Photoactivation of silicon rhodamines via a light-induced protonation

Photoactivatable fluorophores are important for single-particle tracking and super-resolution microscopy. Here we present a photoactivatable fluorophore that forms a bright silicon rhodamine derivative through a light-dependent protonation. In contrast to other photoactivatable fluorophores, no caging groups are required, nor are there any undesired side-products released. Using this photoactivatable fluorophore, we create probes for HaloTag and actin for live-cell single-molecule localization microscopy and single-particle tracking experiments. The unusual mechanism of photoactivation and the fluorophore’s outstanding spectroscopic properties make it a powerful tool for live-cell super-resolution microscopy.

A dvances in super-resolution microscopy (SRM) have led to insights into cellular structures and processes over the past decade 1,2 . One of these SRM approaches is singlemolecule localization microscopy (SMLM), which relies on the switching of fluorophores between an off and an on state [3][4][5] . The switching can be achieved by using photoactivatable or switchable fluorophores [6][7][8] . Up to date, fluorescent proteins and small-molecule fluorophores are the two most commonly used fluorophore classes in SMLM. For live-cell experiments with intracellular proteins, fluorescent proteins have the main advantage that no chemical labeling steps are required. However, smallmolecule fluorophores are generally brighter and more photostable than fluorescent proteins 9 , and therefore of advantage for SMLM experiments 10,11 . Photoactivatable (or caged) smallmolecule fluorophores are known throughout many of the different fluorophore families and are mainly synthesized using photolabile protecting groups [11][12][13][14] . Photoactivatable rhodamine derivatives have been obtained through the attachment of orthonitrobenzyl moieties 11 . However, these probes are mostly used in fixed-cell microscopy due to their decreased solubility and poor cell-permeability [15][16][17] . Furthermore, they result in the stoichiometric formation of very electrophilic nitroso-aldehydes or ketones as reactive byproducts, which are toxic and of concern in live-cell imaging 18 . Rhodamines have also been rendered photoactivatable through a diazoketone group 19 , leading to the introduction of the photoactivatable Janelia Fluor dyes PA-JF 549 and PA-JF 646 20 , which have been successfully used for fixed-cell and live-cell SMLM. However, photoactivation of these fluorophores leads to the formation of a dark side-product. The extent, to which the undesired side-product is formed, depends on the structure and environment of the fluorophore complicating applications of the diazoketone approach. In addition, photoactivation of fluorophores caged with the diazoketone group proceeds through a carbene, which can react with intracellular nucleophiles (Supplementary Fig. 1) 21 . In light of the limitations of the existing caging strategies, alternative chemical strategies are needed to generate photoactivatable fluorophores.
Here, we report the discovery, synthesis and characterization of a class of cell-permeable, photoactivatable fluorophores (PA-SiRs), which are based on the silicon rhodamine (SiR) scaffold and activated through a light-induced protonation. We demonstrate the utility of these fluorophores for live-cell SMLM of intracellular targets and single-particle tracking experiments.

Results
Synthesis and characterization. The first analog of this class of fluorophores was serendipitously found during the attempted synthesis of a SiR derivative bearing an alkyl chain in place of the aromatic substituent at the 9 position of the xanthene scaffold (Fig. 1a). Instead of the desired fluorescent SiR 2 we isolated the non-fluorescent analog PA-SiR (1) (Supplementary Fig. 2). PA-SiR possesses an exocyclic double bond and the two aromatic ring systems are not conjugated, reflected by its λ abs,max value of 290 nm and demonstrated by the X-ray crystal structure of PA-SiR analog 4 (Fig. 1b). However, PA-SiR underwent protonation upon ultraviolet (UV) irradiation in aqueous solution, reestablishing the fluorescent xanthene core of SiR 2 (Fig. 1a, c and Supplementary Figs. 3 and 4). The photoproduct SiR 2 showed an absorption maximum at λ abs,max = 646 nm and emitted at around 660-670 nm. Its extinction coefficient of ε 646 = 90,000 ± 18,000 M −1 cm −1 and fluorescence quantum yield φ = 19.0 ± 2.4% in aqueous buffer (mean ± 95% confidence interval, N = 3 and 4 samples, respectively) were only marginally smaller than those of the previously described SiR-carboyxl 22 (Fig. 1c and Supplementary Table 1). However, 2 is susceptible to nucleophilic attack by water leading to rapid establishment of an equilibrium between 2 and 3 ( Fig. 1a, d, f and Supplementary Fig. 5). Structural modifications on PA-SiR can influence this equilibrium as demonstrated by several synthetized analogs ( Supplementary  Fig. 6). Moreover, both photoactivation of PA-SiR as well as the equilibrium between 2 and 3 are pH sensitive ( Fig. 1e and Supplementary Fig. 7). Photoactivation is prevented by protonation of the aniline groups and is therefore highest at pH values above pH = 6 as revealed by measuring the maximal absorbance at 646 nm reached directly after activation (A max ). The equilibrium between 2 and 3, as measured by recording the absorbance at equilibrium and correcting for A max at 646 nm (A eq ), was shifted toward 3 at higher pH values (Fig. 1e). At physiological pH only about 10% of the activated PA-SiR was present as SiR 2 in comparison to 80% at pH = 6.1. Noteworthy is also the quantitative nature of the photoconversion of PA-SiR, which becomes apparent when following the conversion of PA-SiR to 3 by nuclear magnetic resonance (NMR). These experiments also revealed that the photoactivation is reversible on a time scale of days ( Fig. 1f and Supplementary Fig. 5).
