Ultrafast Excited-State Dynamics of Thiazole Orange

Abstract

Thiazole orange (TO), an asymmetric cyanine dye, has been widely used in biomolecular detection and imaging of DNA/ RNA in gels, due to its unique fluorogenic behavior: fluorescence of free dye in aqueous solution is very weak but emission can be significantly enhanced in nucleic-acid-bound dye. Herein we describe the ultrafast excited-state dynamics of free TO in aqueous solution by exploiting both a femtosecond upconversion spectrophotofluorometer and a picosecond time-correlated single-photon counting (TCSPC) apparatus. For the first time, the fluorescence lifetime of TO monomer in water was found to be ∼1 ps, mixed with concurrent solvent relaxation (which was confirmed by the experimental results of TO in DMSO). Even at moderate concentration, this lifetime has an amplitude (a measure of molecular fraction) that significantly dominates other lifetimes, and this is the origin of weak steady state fluorescence of free TO in water. We also found a novel slower decay component around 34 ps. Interestingly and in addition, the lifetime component on the 30–40 ps timescale was also found in TO-γ-Cyclodextrin (CD) complexes. The fraction of this component increased with the addition of γ-CD. Cyclodextrin has been reported to promote the aggregation of TO. Thus, although a very coincidental match of this time constant by one for a torsional process within the cavity can not be ruled out, we ascribe the shared 30–40 ps component to the lifetime of a highly quenched TO dimer experiencing intra-and inter-molecular rearrangement.

1. INTRODUCTION

Thiazole orange (TO), one of the asymmetric cyanine dyes (shown in Figure 1), is very weakly fluorescent (quantum yield ~0.0002). This has been presciently rationalized in aqueous solution as due to nonradiative photoisomerization combined with the free torsional motion between the benzothiazole and the quinolone heterocycles¹. When binding to DNA/RNA, the intramolecular rotation axis is somehow restricted, leading to a very large enhancement of fluorescence and high quantum yield². Further studies showed TO has a higher binding affinity to G-quadruplexes (G4s) compared to other forms of DNA. This unique characteristic allows one to distinguish G4 from double-stranded DNA³. However, the poor selectivity towards G4 severely limits its applications⁴. Recently, much effort has been taken to improve TO selectivity for G4s by introducing different modified functional groups⁵. More important, RNA aptamers based on thiazole orange derivatives have attracted much attention in recent years⁶. Dolgosheina and coworkers report the RNA Mango aptamer which binds thiazole orange dyes with nanomolar affinity (K_D = 3.2nM) and enhances its fluorescence up to 1,100-fold, making it valuable fluorogenic probe to track and visualize RNA in living cells⁷.

Figure 1.

Chemical structure of thiazole orange

Self-association in aqueous solutions is another important character of TO. Free TO in aqueous solution has three characteristic absorption bands (based on a progression with increasing concentration) at 500, 470 and 430 nm, apparently corresponding to monomer, dimer and higher order aggregates, respectively^{2c, 8}. Biancardi and coworkers studied the steady state spectra of TO under different conditions (temperature, concentration, solvent viscosity)⁹. Their results demonstrated that the self-association of TO was sensitive to the solvent environment. It is also well known that TO can be forced to aggregate in organized assemblies or cavitand macrocycles, such as surfactants^8c, cyclodextrins, etc. Khurana et al. studied the ns-resolved fluorescence kinetics of TO in aqueous solution at different concentrations of sulfobutylether β-cyclodextrin (SBE₇βCD) using the time-correlated single photon counting (TCSPC) instrument¹⁰. They found the fluorescence decay of TO (3 μM) was clearly nonexponential. Properly fitting the decay profile results in three lifetime components of 40 ps, ~900 ps and ~3 ns, which they ascribed to monomer, dimer and aggregates, respectively¹⁰. However, the ultrafast component could be incorrectly recovered based on the low time resolution (~>40 ps) of any TCSPC instrument.

