Synergizing Exchangeable Fluorophore Labels for Multitarget STED Microscopy

Investigating the interplay of cellular proteins with optical microscopy requires multitarget labeling. Spectral multiplexing using high-affinity or covalent labels is limited in the number of fluorophores that can be discriminated in a single imaging experiment. Advanced microscopy methods such as STED microscopy additionally demand balanced excitation, depletion, and emission wavelengths for all fluorophores, further reducing multiplexing capabilities. Noncovalent, weak-affinity labels bypass this “spectral barrier” through label exchange and sequential imaging of different targets. Here, we combine exchangeable HaloTag ligands, weak-affinity DNA hybridization, and hydrophophic and protein–peptide interactions to increase labeling flexibility and demonstrate six-target STED microscopy in single cells. We further show that exchangeable labels reduce photobleaching as well as facilitate long acquisition times and multicolor live-cell and high-fidelity 3D STED microscopy. The synergy of different types of exchangeable labels increases the multiplexing capabilities in fluorescence microscopy, and by that, the information content of microscopy images.

Experimental Methods

Chemical Procedures
All chemical reagents for synthesis were purchased from commercial suppliers (Acros, Fluka, Merck KGaA, Roth, Sigma-Aldrich, TCI, TOCRIS, AAT Bioquest) and used without further purification. Water-free solvents were stored over molecular sieves and used directly from a sealed-bottle. Reactions performed under air and moisture exclusion were carried out in heat-dried glassware and under inert argon atmosphere (Schlenk technique). Evaporation in vacuo was achieved using a rotary evaporator (Heidolph) or by lyophilization on a lyophilizer (Christ) equipped with a vacuum pump (Vacuubrand). Reaction progress was monitored by analytical thin-layer chromatography (TLC, POLYGRAM® SIL G/UV254, 0.2 mm layer pre-coated polyester sheet, 40 x 80 mm, Roth) or liquid chromatography coupled to mass spectrometry (LC-MS, Shimadzu LCMS2020 connected to a Nexera UHPLC system. Column: C18 1.7 µm, 50 × 2.1 mm (ACQUITY UPLC BEH, Waters). Buffer A: 0.1% Formic acid (FA)/MiliQ® water (ddH2O), buffer B: acetonitrile (MeCN). Typical gradient was from 10% to 90% B within 6 min with 0.5 mL min-1 flow.
Flash column purification was performed using a Biotage (IsoleraTM One) flash system equipped with pre-packed SiO2 columns (SiliaSepTM Flash Cartridges, 40 -63 µm, 60 Å). Preparative reversed-phase high-performance liquid chromatography (RP-HPLC) was carried out on an UltiMate 3000 system (Thermo Fisher Scientific 4-azido-1-iodobutan (4): 3 (1.8 g, 9.4 mmol, 1 eq.) and NaI (14 g, 94 mmol, 10 eq.) were mixed in 15 mL acetone in a round-bottom flask. The suspension was stirred at rt overnight. Residual solvent was removed in vacuo and the remaining solid was dissolved in 100 mL DCM and H2O each. The aq. layer was extracted twice with 100 mL DCM. The organic layers were combined, washed with 100 mL sat. Na2S2O3 and brine, dried over MgSO4, filtered and the solvent was removed. Flash column chromatography (5 to 30% EtOAc in nhexane in 6 CV) afforded compound 4 (1.

Intensity time traces of covalent and exchangeable HaloTag ligands
Intensity time-trace analysis of covalent and exchangeable HaloTag ligands was performed using a custom-written ImageJ macro (as described by Spahn et al. 1 ). In brief, image sequences were drift-corrected using the ImageJ plugin "StackReg". User defined thresholding was then used to generate binary masks for image segmentation and extraction of signal and background intensity values. After correction for edge-effects, for each frame the signals from structures were averaged, background corrected and further analyzed in Origin software (Origin2019).

