The Force of the Dark Side – Embedding Media for GSDIM Super-Resolution Localization Microscopy

The key to overcoming the diffraction limit of ~200 nm with GSDIM microscopy is the sequential shutting on and off of fluorophores and consequently their temporal separation. In this context, electrons of a fluorophore accumulate in a metastable, non-fluorescent dark state (triplet state and others) by application of a high-power laser beam [2]. From the dark state, molecules return stochastically to the ground state to enter fluorescence cycles which are detected by the microscope – the so called "blinking". These fluorescent events are then captured over time. To get a high-resolution image, the diffraction-limited events are read out by applying a fitting algorithm. Eventually, the ultimate high-resolution image is reconstructed by plotting the measured positions of thousands of recorded events (Figure 1).

With the GSDIM technique a lateral resolution down to 20 nm can be achieved [2]. In practice, the final resolution depends on the localization precision of an individual fluorophore, which is in turn determined by the properties of the fluorophore and the embedding medium. The following equation [3] shows that the localization precision is dependent on the number of collected photons emitted by one individual fluorophore:

Δr denotes the diffraction limit of the microscope/objective, n the number of photons emitted by an individually localized fluorophore and Δr_sms the localization precision of the distinct fluorophore.

Ultimately, the aim of achieving the best localization precision is directly linked to the effort of a) maximizing the number of emitted photons per fluorophore/event and b) maximizing the number of fluorescence cycles (= minimizing photobleaching). These characteristics are basically influenced by the interplay of the fluorophore's properties and the surrounding embedding medium affecting the dark state and thus the blinking performance.

In the 3D GSDIM system a cylindrical lens is introduced into the light path of the microscope. The resulting optical astigmatism leads to an ellipticity of those blinking events which are not in the focal plane [4]. The point-spread-function (PSF) of events which are above the focal plane is elongated vertically, whereas the PSF of those events which are beneath the focus is elongated horizontally. The software finally correlates each blinking event with a distinct z-coordinate. It is evident that this approach requires a well calibrated acquisition of each blinking event and high localization precision. Accordingly, impressive three-dimensional data can be obtained with an optimized embedding method (Figure 2).

Fig. 2: 3D GSDIM high-resolution image of microtubules and mitochondria
Microtubules of COS-7 cells (A) and mitochondria of MDCK cells (B) in false color code indicating the z-position 0–800 nm. Cells were stained against αTubulin or ATP5B, both with AlexaFluor^®647, and embedded in Vectashield^®/Glycerol-TRIS. For comparison, widefield images are depicted above the respective 3D super-resolution image. Scale bar 10 µm.

The most important step in GSDIM microscopy is to get nearly all fluorophores into the long-lived dark state, which is dependent on special embedding media. Some of the currently known media, like glucose-oxidase-mix or polyvinyl alcohol (PVA), tend to decrease the amount of molecular oxygen in the specimen. Oxygen acts as a triplet state quencher and thus reduces the amount of fluorophores in the triplet state. This is absolutely counterproductive in GSDIM, as the triplet state is an important pathway to the dark states. Accordingly, the decrease in the population of triplet state fluorophores leads indirectly to a loss of localization precision and final resolution, because more fluorophores are in the “on” state interfering with the super-resolution image acquisition process.

In contrast to other high-resolution techniques, the big advantage of GSDIM microscopy is the usage of standard fluorophores and conventional staining protocols which are employed in common light microscopy methods. Only the combination of fluorophore and embedding medium leads to limitations. However, the currently used GSDIM media already allow the application of a very wide range of different fluorophores and even fluorescent proteins. Nevertheless, all of them have their advantages and disadvantages (Figure 3). The most frequently used medium Cysteamine or β-Mercaptoethanolamine (MEA) for example is applicable for a large variety of fluorophores, but unfortunately the medium must be used freshly and its efficiency drops significantly after six hours of usage, leading to a reduced blinking performance. The search for other embedding media and suitable fluorophores is therefore an important step to fully tap the great potential of this super-resolution microscopy method.

Vectashield^® – an alternative embedding medium for 3D GSDIM microscopy

In this context Olivier et al. were able to identify the mounting medium Vectashield^® (Vectorlabs™) as an adequate embedding medium for dSTORM microscopy [1]. Here we tested if this new embedding method could be adopted for 3D GSDIM microscopy and searched for advantages compared to currently used media. In addition, we performed a systematic screen for other fluorophores working in combination with Vectashield^®.

Notably, we found that the fluorophore pair AlexaFluor^®647 and AlexaFluor^®555 (Life Technologies™) showed outstanding properties in this embedding medium with an excellent blinking performance for highest localization precision in 3D GSDIM. This is in line with the findings of Olivier et al. [1]. With AlexaFluor^®647 and AlexaFluor^®555 we were able to generate impressive 3D GSDIM data (Figure 5). In combination with Vectashield^®, AlexaFluor^®647 and AlexaFluor^®555 even exhibited a superior blinking performance compared to currently used fluorophore pairings like AlexaFluor^®647 and AlexaFluor^®488 in MEA or glucose-oxidase-mix. This can be visualized by the comparison of two blinking images of the lists of events performed with Vectashield^® or MEA respectively (Figure 4) and underlines the great potential of this embedding medium.

Indeed, the blinking properties of the fluorochrome pair is not the only great advantage of this medium: Specimens treated with Vectashield®/Glycerol-TRIS can be stored for months at 4 °C without any loss of their excellent blinking properties (Figure 6).