Surprisingly, the ultrafast dynamics of TO has, to date, been seldom investigated by the methods of modern femtosecond time-resolved spectroscopy. Based on the fluorescence quantum yields of the free and bound forms and assuming they share the same radiative lifetime, the excited-state lifetime of the asymmetric cyanine dye monomer had been predicted to be about 1–5 ps¹¹. Classic works by Karunakaran and coworkers reported that the lifetime of TO monomer is less than 4 ps in methanol solution using both transient absorbance and a broadband femtosecond fluorescence up-conversion spectrophotofluorometer^1b. Furstenberg et al. systematically studied the ultrafast fluorescence dynamics of the related oxazole yellow (YO) derivatives in free and DNA-bound forms¹². An ultrafast lifetime of several picoseconds was directly measured and was assigned to the monomers of free YO derivatives. In addition, they found a component with a time constant of 30–50 ps in both free and DNA-bound forms, but the detailed mechanism of this ultrafast component remains unclear.

To the best of our knowledge, the femtosecond time-resolved spectra of TO in neat aqueous solution has not been reported. We report here a detailed study of the ultrafast dynamics of TO and TO-γ-CD complexes in water with femtosecond resolution. For the first time, the lifetime of TO monomer in water was found to be ~1 ps, which is in the same range as water dipolar relaxation. Interestingly, an additional ultrafast decay component with a time constant of 30–40 ps was also found in both TO and TO-γ-CD complexes. The fraction of this component increased with addition of γ-CD. While a coincidental 30–40ps torsionally restricted isomerization of monomers within β-CD can’t be formally excluded, we currently ascribe the 30–40 ps component to the lifetime of highly quenched TO dimers (since cyclodextrin has been reported to promote the aggregation of TO). This is further supported by the fact that in dimethyl-sulfoxide (DMSO), where shielding-induced aggregation is not operative, the ~30 ps component was not present.

2. EXPERIMENTAL SECTION

Upconversion Spectrophotofluorometer.

The general experimental setup has been described in detail elsewhere¹³. An ultrafast Ti: sapphire regenerative amplifier (Astrella, Coherent Inc.) was used to generate a fundamental near infrared pulse with an average power of 7 W. This output pulse centered at 800 nm had a duration of ~90 fs and a repetition rate of 1 KHz. For ultrafast fluorescence up-conversion measurements, the excitation pulse centered at 470 nm was generated from a portion of the fundamental beam with an optical parametric amplifier (OPerA Solo, Coherent Inc.). To avoid photodegradation or hole burning, the excitation pulse average power was carefully attenuated to ~0.1 mW (100 nJ). The sample was held in a UV quartz disk-shaped cuvette whose diameter was 80 mm and this disk was continuously spun (tangential velocity of ~5 m/s) to avoid overexposure at the sample spot. The fluorescence was collected by a pair of parabolic mirrors and focused into a 0.2 mm thick BBO mixing crystal, together with a delayed fundamental near infrared beam (800 nm) as the gate pulse. The upconverted signal, while the fluorescence wavelength was in the range of 510–670 nm, was obtained via type I sum frequency generation and a noncollinear geometry between the fluorescence and gate pulse, in order to avoid various confounding background signals (unmixed fluorescence, gate pulse or remnants of 470 nm beam). The upconverted signal was resolved in a monochromator (Omnik500, Zolix), and then detected by a relatively slow photomultiplier tube (CR317, Hamamatsu). In order to eliminate the influence from fluorescence anisotropy, the magic-angle (54.7°) condition between excitation and emission was used throughout the experiment. The instrument response function was measured to be 450 fs and was derived from cross-correlation between Raman scattering (near 560 nm) of water and the gate pulse.

Time-Correlated Single-Photon Counting (TCSPC) Apparatus.

Fluorescence lifetime on the nanosecond time scale was measured on a home-built time-correlated single photon counting apparatus. Excitation pulses centered at 470 nm were generated by a picosecond super-continuum fiber laser (SC400-pp-4, Fianium, UK) with a repetition rate of 20 MHz. The fluorescence was recorded by a stand-alone time-correlated single-photon counting (TCSPC) module (PicoHarp 300, PicoQuant) and a single-photon counting photomultiplier (PMA165A-N-M, PicoQuant). The instrument response function was measured to be ∼190 ps from the Rayleigh scattering of SiO₂ nanoparticles in water. The details of this experimental setup have been described previously¹⁴.

The steady-state absorption and fluorescence spectra were measured with a UV/Vis spectrophotometer (TU1901, Beijing Purkinje General Instrument Co. Ltd.) and a FluoroMax-4 spectrofluorometer (Horiba, Jobin Yvon), respectively.