Calculations of the frontier molecular orbitals of model compound PA-SiR 4 and data published on cross-conjugated 1,1-diphenyl alkenes 23,24 indicate that the photoactivation could proceed through a twisted intramolecular charge transfer followed by protonation of the intermediate ( Supplementary  Fig. 3). Further studies are needed to clarify the mechanism of this remarkable reaction and to the best of our knowledge this type of light-induced protonation has not previously been reported for rhodamine derivatives or other xanthenes. Carbopyronine and fluorescein derivatives with such an exocyclic double bond have been described (see Supplementary  Fig. 3 for an overview of related structures and reactions), but have not been reported to undergo light-induced protonation. However, with the many strategies available to tune the HOMOs and LUMOs of xanthene derivatives, we believe it should be feasible to generate photoactivatable rhodamine and carbopyronine derivatives, thereby expanding the scope of this reaction.
The susceptibility of activated PA-SiR toward nucleophiles and its half-life of minutes at physiological pH are a disadvantage of these fluorophores for standard diffraction limited imaging. However, this is less relevant for single-molecule based superresolution microscopy since the observation period of individual fluorophores in SMLM is on the order of milliseconds and the reaction of activated PA-SiR with nucleophiles should not interfere in such experiments. Furthermore, the equilibrium of the reaction of activated PA-SiR with nucleophiles is environmentally sensitive. In fact, when we prepared conjugates of PA-SiR with ligands for protein labeling (Supplementary Figs. 8 and 10) 25-28 , we discovered that PA-SiR-Halo attached to HaloTag (Supplementary Fig. 9) can be very efficiently activated and its fluorescent form 2 is stable over hours at physiological pH, whereas PA-SiR-Halo not conjugated to HaloTag is inefficiently activated and the activated probe decays quickly (Fig. 2a, Supplementary Fig. 10). We attribute the remarkable stability of PA-SiR-Halo attached to HaloTag to specific interactions between HaloTag and the activated probe. It has been previously observed that HaloTag possesses a high affinity toward zwitterionic rhodamine derivatives, including SiR-carboxyl 22 . However, in the absence of structural information on HaloTag labeled with PA-SiR-Halo we cannot provide more detailed insights on the nature of these interactions. This apparent fluorogenicity of the probe should prove beneficial for live-cell imaging as unconjugated PA-SiR-Halo is not fluorescent, which increases the signal-to-background ratio. In addition, PA-SiR-Halo conjugated to HaloTag and photoactivated showed much greater stability toward other nucleophiles such as cysteamine than free PA-SiR (Fig. 2c). The generated fluorescent product had an extinction coefficient of ε 646 = 180,000 ± 30,000 M −1 cm −1 and a fluorescence quantum yield of φ = 29.2 ± 1.2% in aqueous buffer (mean ± 95% confidence interval, N = 3 samples) making it an outstanding fluorophore. Its quantum yield of activation was found to be φ act = 0.86 ± 0.07% at 340 nm and φ act = 0.09 ± 0.04% at 405 nm (mean ± standard error of the mean, N = 3 samples), similar to that of PA-JF 646 (Supplementary Tables 1 and 2) 29 .  (2). The C7-C14 bond is the shortest bond followed by the aromatic bonds exemplified by C4-C5 or C5-C6. Bonds C6-C7 and C7-C8 are considerably longer. Further information can be found in Supplementary Table 11. c Normalized absorption spectra of PA-SiR in PBS (10 μM) before and after UV irradiation as well as emission spectra after activation. d Absorption spectra of PA-SiR in PBS (10 μM) before activation and directly after UV irradiation measured every 1 min, revealing the reaction from 2 to 3. e pH dependence of the equilibrium system of PA-SiR in PBS (10 μM) at different pH after brief photoactivation through UV irradiation.
