mixture (1 : 1) and DNA FL -p53CD_GST complex are 70 cm À1 , 130 cm À1 , and 40 cm À1 , respectively. These values are negligible compared to the total amount of Stokes shi of 5320 cm À1 , 5020 cm À1 , and 5120 cm À1 , respectively, obtained  by the "time-zero" estimation (Fig. 7). These ndings indicate that the nano-vicinity of the probe is different to the probe in plain water (estimated total amount of Stokes shi of dC FL TP is about 5600 cm À1 ), but highly hydrated. The signicantly smaller Stokes shi for the DNA FL -p53CD_GST complex indicates a less hydrated probe environment when compared to the free DNA molecule. As practically the entire solvation dynamics are occurring on a time-scale faster than that of the resolution of the experiment (i.e. faster than 40 ps), the ensemble solvent dynamics are due to fast segmental motions of the DNA with possibly some contribution of bulk water. These observations are comparable with literature reported dynamic Stokes shis: independent of chromophore (i.e. coumarin 102, 4 0 ,6-diamidino-2-phenylindole, Hoechst 33258 and 2-hydroxy-7-nitro-uorene (HNF)) and way of labelling (nucleobase replacement or non-covalent minor groove-binding), the largest part of the transient Stokes shi occurs on a time scale faster than 40 ps. 37,40 However, in all these cases, except for HNF, a minor, but signicant nanosecond component is present. 36,41 A signicantly different picture was observed for DNA FL incorporated into the DNA FL /DOTAP lipoplexes. The total amount of the Stokes shi decreased to 4650 cm À1 . Even more strikingly, the experimentally observed transient Stokes shis increased to 65% and 72% of the total amount of the Stokes shis in the cases of 1 : 1 DNA FL /DOTAP charge ratio and 1 : 4 DNA FL /DOTAP charge ratio, respectively (Fig. 7). In this case, almost all solvation dynamics occurred on the nanosecond time scale. Upon addition of DOTAP to DNA FL accompanied by lipoplex formation between DNA FL and DOTAP, the nano-vicinity of DNA FL became signicantly less hydrated compared to DNA FL alone in the buffer and the microenvironment mobility decreased dramatically. Addition of more DOTAP to DNA FL / DOTAP lipoplexes resulted in a further decrease of the slower relaxation time (s R ) while the hydration stayed at almost the same level. This observation suggests that the lipoplex structure is getting more compact upon addition of the DOTAP, which is in line with previously reported results, where the experiments with the addition of DOTAP resulted in more compressed lipoplexes and growth in their size. 28b As the time-scale of the TDFS is comparable to probes located at the external interface of a lipid, 42 we speculate that the dC FL dye incorporated in the DNA FL molecule is probing mainly the hydrated positively charged DOTAP head-groups which are in close contact with DNA FL incorporated into the DNA FL /DOTAP lipoplexes. We are now performing more experiments to explain the DNA FL / DOTAP dynamics in detail.
Notably, such a nanosecond TDFS has not been described in a DNA system. Together with the large total Stokes shi, we believe that dC FL has good potential as a uorescence probe in applications in the TDFS solvation studies of DNA. Furthermore, lipoplex investigation is of interest for the gene transfer to the cells. Studying DNA FL lipoplexes could reveal not only DNA dynamics themselves but also the processes involved in DNA delivery into cells. 43

Conclusions
To conclude, we developed a new solvatochromic nucleoside (dC FL ) featuring a unique combination of properties: (i) its triphosphate (dC FL TP) is recognized as a substrate by DNA polymerases in primer extension and PCR; (ii) it does not suffer from hydrolytic instability in aqueous solution; (iii) it exhibits exceptional solvatochromicity, namely the shi in emission band from 421 nm in dioxane to 584 nm in water, covering nearly the whole visible spectrum; (iv) the dye retains a relatively high quantum yield of uorescence in water. The changes in emission color of the labeled DNA upon interactions with DNA-binding protein p53 and lipid are quite pronounced and detectable not only by uorescence spectroscopy but also by digital photography with image analysis and by naked eye inspection. To the best of our knowledge, this is the rst nakedeye-visible sensor to study DNA protein interactions and the push-pull uorene-based solvatochromic uorophore performs better than environment-sensitive uorophores previously used for the labeling of nucleic acids. [12][13][14][15][16] The physico-chemical origins of the observed solvatochromism were rationalized by DFT and ADC(2) calculations correlated with the computational model of the DNA + protein complex which show that, depending on the polarity of the solvent, emission either from the planar or twisted conformation of the excited state occurs resulting in a change of the emission wavelength. Finally, the modied nucleoside was shown to be suitable for probing site-specic dynamics and hydration of DNA using TDFS measurements. The combination of these properties makes this probe a prospective versatile tool for investigating DNA interactions in biophysical and bioanalytical studies.