Samples.

All compounds were commercially available and used without further purification. Thiazole Orange (TO) and DMSO were purchased from Sigma-Aldrich and γ-CD was purchased from Aladdin (Shanghai, China). Stock solutions of TO and TO-γ-CD in deionized water were prepared and stored in the dark. A stock solution (200 μM) of TO was prepared in DMSO. All samples were prepared fresh from the stock solutions under dim light conditions. For steady state measurements, the dye concentration was in the range of 5–140 μM.

3. RESULTS AND DISCUSSION

3.1. Steady-state spectra

It has been reported that the self-association characteristic of TO in aqueous solution leads to an equilibrium of monomer, dimer and aggregates^8–10. At relatively high concentration, the absorption spectrum of TO in aqueous solution usually shows a peak at ~500 nm, a shoulder at 470 nm and a weaker band at 430 nm corresponding to the monomer, dimer and higher aggregate form of the dye, respectively^{8a, 15}. The absorption spectra of TO at various dye concentrations have been shown in Figure 2a. As it can be seen, TO shows a single absorption band at 500 nm at lower concentration. With a rise of TO concentration, a ‘shoulder’ band at ~472 nm gradually appears. Inset of Fig.2a shows the ratio of absorbance at the 472 nm peak to that at the 500 nm peak. Clearly, the ratio shows a monotonic increasing trend with TO concentration, indicating the transformation from monomer to dimer at higher TO concentration.

Figure 2.

(a) Absorption and (b) fluorescence spectra of TO in aqueous solution at different concentrations. The inset shows the absorbance ratio of A₄₇₂/A₅₀₀ (a) and the total fluorescence intensity (b) vs the increasing TO concentrations. Excitation: 470 nm.

Figure 2b shows the corresponding fluorescence emission spectra of TO at different concentrations excited at 470 nm. At low concentration (~5 μM), TO appears to yield negligible fluorescence. With increasing concentration of TO, a very broad emission band covering from 520 nm to over 800 nm with a peak at 665 nm is clearly observed. Corresponding to the absorption bands, monomeric, dimeric and higher aggregate forms of TO in aqueous solution also display their intrinsic fluorescence at around 525, 600 and 650nm, respectively^{8a, 12, 15}. Inset of Fig.2b shows the variation in the total fluorescence intensity with the concentrations of TO. The total fluorescence intensity (mainly corresponding to the dimer and aggregate emission) is a linear function of TO concentration, which is also indicative of the formation of TO dimers/aggregates from low yield monomers. According to Kasha’s exciton theory, the observation of fluorescence enhancement from TO dimer and aggregate is abnormal (the fluorescence should normally be quenched for H-type aggregates). This could result from the slightly twisting between the transition dipoles of the TO molecules which have formed H-aggregates, and the rigidification of the TO aggregate which also reduced the nonradiative decay channel for the TO excited state^8b. As discussed below, the monomer, dimer and aggregates of TO exhibit very different spectral shape, and spectral mixing could be the major reason for the very broad emission band. The shape, however, is nearly concentration invariant, which suggests that some components in the mixture have very low yield.

3.2. Picosecond-resolved fluorescence spectra

The ps-resolved fluorescence decay profiles of TO (140 μM) in water at different emission wavelengths from 510 to 690 nm in TCSPC have been collected, and several typical decay curves are shown in Figure 3a. As it can be seen, these decay curves are apparently multiexponential. Further, it is apparent that the decay behavior depends strongly on wavelength, reflecting the clearly different spectra shapes. The decay data were globally fitted using a tri-exponential decay model, resulting in three lifetime components of 60 ps, 0.4 ns and 2.6 ns, having an average lifetime of 1.65 ns. The shortest lifetime 60 ps is poorly resolved since it is much shorter than the IRF (~190 ps) of our TCSPC setup. The previous research has proved that two longer components might be from the lifetime of dimers (0.4 ns) and aggregates (2.6 ns), respectively^{10, 15}.

Figure 3.

Fluorescence decay traces of 140 μM (a) and 75 μM (b) TO in water and γ-CD (c) at different wavelengths.

