Protein‐Specific, Multicolor and 3D STED Imaging in Cells with DNA‐Labeled Antibodies

Abstract Photobleaching is a major challenge in fluorescence microscopy, in particular if high excitation light intensities are used. Signal‐to‐noise and spatial resolution may be compromised, which limits the amount of information that can be extracted from an image. Photobleaching can be bypassed by using exchangeable labels, which transiently bind to and dissociate from a target, thereby replenishing the destroyed labels with intact ones from a reservoir. Here, we demonstrate confocal and STED microscopy with short, fluorophore‐labeled oligonucleotides that transiently bind to complementary oligonucleotides attached to protein‐specific antibodies. The constant exchange of fluorophore labels in DNA‐based STED imaging bypasses photobleaching that occurs with covalent labels. We show that this concept is suitable for targeted, two‐color STED imaging of whole cells.

U2OS cells were seeded on fibronectin-coated (Sigma Aldrich, Germany, 15 µg/ml fibronectin for 30 min) 8-well chamber slides (Sarstedt, Germany, 2 x 10 4 cells/well) and incubated at 37°C and 5% CO2 in DMEM containing 4.5 g/l glucose, 10% FBS and 1% GlutaMAX (all purchased from Gibco, Thermo Fisher, USA). After 24 h, cells were chemically fixed using optimized protocols either for conservation of the microtubule cytoskeleton or conservation of microtubules and mitochondria. Single colonies of E. coli MG1655 WT cells (CGSC #6300) were picked from plate and grown overnight in LB Miller (Carl Roth, Germany) at 37°C. Cells were diluted 1:500 in 5 ml LB Miller and grown at 30°C to exponential phase while shaking (230 rpm).

Fixation and labeling procedure for microtubules
For labeling of microtubules, the cytosol of U2OS cells was extracted twice for 30-60 s using 37°C prewarmed microtubule stabilizing buffer (MTSB) (80 mM PIPES pH 6.8, 1 mM MgCl2, 5 mM EGTA, 0.5% TX-100, Sigma Aldrich, Germany and Thermo Fisher, USA) (protocol published by Cramer and Desai [1]), followed by chemical fixation using MTSB + 0.5% GA (Electron Microscopy Sciences, USA) for 10 min. Cells were washed thrice with PBS and quenched for 7 min using 0.2% NaBH4 (Sigma Aldrich, Germany) in PBS (Gibco, Thermo Fisher, USA Fixation and labeling procedure for microtubules and mitochondria Cells were fixed for 20 min at room temperature using 3% FA and 0.2% GA in PHEM buffer [2].
Fixation and labeling procedure of bacterial cells E. coli MG1655 cells were grown to OD600 ~ 0.5 (mass doubling time 36.8 ± 1.9 min) and fixed using 2% FA and 0.1% GA in NaPO4 buffer for 15 min [4]. Cells were washed with PBS, incubated 30 min with PBS + 50 mM NH4Cl and immobilized on KOH-cleaned (3M, 1h) and poly-L-lysine-coated (0.01% for 15 min) 8-well chamber slides. After immobilization, cells were blocked for 20 min at room temperature using PBS + 3% IgG-free BSA. 10 µg/ml polyclonal rabbit anti-E. coli antibody (BIO-RAD, #4329-4906) diluted in PBS + 1% BSA were added to each chamber. After 1 h incubation at room temperature, chambers were washed thrice with PBS and custom-labeled secondary antibodies (25 µg/ml P1-donkey-anti-rabbit or P5-donkey-anti-rabbit) diluted in PBS + 1% BSA were added for 1 h (RT). Cells were washed thrice with PBS, twice with PBS + 1% BSA, once again with PBS and postfixed for 10 min at room temperature using 2% MeOH-free FA in PBS. Finally, chambers were washed once with PBS and excess formaldehyde was quenched for 20 min using PBS + 50 mM NH4Cl.

