STED-PAINT for

The following examples and protocols provide a practical introduction and can be taken as a starting point for own studies. For a detailed description of the different approaches see Appendix 1.

Figure 1. STED-PAINT Imaging of bacterial membranes (magenta) and DNA (green) using a high concentration of the exchangeable labels Nile Red and JF646-Hoechst. (A) Frame accumulation, enabled by dye exchange, yields higher image brightness showing details that are otherwise lost in noise (arrow). (B) Dye replenishment allows repetitive volume imaging with 3D super-resolution. Bottom row shows isolated DNA channel for clarity. Scale bars, 1 µm.

Bacterial membranes and DNA

For the first imaging example (Fig. 1), we recorded bacterial membranes and DNA with STED-PAINT microscopy. For labeling fixed bacterial cells, we chose a combination of two dyes that were previously shown to work well with the STED-PAINT approach: the lipophilic dye Nile Red for staining membranes, and the DNA-binding dye conjugate JF₆₄₆-HOECHST. Similar results can be obtained using SiR-Hoechst ^{15, 19}. In line with the general approach, the dyes were added to the imaging medium at concentrations around 300 nM.

The small size of bacterial cells limits the ability of light microscopy to distinguish subcellular bacterial organelles or macromolecular assemblies, let alone determining the organelle shape or the distribution of proteins within. However, many processes involve specialized structures that require higher resolution for sufficient visualization, as seen for cell division that involves the FtsZ ring. Here, optical super-resolution microscopy not only resolves the ring structure, but also reveals the interplay of key components owing to its multicolour capabilities ²⁰. However, to study such dynamic processes at high resolution over time, one requires an approach such as PAINT.

Sample: E.coli cells, fixed in mid-exponential phase (2 % FA/0.05 % GA) were washed with PBS and immobilized on KOH-cleaned and poly-L-lysine-treated glass coverslips and permeabilized using 0.5 % TritonX-100. Staining and imaging was performed in 300 nM Nile Red and 300 nM JF₆₄₆-Hoechst in 150 mM Tris pH 8.0 at RT.

Imaging: Nile Red shows fast replenishment, while JF646-Hoechst replenishes more slowly. Due to their small size, bacteria are well-suited for approach 2 using small image frames that can be recorded fast. At high frame acquisition rates, stage drift does not pose a problem. For 2D-STED, 17 or more image frames were accumulated to improve signal (Fig 1 A). For volume imaging with 3D super-resolution, closely spaced z-planes (pixel size: 50 x 50 x 50 nm) were recorded without noticeable bleaching, even after acquisition of ten full volumes.

Example settings 2D-STED: Approach 2, frame size 5.6 x 6.8 µm (XY), pixel size 30 nm (XY), dwell-time 10 µs, Nile Red 15-23 % excitation at 561 nm and 40 % STED at 775 nm (high-power variant), JF646-Hoechst 4-4.5 % excitation at 640 nm and 10 % STED at 775 nm (high-power variant), line accumulations 1x (confocal) / 3x (STED), frame signal accumulation ≥17x.

Example settings 3D-STED: Approach 1, frame 2.4 x 2.4 x 2.5 µm (XYZ), pixel size 50 nm (XYZ), Nile Red as above, JF646-Hoechst 5-6.2 % excitation at 640 nm and 20  % STED at 775 nm (high-power variant), line accumulations 1x (confocal) / 3x (STED).

Figure 2. STED-PAINT imaging of a Nile Red labeled HeLa cell provides bright images of mitochondrial membranes and a range of other cellular membranes including endoplasmatic reticulum, plasma membrane, vesicles and lipid droplets. Dye replenishment during acquisition allows bright STED images without bleaching. Raw data shown.

Mitochondrial membranes in fixed and living mammalian cells

Imaging the structure and dynamics of the mitochondrial inner membrane is currently a hot topic in organelle biology. In addition to tagging and labeling of mitochondrial proteins using small tags such as the SNAP tag ¹³, results have been published using mitochondria specific probes ¹⁴.

Moreover, Span et al. ¹⁵ described the use of Nile Red for labeling and imaging of a range of cellular membranes as well as lipid droplets. Amongst other results, they show that this dye strongly labels mitochondrial membranes and can reveal the inner structure of cristae (Fig. 2). However, due to the broad specificity of the dye, a counterstain might be required to unambiguously identify structures of interest.