Further decreasing the concentration of TO to 75 μM, we repeated the above experiment and fitted the decay traces presented in Figure 3b. The lifetime of dimers decreases from 0.4 ns to 0.2 ns along with the reduction in the one of aggregates from 2.6 ns to 2.1 ns. The average lifetime of two longer components decreases significantly from 1.65 ns to 0.7 ns. Therefore, these changes suggest more than one aggregate species is present but unresolved when TO concentration increased from 75 μM to 140 μM.

It has been reported that cavitand macrocycles, such as surfactants and cyclodextrins, can promote the aggregation of TO¹⁰. To confirm it and for comparison, we have studied the effect of the α-CD, β-CD and γ-CD on TO solution. Herein, we only showed picosecond-resolved fluorescence decay of TO (140 μM) in γ-CD as in Figure 3c. The γ-CD concentration was 140 mM because of low binding constant of complex formation processes (the detailed concentration study for TO and γ-CD not shown). Global fitting here also extracts three lifetime components of 120 ps, 0.65 ns and 3.2 ns (average lifetime is 2 ns), which is in accordance with previous reports^8c,10,15. It appears both dimer and aggregate lifetimes became longer when γ-CD was added, suggesting both TO dimer and aggregates could be encapsulated in the cyclodextrin, where more restricted intramolecular torsional motion yields increased lifetimes, along with secondary environmental (e.g. polarity) influences.

3.3. Ultrafast Fluorescence Dynamics

Since the shortest lifetime cannot be correctly recovered with limited TCSPC time resolution, the fluorescence upconversion traces of TO and TO-γ-CD complex in water were collected on the 350 fs - 50 ps time scale at different emission wavelengths (510 to 670 nm). As shown in Figure 4, the decay curves are obviously multiexponential and wavelength-dependent for both 140 μM TO (similar results of 75 μM TO not shown) and TO-γ-CD complex. At least 3–4 lifetime components were needed to reproduce the experimental data. An ultrafast lifetime (0.34 ps in free TO, 0.51 ps in TO-γ-CD complex) corresponding to the Gaussian component, was required to accurately describe the initial rise of the fluorescence intensity^12a, which needs to be determined by a more precise instrument (our noncollinear IRF is ~0.5 ps).

Figure 4.

Upconversion fluorescence decay of 140 μΜ TO (a) and TO-γ-CD (b) in water at different emission wavelengths in 0–50 ps time windows.

In order to fully resolve the spectra shape and relative proportion of different forms of TO, we constructed the decay associated spectra (DAS) by properly normalizing these decay profiles to steady state emission spectra. DAS has been widely used to distinguish the relative contribution of ground state heterogeneity versus solvent relaxation. Positive DAS at all emission wavelengths are indicative of simple decay, while DAS with the “positive blue, negative red” are characteristic of excited state reactions such as solvent relaxation¹⁶.

The properly normalized fs DAS are shown in Figure 5. The longer lifetime components (1.65 ns, 0.7 ns and 2 ns for 140 μM TO, 75 μM TO and TO-γ-CD complex, respectively) certainly represent the mean lifetime of TO dimer or aggregate. The normalization to the long time scales is not very sensitive to the absolute values, so a single mean lifetime term will adequately represent the biexponential at short times. A component with a time constant of 1–4 ps was observed for both TO and TO-γ-CD complex. The amplitude of this component is positive at all measured wavelengths, indicating ground state heterogeneity dominates here. We ascribe this ultrafast component to the lifetime of TO monomer, though this term could be mixed or distorted by superimposed water relaxation (known to be in 1–2ps range) amplitude contributions. The excited state of free TO is very short-lived because of a nonradiative process associated with a large amplitude intramolecular twisting motion. When γ-CD was added, this lifetime constant increased from ~1.1 ps to 3.3 ps, which means the TO monomer encapsulated in the hydrophobic cavity of cyclodextrin experiences effective viscosity or restrictions slowing the rate of torsional motion. It is critical to note that the short TO monomeric lifetime presents the most significant contribution of amplitude, than other components even at this concentration, which leads to TO effectively being nonfluorescent in aqueous solution. Once encapsulated in the cavity or intercalated with nucleobases, this amplitude (a measure of molecular fraction that is “dark”) decreases from ~90% (in water) to ~75% (in γ-CD) and ~15% (in DNA, data shown in other work), causing a substantial fluorescence enhancement.

