Multicolor super-resolution imaging with photo-switchable fluorescent probes

Abstract

Recent advances in far-field optical nanoscopy have enabled fluorescence imaging with a spatial resolution of 20 to 50 nanometers. Multicolor super-resolution imaging, however, remains a challenging task. Here, we introduce a family of photo-switchable fluorescent probes and demonstrate multicolor stochastic optical reconstruction microscopy (STORM). Each probe consists of a photo-switchable "reporter" fluorophore that can be cycled between fluorescent and dark states, and an "activator" that facilitates photo-activation of the reporter. Combinatorial pairing of reporters and activators allows the creation of probes with many distinct colors. Iterative, color-specific activation of sparse subsets of these probes allows their localization with nanometer accuracy, enabling the construction of a super-resolution STORM image. Using this approach, we demonstrate multicolor imaging of DNA model samples and mammalian cells with 20- to 30-nanometer resolution. This technique will facilitate direct visualization of molecular interactions at the nanometer scale.

Fig. 1

Photo-switchable probes constructed from activator-reporter pairs. (A) Spectrally distinct reporters exhibit photo-switching behavior. The lower panel shows the fluorescence time traces of three photo-switchable reporters, Cy5 (dark yellow line), Cy5.5 (red line), and Cy7 (brown line) when paired with a Cy3 dye as the activator on a DNA construct. Traces were shifted relative to each other for clarity. The upper panel shows the green laser pulses (532 nm) used to activate the reporters. The red laser (657 nm) was continuously on, serving to excite fluorescence from the reporters and to switch them off to the dark state. (B) Switching rate constants k_on and k_off of the Cy3-Cy5, Cy3-Cy5.5, and Cy3-Cy7 pairs as a function of green and red laser power. Error bars indicate SEM from ∼ 3 data sets. We note that the laser power to intensity calibration may vary between different samples due to moderate differences in the laser spot size at the sample. (C) The same reporter can be activated by spectrally distinct activators. The lower panel shows the fluorescence time traces of Cy5 paired with three different activators, Alexa 405 (magenta line), Cy2 (blue line) and Cy3 (green line). The upper panel shows the violet (405 nm, magenta line), blue (457 nm, blue line), and green (532 nm, green line) activation pulses. (D) Normalized activation rate constants per unit laser power of the three pairs at three activation wavelengths, 405 nm, 457 nm, and 532 nm. The values obtained for Alexa 405-Cy5, Cy2-Cy5 and Cy3-Cy5 were used for normalization at 405 nm, 457 nm and 532 nm, respectively. The absolute activation rates were rapid for each pair at its corresponding optimal wavelength, with values ∼10 s^-1 or greater at only a few hundred μW of laser power. The activation rate of the Alexa 405-Cy5 pair by the 532 nm laser was too small to be measured reliably.

Fig. 2

Three-color STORM imaging of a model DNA sample. (A) Three-color STORM image of three different DNA constructs labeled with Alexa 405-Cy5, Cy2-Cy5, or Cy3-Cy5 mixed at a high surface density on a microscope slide. The image was plotted by rendering each localization as a Gaussian peak, the width of which was scaled with the theoretical localization accuracy given by the number of photons detected (26). Each colored spot in this image represents a cluster of localizations from a single DNA molecule. A conventional fluorescence image of the same area is shown in fig. S3 for comparison. (B, C) Higher magnification views of the boxed regions in (A) show several examples of closely spaced DNA molecules. Here, each localization was plotted as a cross, colored according to following code: if the molecule was activated by a 405 nm, 457 nm, or 532 nm laser pulse, the color of the cross was assigned as blue, green, or red, respectively. (D) The localization distributions of the blue, green, or red clusters. The 2D histograms of localizations were generated by aligning multiple (50 - 60) clusters by their center of mass. The histograms were fit to a Gaussian profile to determine their FWHM.

Fig. 3

STORM imaging of microtubules in a mammalian cell. (A) Conventional immunofluorescence image of microtubules in a large area of a BS-C-1 cell. (B) STORM image of the same area. (C, E) Conventional and (D, F) STORM images corresponding to the boxed regions in (A). (G) Cross-sectional profiles of two nearby microtubule filaments in the cell. The inset shows the STORM image, and the histogram shows the cross-sectional distribution of localizations with in the small regions specified by the white box. (H) Cross-sectional profile of a microtubule segment determined from the STORM image. A relatively long segment (∼ 7 μm) was chosen to obtain good statistics. The histogram shows the cross-sectional distribution of localizations. The green line is a single Gaussian fit with FWHM = 51 nm. The red line shows the fit obtained by convolving a rectangular function of width = d with a Gaussian function of FWHM = r. The fit yields d = 56 nm and r = 22 nm, corresponding to the microtubule width and the imaging resolution, respectively.

Fig. 4

Two-color STORM imaging of microtubules and CCPs in a mammalian cell. (A) STORM image of a large area of a BS-C-1 cell. The secondary antibodies used for microtubule staining were labeled with Cy2 and Alexa 647, while those for clathrin were labeled with Cy3 and Alexa 647. The 457 nm and 532 nm laser pulses were used to selectively activate the two pairs. Each localization was colored according to the following code: green for 457nm activation and red for 532 nm activation. (B) STORM image corresponding to the boxed region in (A) shown at a higher magnification. (C) Further magnified view of the boxed region in (B).