Determination of spatial resolution in STED images
To determine the spatial resolution in STED images, chemically fixed vimentin-HT7 expressing U2OS cells were labeled using xHTLs at concentrations of 100 nM and 500 nM. At high label densities, continuous signal was detected along the filaments, allowing the determination of the full-width half-maximum (FWHM) of the structure. For this purpose, a custom written analysis pipeline was established in python (https://github.com/MariusGlg/Filament_width_analyzer). In brief, 10-50 lines equally spaced and perpendicular to the vimentin filaments were drawn. Intensity profiles along the line segments (n = 691) were fitted with a Gaussian function and full-width half-maximum values (FWHM) were extracted for each segment. FWHM values were then plotted as a relative frequency distribution and fitted with a Gaussian function in Origin software. At low labeling densities, the point-spread-functions (PSF) of single vimentin-HT7 spots was analyzed as a measure of the spatial resolution. For this purpose, local intensity maxima of STED images reflecting the vimentin-HT7 positions were identified in ImageJ using intensity value thresholding ("Find Maxima" function) based on background and signal intensities. The ImageJ plugin "GaussFit OnSpot" was then used for fitting Gaussian profiles onto selected positions and extracting the FWHM (n = 88). As a third measure of the spatial resolution, the intensity profiles of vimentin structures at intersection points were analyzed. In brief, intensity line profiles perpendicular to the axes of vimentin filaments that were present in close proximity to each other were analyzed using ImageJ. The measured line profiles were then fitted using a multi-peak Gaussian function in Origin software to determine their full FWHM and mean. The FWHM and the distance between the means were taken to be the physical resolution achievable with biological samples in the setup used. In Fourier ring correlation based analysis of the image resolution, exchangeable and covalent HT7 labels were compared. For this purpose, chemically fixed cells were either covalently labeled with SiR-HTL (300 nM, incubation for 30 min at RT, followed by three washing steps) prior to imaging or directly imaged in 500 nM SiR-S5 (PBS). Following parameters were used for STED microscopy: 3% excitation power, 6% depletion laser power, 20 nm pixel size, pixel dwell time of 3 µs, 10x line accumulations and a pinhole size 0.71 AU. To create image pairs for the FRC analysis, the same ROI was imaged twice with no time delay between the imaging rounds to minimize sample drift. The FRC analysis was conducted using the NanoJ plugin in Fiji 2 with 10 measurements per condition and 20 blocks per image for good sampling of signal and background. Average FRC values per image were finally plotted in Origin software. For the determination of signal-to-noise and signal-to-background ratios, chemically fixed vimentin-HT7 expressing U2OS cells were prepared and imaged as described for the FRC analysis. The analysis was performed in Fiji. In brief, user defined intensity thresholds were used to create binary masks for the individual STED images. The mask was then applied to an image for segmentation of signal and background and mean pixel intensity values were then determined for both. Noise was determined as the standard deviation of the mean background pixel value in each image.

3D reconstruction of STED z-stacks
For volumetric rendering of STED z-stacks, images were first background subtracted in ImageJ using a rolling ball radius of 50 pixels. Images were then deconvoluted using the ImageJ plugin "DeconWithGaussian" and the zposition was color-coded using the plugin "Z-stack Depth Colorcode" (LUT spectrum). For single-color volumetric rendering and generation of 3D movies the plugin "3D viewer" was used. Dual-color 3D rendering and generation of dual-color movies was conducted in Napari open source software 3 .

Supplementary Tables
Supplementary Table 1 Primary and secondary antibodies used for DNA-PAINT based labeling in this study.

Antibodies
Target

Supplementary Figures
Supplementary Figure 1. Confocal and STED microscopy of lysosomes and mitochondria using exchangeable fluorophore labels. A) Confocal laser scanning microscopy (CLSM, green) and STED (magenta) image of lysosomes (LamP1-HT7, transient transfection) labelled using xHTLs (JF635-S5) in a U2OS cell. i) Overview composite image with white box indicating a region of a single lysosome magnified in ii) and iii). iv) Normalized intensity profile across the lysosome (marked as white lines in ii) and iii)). The intensity distribution was fitted with a single (green) or two Gaussian functions (magenta). Scale bars are 10 µm (i) and 1 µm (ii-iii). B) CLSM (green) and STED (magenta) image of mitochondria (TOM20) labelled via DNA-PAINT (P1-Abb635 P) in a U2OS cell. i) Overview composite image with white box indicating a region magnified in ii) and iii). iv) Normalized intensity profile across single mitochondria (marked as white lines in ii) and iii). The intensity distribution was fitted with a single (green) or multi-gaussian function (magenta). Scale bars are 10 µm (i) and 2 µm (ii-iii).