Normalized absorbance values A max directly after activation and A eq in equilibrium at different pH values are given, reflecting changes in activation (A max ) and equilibrium constant (A eq ). Values displayed are means from three individual measurements, error bars correspond to 95% confidence intervals. Source data are provided as a Source Data file. Live-cell confocal imaging. PA-SiR-Halo possesses a number of properties that make it an attractive candidate for live-cell imaging such as the absence of side-products during photoconversion, the absence of caging groups that affect solubility and permeability, the efficiency of photoactivation and stability of the HaloTag-bound probe compared to unconjugated probe, and its outstanding spectroscopic properties. We, therefore, incubated U-2 OS cells expressing a histone H2B-HaloTag fusion protein with 0.5 μM PA-SiR-Halo for 2 h and imaged the cells prior and after UV activation at 365 nm (Fig. 2d, e). We found that PA-SiR-Halo showed an excellent signal-to-background ratio after activation under no wash conditions (32 ± 5, mean ± 95% confidence interval, N = 119 cells) and that the fluorescence signal after activation was stable over time. In comparison, PA-JF 646 -Halo showed faster activation kinetics but a lower signal-tobackground ratio after activation (13.2 ± 1.9, N = 121 cells) (Supplementary Fig. 12d-f) 20 . Moreover, PA-SiR-Halo was used to image various other intracellular HaloTag fusion proteins ( Fig. 2f- Fixed-cell SMLM. The photophysical properties such as the number of detected photons per frame and fluorophore are decisive for SMLM as the attainable localization precision scales with the inverse square root of the number of detected photons 30 . In order to determine these numbers, we immobilized HaloTag labeled with PA-SiR-Halo on coated glass coverslips and imaged the fluorophore using total-internal reflection (TIRF) microscopy ( Supplementary Fig. 13a). In these experiments, we used a 405 nm laser for photoactivation, generally used to create a sparse subset of fluorescent molecules in SMLM. We found that the photon numbers per particle per frame for PA-SiR-Halo at a power density of 1.2 kW cm −2 suitable for live-cell single-particle tracking were roughly 30% higher than for PA-JF 646 -Halo and considerably higher than those measured for mEos3.2 (Supplementary Fig. 13b) 20 . To test the performance of PA-SiR-Halo in fixed-cell SMLM, we expressed the microtubule binding protein Cep41 as a HaloTag fusion in U-2 OS cells and labeled it with PA-SiR-Halo. The microtubule diameter was determined to be FWHM PA-SiR-Halo = 38.7 ± 7.7 nm (mean ± 95% confidence interval, N = 20 tubules) which corresponds well to the microtubule diameter of 25 nm if one takes the size of Cep41-Halo (74 kDa, ca. 5 nm) into account. Moreover, the data is consistent with previously reported data (Fig. 3a, Supplementary Fig. 14a, g) 8,31 .
Furthermore, we imaged a HaloTag fusion of Nup96 32 , a protein of the nuclear pore complex. Nuclear pores possess a regular circular shape with an internal diameter of about 100 nm 33,34 . Using PA-SiR-Halo labeled Nup96-Halo in fixed U-2 OS cells we were able to reveal the circular structure of the nuclear pore ( Fig. 3c-e). In addition, PA-SiR-Actin was tested for SMLM in fixed COS-7 cells revealing stress fibers and connecting thinner fibers ( Supplementary Fig. 14c). Both PA-SiR-Halo and Live-cell single-particle tracking and SMLM. We next tested the performance of PA-SiR-Halo in live-cell single-particle tracking photoactivated localization microscopy (sptPALM) (Fig. 3b) 36 .