Figure 5.

Decay associated spectra of 140 μΜ (a) and 75 μΜ (b) free TO in water and in γ-CD (c) extracted from upconversion data.

A new additional decay component, with a nearly identical time constant of 34 or 40 ps, was also found for both TO and TO-γ-CD complex, respectively. The DAS of this component are also positive at all investigated wavelengths and similar to the steady-state emission spectrum. Most importantly, the fraction of this component decreased from ~6% to ~2.5% at lower TO concentration, and the fraction of this component increased from ~6% to ~20% with the addition of γ-CD. Cyclodextrin has been shown to promote the aggregation of TO¹⁰. This component is likely associated with early aggregates, most likely from TO dimer. Alternatively, since the original bluest state of TO in MeOH was also near 570 nm, the path to the conic intersection described in previous work^1b could be very coincidentally slowed to 30–40 ps in the cavitand environment.

It is known that self-association is one of the features of TO, especially in aqueous solutions. The feature of water as the most favorable solvent for aggregation of TO is no doubt associated with its high dielectric constant reducing the repulsive force between the similarly charged dye cations or anions in the aggregate¹⁷. However, in comparison to the water, the dielectric constant of DMSO is low, obviating aggregation. Therefore, to gain further insight the origin of this 30–40 ps component, the ultrafast dynamics of TO in DMSO have been performed. According to the absorption spectra of TO at various dye concentrations in DMSO (data not shown), the ratio of absorbance at the 472 nm peak to that at the 500 nm peak remains constant, which implies TO all appears in the monomer form.

The ultrafast fluorescence dynamics of free TO in DMSO was investigated by fluorescence upconversion and monitored in the same way as above (Figure 6a). Global fitting results in two lifetime components of 0.67 ps and 2.7 ps with accompanying DAS. The first component (0.67 ps) could arise from solvation alone, since amplitude is positive at short wavelengths and crosses to negative when going to the red. The second component (2.7 ps) can best be ascribed to the lifetime of TO monomer. The amplitude of this component was always positive and matched rather closely the very weak steady-state emission spectrum (Figure 6b), indicating that this component was associated with the decay of the excited-state population of the free dyes.

Figure 6.

(a) Decay associated spectra of TO in DMSO extracted from upconversion data. (b) Absorption and emission spectra of TO in DMSO.

Our upconversion measurements, performed with the required sub-picosecond resolution, indeed confirm that a 34 ps component was not present under such conditions. According to these experimental results, the origin of the 30–50 ps component at least in neat water, and possibly also in γ-CD could be understood from the higher rigidity of TO dimer in comparison to the monomer. Therefore, both the component of 0.4 ns (mentioned above) and 34 ps might be dimer lifetimes due to the different structures. TDDFT studies a decade ago⁹ suggested 2–3 differing dimeric structures (displayed in figure 7) with roughly similar stokes’ shifts and oscillator strengths. However, there is no measurement of known exciton lifetime for us to use as a guide, we could only speculate about the quenched H-dimer (34 ps) in a large twist angle that would experience a cooperative intramolecular torsional motion an order of magnitude faster than the H-dimer (0.4 ns). The nonradiative decay rate ranking of these putative structures would require an extensive QM-MM study (beyond the scope of this work), and even more involved minimization and dynamics studies would be required to explain the higher order multimers exhibiting the much longer ns components (2.6 ns). Future studies should employ other loose cavitand and nucleic acid hosts, and polarized experiments seeking evidence of exciton coupling and/or internal rotation may also be of use.

Figure 7.

Molecular structures of TO monomer and various TO dimers in water⁹.

4. CONCLUSIONS

The ultrafast dynamics of Thiazole Orange in the biologically relevant solvent water has been dissected for the first time, and the complexities of aqueous aggregation have been assessed by comparing dynamics in a cavitand (encouraging aggregation) and in DMSO (discouraging charge-shielding and aggregation). The long anticipated picosecond decay of the monomer internal isomerization has been seen, but order of magnitude slower cooperative processes in dimers yield 30–40 ps lifetimes as well. These are also present in the cavitand, known to encourage dimers. The presence at higher concentrations of even higher TO aggregates and their nanosecond shared states presents a different challenge that future studies encompassing QM-MM simulation and exciton coupling must address.