Drawbacks like its relatively strong autofluorescence and resultant background, especially in the 532 nm laser range, can be diminished by mixing Vectashield^® with a buffer containing glycerol and 50 mM TRIS pH 8 at a ratio of 1:10 [1]. The so-called backpumping laser with 405 nm wavelength, which is used to augment the number of fluorescent cycles of the fluorophore, is not usually required for either AlexaFluor^®647 or AlexaFluor^®555 because of their excellent blinking properties in Vectashield^®. This also prevents background formation.

In our further studies to identify other suitable fluorophores, we found, interestingly, an acceptable blinking performance of the fluorescent protein EYFP comparable to the performance in glucose-oxidase-mixture.

Unfortunately, we were not able to find other fluorophores, especially not in the 488 nm range, where AlexaFluor®488 and Atto^®488 did not show the expected blinking performance. Regarding the great number of still untested candidates and the increasing number of fluorophores which are specially designed for super-resolution microscopy, the drawback of a limited number of fluorophores will certainly be diminished in the future. As in the case of MEA, the list of blinking fluorophores in combination with Vectashield^® is sure to grow with time and ongoing testing.

Tab. 1: Verified fluorophores (also compare Olivier et al. [1]). The listed fluorophores were tested by immunostaining of ATP5B and αTubulin in COS-7 cells embedded in Vectashield^®/Glycerol-TRIS (1:10). Their blinking behavior was evaluated with the Leica SR GSD 3D system.

Fig. 5: 3D GSDIM high-resolution image – dual staining of mitochondria and microtubules. COS-7 cells were stained against αTubulin and ATP5B with AlexaFluor^®647 and AlexaFluor^®555 respectively. The specimens were embedded in Vectashield^®/Glycerol-TRIS. Comparison of GSDIM and deconvoluted widefield 3D stacks. Microtubules (αTubulin) are displayed in green and mitochondria (ATP5B) in red.

Fig. 6: 3D high-resolution image comparison: fresh (A) vs. two months old (B) specimen. (A) 3D GSDIM images of microtubules in green and mitochondria in red or in false color code respectively, indicating the z-position. As in Figure 5, the specimens were embedded in Vectashield/Glycerol-TRIS. (B) The specimens used in Figure 5 and Figure 6A were stored for two months at 4 °C and then imaged again, still embedded in Vectashield^®/Glycerol-TRIS.

COS-7 cells were cultured for one day on precision coverslips (Marienfeld-Superior™) and then fixed with ice-cold methanol. After a block with PBS containing 1 % milk powder for 1 h, cells were incubated for 2 h with the primary antibodies directed against ATP5B (Santa Cruz™) and α-Tubulin (Sigma™) in a ratio of 1:100 in PBS/milk powder. To label the primary antibodies with AlexaFluor^®647 or AlexaFluor^®555 respectively, secondary antibodies were incubated for 1 h in a dilution of 1:200 in PBS. Each step except the block with milk powder was accompanied by four washing steps with PBS.

As depicted in Figure 7, cells were embedded in Vectashield^®/Glycerol-TRIS-buffer (Vectashield^® and Glycerol-TRIS-buffer in a 1:10 ratio; glycerol with 50 mM TRIS pH 8) as follows: 75–100 µL medium was placed upon the cavity of the depression slide. Then the coverslip was positioned on the depression slide with the cells facing the medium and the depression. In this step it is important to avoid any air bubbles. An excess of medium has to be removed by filter paper to ensure that the two-component silicone Twinsil^® is able to dry correctly. The yellow and blue components of Twinsil^® were mixed 1:1 and then applied to the edges of the coverslip. After 10 minutes, the silicone was hardened and the specimen was ready for GSDIM imaging.

Breaking the diffraction barrier by regional or temporal separation of single fluorescent events has revolutionized light microscopy. In the future, there will be no more resolution constraints due to the diffraction limit, but the development of microscopes with ever-decreasing limits in resolution reveals new challenges such as labeling or embedding media in the case of GSDIM microscopy. As mentioned earlier, the critical step in all super-resolution methods is to get fluorophores on the dark side, which is the basis for good localization precision. In GSDIM microscopy, this process is mainly influenced by the presence of the embedding medium. Here, the interplay between the embedding medium and the properties of the fluorophore is the decisive factor for successful high-resolution imaging.

Nowadays, we have powerful fluorophore pairings like AlexaFluor^®647 and AlexaFluor^®488 in MEA or AlexaFluor^®647 and AlexaFluor^®555 in Vectashield^®/Glycerol-TRIS which allow the acquisition of impressive 3D super-resolution images in two colors. In the near future our repertoire of efficient fluorophores in combination with the embedding medium Vectashield^® is sure to be expanded by ongoing tests and practical experience. Together with other high-resolution techniques like STED microscopy, this will give us a completely new view of cellular life and its biochemical regulation.

1. Olivier N, Keller D, Gönczy P and Manley S: Simple buffers for 3D STORM microscopy. Biomed. Opt. Express 4 (6): 885–99 (2013).
2. Fölling J, Bossi M, Bock H, Medda R, Wurm CA, Hein B, Jakobs S, Eggeling C and Hell SW: Fluorescence nanoscopy by ground-state depletion and single-molecule return. Nature Meth. 5: 943–45 (2008).
3. Thompson RE, Larson DR and Webb WW: Precise Nanometer Localization Analysis for Individual Fluorescent Probes. Biophysical Journal 82: 2775–83 (2002).
4. Huang B, Wang W, Bates M and Zhuang X: Three-Dimensional Super-Resolution Imaging by Stochastic Optical Reconstruction Microscopy. Science 319: 810–13 (2008).