Controlling the fluorescence of ordinary oxazine dyes for single-molecule switching and superresolution microscopy

Abstract

Fluorescent molecular switches have widespread potential for use as sensors, material applications in electro-optical data storages and displays, and superresolution fluorescence microscopy. We demonstrate that adjustment of fluorophore properties and environmental conditions allows the use of ordinary fluorescent dyes as efficient single-molecule switches that report sensitively on their local redox condition. Adding or removing reductant or oxidant, switches the fluorescence of oxazine dyes between stable fluorescent and nonfluorescent states. At low oxygen concentrations, the off-state that we ascribe to a radical anion is thermally stable with a lifetime in the minutes range. The molecular switches show a remarkable reliability with intriguing fatigue resistance at the single-molecule level: Depending on the switching rate, between 400 and 3,000 switching cycles are observed before irreversible photodestruction occurs. A detailed picture of the underlying photoinduced and redox reactions is elaborated. In the presence of both reductant and oxidant, continuous switching is manifested by "blinking" with independently controllable on- and off-state lifetimes in both deoxygenated and oxygenated environments. This "continuous switching mode" is advantageously used for imaging actin filament and actin filament bundles in fixed cells with subdiffraction-limited resolution.

Fig. 1.

Photoinduced processes of the fluorophore ATT0655. (A) Scheme. After excitation to the first excited singlet state S₁, fluorescence is emitted at rate k_Fl (neglecting nonradiative processes). Intersystem crossing competes with fluorescence and leads to the infrequent formation of triplet states T₁ with the rate constant k_Isc. The triplet state is depopulated by intersystem crossing with rate constant k_T or by electron transfer. This may occur either through oxidation by methylviologen (MV) forming a radical cation F^•+ or through reduction by AA yielding a radical anion F^•−. The radical ions could be recovered to singlet ground-state fluorophores by the complementary redox reaction. Direct electron transfer from the singlet manifold (k_RedS, k_OxS) has to be taken into account at higher redox-agent concentration. (B and C) Fluorescence transients of ATTO655-labeled dsDNA immobilized in aqueous environment after oxygen removal, addition of 100 μM AA (B) and additionally 100 μM MV (C), respectively. The transients are binned in 10 ms. Samples were excited at 640 nm with an average excitation intensity of 1.5 kW/cm².

Fig. 2.

Confocal fluorescence images (6 × 6 μm) of ATTO655-Cy3B-labeled dsDNA. The colors green and red encode for the overall fluorescence intensity after green and red excitation, respectively. dsDNA bearing the 2 dyes are hence visible as yellow spots. (A) The first image was acquired in PBS. (B, D, F) The second scan shows the same part of the cover slide after the buffer was deaerated and 100 μM AA was added. (C, E, G) Rinsing of the surface and refilling the chamber with standard PBS switches the molecules back into their fluorescent form. This cycle of reversible redox-switching of single ATTO655-molecules was repeated several times, 3 cycles are shown in A–G.

Fig. 3.

Dependence of the on- and off-state lifetime τ_on and τ_off on the concentration of the reductant AA and the oxidant MV in the absence and presence of oxygen. (A and D) AA dependence of switching at constant 100 μM MV concentration. (B and E) MV dependence of switching at constant 100 μM AA concentration. (C and F) The on- and off-times with simultaneous change of MV and AA. The first row shows data measured without oxygen, whereas all data shown in the second row were measured in the presence of oxygen. The error bars indicate the standard deviation from the mean. The data were fitted by using an allometric function (f(x) = a·x^b) to guide the eye.

Fig. 4.

Plot of the detected photons during an on-cycle, on-counts N_on, obtained from the fluorescence transients, versus the AA concentration. The error bars indicate the standard deviation from the mean. A fit using Eq. 2 is shown as black solid line.

Fig. 5.

Total internal reflection fluorescence microscopy of ATTO655-phalloidin-labeled single actin filaments (A and B) and bundled actin filaments in fixed NIH/3T3 cells (D and E). (A and D) TIRF microscopy images and (B and E) Blinkmicroscopy images with subdiffraction resolution. (C and F) Histograms over the regions marked with a white rectangle in B and E and the corresponding peak to peak distances derived from Gaussian fits. The images were recorded with reductant and oxidant concentrations optimized for imaging speed, fluorophore density, and excitation intensity (see SI Text for details).