Confocal microscopy
Time series were recorded on a Zeiss LSM710 bearing a Plan-Apo 63x oil objective (DIC M27, 1.4 NA) (Zeiss, Germany). An internal 633 nm diode laser was used for excitation at varying laser intensities, ranging between 15.5 µW and 156 µW (10% and 90% of maximal laser power, respectively). A heating module consisting of a TempModule S (Zeiss, Germany) and a heating insert P (PeCon GmbH, Germany) were used to maintain a temperature of 25°C, which is crucial to maintain constant hybridization/dissociation rates of the docking/imager DNA oligo pairs and to reduce sample drift. The microscope was controlled using Zeiss ZEN 2008 software (v. 5,0,0,267).
Imaging was performed in hybridization buffer consisting of PBS (without Ca 2+ and Mg 2+ ) + 500 mM NaCl pH 8.2 (adjusted using NaOH). Confocal measurements contributing to Figure S2 were recorded on TCS SP8 confocal microscope (Leica Microsystems, Germany). Exact parameters for each measurement are listed in Supplementary For STED imaging, a final concentration of 500 nM of fluorophore-labeled imager strands (see Supplementary Table 1), diluted in the hybridization buffer as used for CLSM imaging was added to the immunostained samples and three time course measurements were recorded for both the covalent and dynamic labeling approach (1 min/frame, 50 frames). For Figure S5B, an oxygen scavenger system consisting of 10 nM protocatechuate-3,4-dioxygenase (PCD, #03930590, Sigma Aldrich, Germany) and 2.5 mM protocatechuic acid (PCA, # P8279, Sigma Aldrich, Germany) was added to the imaging buffer.

Analysis of intensity time traces
Image stacks were exported in TIFF format and further processed using the open-source image analysis package Fiji (v1.51w). Image stacks of fixed cells (z-stacks or multi-frame measurements) were aligned using the Fiji plugin "Stackreg" (translation). Intensity time traces were extracted using a custom-written Fiji macro. In short, an average image was calculated from stabilized time series using the plugin "Z project". The resulting image was manually thresholded and the resulting binary image was segmented using the plugin "Watershed" to prevent inclusion of label-free regions in the analysis. Further, 8 pixels at the image borders were excluded, as they contain artifactual intensity values resulting from the alignment procedure. Resulting ROIs were combined in the "RoiManager" and mean pixel values were extracted for each frame in the original time series. The binary image was inverted in order to obtain the background ROI. We found that the background introduced by freely diffusing imager strands is a reliable indicator for fluctuations in laser intensity. The average background pixel value was determined for each imaging frame and both signal and background values were automatically saved as text files.
The obtained raw data was then further processed using OriginPro 2018 (OriginLabs). Intensity traces of both signal and background were normalized to the first frame. Signal intensity traces were then corrected by division with the background intensity trace. The resulting corrected intensity traces were averaged and plotted with the respective standard deviation. Time series of the covalent label were processed similarly, except that no background correction was applied due to the lack of an indicator for laser intensity fluctuations.

Image deconvolution
Image deconvolution was carried out using Huygens Professional version 16.10.1p2 (Scientific Volume Imaging).

Determination of microtubule diameters
Diameters of microtubules were determined by measuring the full-width-at-half-maximum (FWHM) perpendicular to straight filaments using the Fiji plugin "Plot Profile" and a line width of 400 nm.
FWHM values were extracted using a Gaussian fit function in OriginPro 2018.
Determination of signal-to-background and signal-to-noise ratios The mean pixel values for signal and background were determined as described above and shown in Figure S3. For CLSM images, background intensity was determined ~ 2 -3 µm above the glass surface to avoid contributions by antibodies/labels unspecifically bound to the glass surface. Noise was determined as the standard deviation of the mean background pixel value in each image.

DNA-PAINT analysis
Single-molecule localization data was analyzed using Picasso v. 0.28 [3]. Hybridization events were localized using the LQ Gaussian fitting algorithm with a box size of 7 px and a min. net. gradient of 30k. Data were drift corrected using RCC (500-1000 frames window size) and subsequent frame localizations within 0.5 px (~80 nm) were grouped allowing a dark time of 2 frames. Localizations were filtered according to their PSF width in x and y direction (95 < ρPSF < 205 nm) in order to reduce outof-focus signal. The number of tracked frames was taken as a measure for the emitter ON time (binding time). At least 60,000 single-molecule spots recorded in at least 2 measurements were analyzed to determine the binding time for the imager strands. Binding times were approximated by fitting the distribution with a mono-exponential decay function omitting the first two time intervals (OriginPro 2018).