Sample: HeLa cells fixed with 4 % FA / 0.1% GA in PHEM buffer for 60 min (adapted from ²⁵) were labeled and imaged with 300 nM Nile Red in 150 mM Tris pH 8.0 at RT. Note: Fix specimen for ≥ 60 min with a mixture of FA and GA and do not permeabilize cells to preserve membrane structures.

Imaging: Nile Red replenishes quickly and is suited for approach 1 (see p. 6). As Nile Red requires relatively high STED laser powers for sufficient resolution, bleaching is dominant and dye exchange crucial. To collect enough signal despite photobleaching, multiple line accumulations at relatively low scan speeds (e.g. via increased pixel dwell-times) are required. Brightest results were achieved for large image frames (>20 µm), as this slows down each line acquisition and provides more recovery time between line repeats.

Example settings: Approach 1, frame 26.2 x 27.4 µm (XY), pixel size 20 nm (XY), dwell-time 10 µs, 2-10 % excitation at 561 nm, 65 % STED at 775 nm (high-power variant), line accumulations 8x (confocal) / 24x (STED), single frame.

Figure 3. STED-PAINT imaging of Lifeact Alexa Fluor 594 in fixed HeLa cells allows bright, high-resolution STED images of the cytoskeleton fine structure as it binds only transiently and is constantly replenished from the imaging medium. Small section of larger frame showing raw data.

Cytoskeleton in fixed and living mammalian cells

The cytoskeleton of mammalian cells lends itself as a beautiful test and demonstration object for practically all types of light microscopy. Nevertheless, numerous open scientific questions are related to the fine-structure and function of cytoskeleton networks ^{14, 16, 17}.

Labeling the cytoskeleton has been done for decades using immunofluorescence, toxin-based labeling, or methods based on GFP-tagging. Many of these approaches are well suited for analyzing the coarse structure of the network. However, the fine structure is often dim and hard to resolve well. Often, this can be attributed to low labeling densities of fine filaments and/or rapid photobleaching.

Here, PAINT labeling was applied in fixed mammalian cells using the Lifeact peptide ¹⁸ coupled to Alexa Fluor 594. As shown in Fig. 3, the exchange of labels in the medium allows the extended accumulation of signal and reveals much more details among the fine structures of the cytoskeleton network than classical immuno- or phalloidin-labeled samples.

Sample: HeLa cells were fixed using microtubule-stabilizing buffer as described previously ¹² to preserve cytoskeletal structures. Staining and imaging at RT with 1-2 µM Lifeact-Alexa Fluor 594 in 100 mM Tris pH 8.0 and oxygen scavenging via 2.5 mM PCA and 10 nM PCD ²¹. Note: Oxygen scavenging helps protect cytoskeletal structures during image acquisition.

Imaging: Approach 1 was employed as Lifeact Alexa Fluor 594 exchanges rapidly. Recording with a high number of line accumulations improved signal for fine structures.
Example settings: Approach 1, frame 73.6 x 74.6 µm (XY), pixel size 20 nm (XY), dwell-time 10 µs, 20-50 % excitation at 561 nm, 25 % STED at 775 nm (high-power variant), line accumulations 8x (confocal) / 24x (STED), single frame.

Example settings: Approach 1, frame 73.6 x 74.6 µm (XY), pixel size 20 nm (XY), dwell-time 10 µs, 20-50 % excitation at 561 nm, 25 % STED at 775 nm (high-power variant), line accumulations 8x (confocal) / 24x (STED), single frame.

STED-DNA-PAINT

The applications presented above demonstrate the performance of the STED-PAINT combination. All are based on toxin or lipid labeling. Nevertheless, most scientific questions are centered around the function(s) of specific proteins or protein complexes, necessitating labeling approaches specific to a defined target protein. In most cases, toxin-based stains are unavailable or lack the required specificity.

Figure 4. The DNA-PAINT principle allows to combine immunolabeling of target proteins with the PAINT approach. For this, three components are required: the primary antibody, binding to the target structure, the secondary antibody, binding to the primary antibody. The secondary antibody is coupled to the ssDNA docking strand and an ssDNA imager strand labeled with dye is exchanged between the bound and unbound state.