Supplementary Note 1
Relative binding times of the imager strands used in this study were determined using E. coli MG1655 WT cells labeled for surface exposed K12 epitopes. We chose this target as E. coli cells exhibit a diameter of ~ 1 µm, resulting in good spatial separation of individual hybridization events when imaging the bacterial mid plane. This is superior to DNA-PAINT imaging of e.g. TOM20, as mitochondria show varying diameters, invaginations and constriction sites, thus increasing the fraction of overlapping single-molecule events. Furthermore, filtering of signals according to their PSF width successfully removes out-of-focus events and background localizations which is crucial for robust determination of imager strand binding times. We did not choose DNA-origami for this experiment, as different imaging buffers are used for their imaging (e.g. tris-buffer containing magnesium for DNA-origami stabilization) and we aimed to use the hybridization buffers applied in CLSM and STED imaging.

Supplementary Tables
Supplementary Table 1 Supplementary 3' -ATC TAC ATA TT -5' -antibody P1 imager strand (9 nt duplex) 5' -C TAG ATG TAT -3' -AbberiorSTAR635P P1 imager strand (8 nt duplex) 5' -AG ATG TAT -3'-AbberiorSTAR635P P5 docking strand 3' -TAT GTA ACT TT -5' -antibody P5 imager strand (9nt duplex) 5' -C ATA CAT TGA -3' -AlexaFluor594 Supplementary Table 2 Supplementary Table 2 is provided as separate .xlsx file and comprises a detailed overview on the imaging conditions for each experiment. Figure S1: Determination of imager strand binding times using single-molecule DNA-PAINT imaging. The K12 envelope of E.coli MG1655 cells were visualized with DNA-PAINT using the 8 nt [A] and 9 nt [B] duplex forming P1-AbberiorSTAR635P imager strand, as well as the 9 nt duplex forming P5-AlexaFluor594 imager strand [C]. Shown are representative diffraction limited images (i) (standard deviation images of DNA-PAINT movies) and the respective super-resolved DNA-PAINT images (ii). Low irradiation intensities (0.3 and 0.45 kW/cm² for AlexaFluor594 and AbberiorSTAR635P, respectively) were used to minimize photobleaching of hybridized imager strands while still bound to the docking strand. Pronounced bleaching of bound imager strands would lead to an artificial shortening of the binding time. Still, the experimental localization precision (NeNA value [5]) were in the range of 8-12 nm. For each imager strand, binding times were extracted (> 60,000 events per strand). Fitting the binding time distribution with a mono-exponential decay (iii) provided binding halftimes of 209 ± 3 ms (P1-AbberiorSTAR635P, 8 nt duplex), 491 ± 7 ms (P1-AbberiorSTAR635P, 9 nt duplex) and 363 ± 8 ms (P5-AlexaFluor594, 9 nt duplex). Scale bar is 1 µm.    Signal-to-noise and signal-to-background ratios of STED measurements using the covalent or dynamic labeling approach (an exemplary ROI is shown in Figure 1C). Signal to background (3.2 ± 1.4 vs. 2.7 ± 0.6) and signal to noise ratios (10.6 ± 6.7 vs. 6.6 ± 3.8) were slightly higher for the covalent label compared to the dynamic label. Note that this could be explained by an increased degree of labeling for secondary antibodies carrying AbberiorSTAR635P dyes.

Supplementary Figures
[B] Signal over time in STED time series using 500 nM P1-AbberiorSTAR635P (8 nt duplex) without (black dots and shaded areas) and with (red dots and shaded areas) the PCA/PCD oxygen scavenger system. The removal of oxygen has no apparent effect on the intensity time trace. Three measurements were analyzed for each condition. Plots show mean values (bars and dots) and the respective standard deviations (error bars and shaded areas).

Supplementary_Video_1.mov
Title: Supplementary Video 1: Two-color STED of U2OS labeled for tubulin and TOM20 with transiently binding DNA-oligomers.

Supplementary_Video_2.mov
Title: Supplementary Video 2: Volumetric rendering of a 3D STED image of a single mitochondrion.