One highly specific labeling approach is DNA-PAINT, based on classic immunostaining ^{10, 12}. Here, a target protein is immunolabeled with target-specific primary antibodies and secondary antibodies which are conjugated to DNA strands (docking strands). For imaging, dye-coupled complementary DNA strands (imager strands) are added, which can temporarily bind to the antibody-coupled DNA strand via base pairing. This allows generalization of the PAINT concept to more protein-specific labeling. Importantly, the binding kinetics of docking and imager strands can be fine-tuned by changing the DNA sequence, buffer composition or temperature. This provides multiple levers to optimize DNA-PAINT for bright STED images and faster image acquisition ¹².

To demonstrate the performance of this approach, we combined simple PAINT labeling via Lifeact-Alexa Fluor 594 with DNA-PAINT labeling via tubulin-specific primary antibodies, DNA-conjugated secondary antibodies and DNA imager strands conjugated to abberior STAR 635P.

In summary, the results of the STED-DNA-PAINT approach stand for its own (Fig. 5). This points towards a general applicability of PAINT for numerous fixed cell imaging experiments, where the localization of one or more target proteins is the question of interest. However, as the DNA-PAINT concept is not directly compatible to live cell imaging, more dynamic questions in field of bioimaging cannot be addressed with this technology.

Figure 5. Combination of STED-PAINT and STED-DNA-PAINT. In addtition to actin labeling as in figure 3, Tubulin was labeled in fixed HeLa cells using DNA-PAINT with DNA-conjugated antibodies and transient binding of complemetary DNA imager strands conjugated to abberior STAR 635P. Both stains are replenished from the buffer rendering bright STED images.

Sample: HeLa cells were fixed using microtubule-stabilizing buffer as described previously ¹² to preserve cytoskeletal structures. Immunostaining used mouse anti-β-tubulin primary antibodies (#32-2600, Thermo Fisher, USA) and custom, DNA-conjugated secondary antibodies ¹². Final staining and imaging at RT with 1-2 µM Lifeact-Alexa Fluor 594 and 500 nM STAR 635P-conjugated imager strand in PBS (without Ca²⁺ and Mg²⁺) with 500 mM NaCl pH 8.2 and oxygen scavenging via 2.5 mM PCA and 10 nM PCD ²¹. Note: Oxygen scavenging helps protect cytoskeletal structures during image acquisition.

Imaging: Approach 1 was employed as Lifeact Alexa Fluor 594 exchanges rapidly and the DNA-PAINT imager strand at medium speed. A high number of line accumulations was used to increase signal for fine actin structures. Note that line accumulations are lower for STAR635P, owing to its higher brightness but slower replenishment (requiring more breaks).

Example settings: Approach 1, frame 51.2 x 42.2 µm (XY), pixel size 20 nm (XY), dwell-time 10 µs, Lifeact-Alexa Fluor 594: 20-50 % excitation at 561 nm, 25 % STED at 775 nm (high-power variant), line accumulations 2x (confocal) / 12x (STED), DNA-PAINT STAR 635P: 10-20 % excitation at 640 nm and 18 % STED at 775 nm (high-power variant), line accumulations 2x (confocal) / 6x (STED), single frame.

Multicolor DNA-PAINT & STED

Recently, multicolour DNA-PAINT has been published and used to create biological insight ^22,10,12,23. Here, multiple biological targets are stained with specific primary antibodies, each conjugated to a distinct DNA docking strand. Each target is imaged after a separate staining step using an DNA imager strand complementary to only one of the docking strands. All imager strands are conjugated to the same dye, allowing the same settings for each target. Importantly, in between two imaging rounds rigorous washings steps are required to ensure only the intended imager strand is present in the sample. Alternatively, imager strands with different sequences (and thus target-specificity) can be added simultaneously and spectrally distinguished by the excitation wavelength and detection window while using the same depletion laser wavelength (e.g. Alexa Fluor 594 and abberior STAR 635P). Several of those combinations have been already tested for STED-DNA-PAINT applications ¹².

Additionally, with DNA-conjugated primary antibodies, the number of parallel stains is not limited by the repertoire of antibodies from different species, further expanding the multiplexing capability of DNA-PAINT and STED-DNA-PAINT.

Although the general applicability of this approach in combination with STED microscopy needs to be confirmed, the results of these initial experiments are extremely promising and point towards a very interesting future perspective. Recent advances in the field of DNA-PAINT, such as SpeedPAINT ²⁴, may further improve imaging speed and signal-to-noise ratio, rendering PAINT-STED an even more and useful tool.

Table 1. Exemplary Labels for STED-PAINT