To this end, we chose to track a G-protein coupled receptor involved in cellular signaling that is located in the plasma membrane: beta-2-adrenergic receptor (β2AR) 37 . Using β2AR fused to HaloTag and labeled with PA-SiR-Halo, we were able to track β2AR for several hundreds of milliseconds before photobleaching (Fig. 3b). These track-lengths are considerably longer than what is commonly found for photoactivatable or photoconvertible proteins 20 and similar to what we found for PA-JF 646 -Halo. Furthermore, β2AR labeled with either PA-SiR-Halo or PA-JF 646 -Halo moved with comparable mean speeds ( Supplementary  Fig. 15). PA-SiR-Halo might prove to be beneficial over PA-JF 646 -Halo in intracellular single-particle tracking experiments, where high signal-to-background ratios are required. It has to be noted that the use of (high-intensity) UV light for photoactivation can cause fluorophore degradation and phototoxicity. UV light shows higher phototoxicity than red light 38,39 and should be used only at low intensities and/or low pulse frequencies. Phototoxicity caused by activation of small-molecule synthetic probes with UV light can be due to the UV light itself as well as toxic side products of the photoactivation. While PA-SiR does not release any toxic side products, the risk of conventional phototoxicity remains. Finally, we investigated the potential of PA-SiR-Halo for livecell SMLM. In such experiments, we could follow the fast dynamics of mitochondria (TOMM20-Halo) labeled with PA-SiR-Halo over one minute in 10 s snapshots without artificial narrowing and collapsing of structures ( Fig. 4 and Supplementary Movie 1, Supplementary Fig. 16). The movie and the snapshots taken thereof revealed intermediate formation of thin tubules between mitochondria (blue arrowheads), as was previously seen with SMLM imaging of MitoTracker Red 40 . It was possible to follow fission events of mitochondria highlighting the dynamic network of connecting and disconnecting mitochondrial units (yellow arrowheads). Most interestingly, localizing the fluorophore to the outer membrane of the mitochondria further enabled us to distinguish the outer membrane from the matrix in several cases (red arrowheads), which has not been observed with live-cell SMLM so far. This will eventually help to study interactions between the inner and outer membrane of mitochondria by two color SMLM. This demonstrates that PA-SiR-Halo enables live-cell SMLM of intracellular targets.

Discussion
In summary, PA-SiR is a photoactivatable, cell-permeable, far-red fluorophore that is activated by an unusual light-induced protonation. Its outstanding spectroscopic properties make it well suited for SMLM in both fixed and live-cells and enabled us to create powerful probes for HaloTag and actin. We expect that the exceptional properties of PA-SiR will be exploited in the future to create various other photoactivatable probes for live-cell imaging. perceived to be hollow as Tomm20 is localized to the outer membrane of mitochondria (red arrowheads). The highly dynamic mitochondria form thin tubules between neighboring mitochondria (blue arrowheads) and disconnect (fission) in other areas (yellow arrowheads). Scale bar, 1 μm. Full rolling frame movie available as Supplementary Movie 1 the squared correlation coefficient. They are defined as follows:

Methods
Fluorescence spectra and quantum yield. Fluorescence spectra were measured on a JASCO FP-8600 fluorimeter in 1.4 mL fluorescence cuvettes (Hellma Analytics). Emission spectra were collected from 610 to 1000 nm exciting at 580 nm; excitation spectra were recorded at 664 nm exciting from 400 to 655 nm unless otherwise stated. Fluorescence intensity upon addition of cysteamine was measured on a plate reader (TECAN Spark® 20 M) equipped with a monochromator exciting at 640/10 nm and collecting the emission at 670/10 nm. Quantum yields were determined using a Hamamatsu Quantaurus QY.
Extinction coefficient and quantum yield of activation. Extinction coefficients at 646 nm after activation were calculated from the equilibrium constants (K 2 ) obtained in the 12 s activation experiments (Fig. 2b, Supplementary Figs. 6 and 10, Table 5 and 7), assuming that during the activation the decay (k 2 and k -2 ) is negligible, and the absorbance reached in equilibrium in the saturation experiment ( Supplementary Fig. 11, Table 8). Quantum yields of activation were determined using standard ferroxialate actinometry 41 along with the activation rates determined in the saturation experiments ( Supplementary Fig. 11). This calculation does not take into account the decay kinetics but was good enough to give an estimate of the quantum yields of activation. Saturation experiments under 405 nm irradiation ( Supplementary Fig. 11e, f) were performed using a Solis405C High-Power LED (Thorlabs) coupled to a 3 mm liquid light guide.
Computational chemistry. Optimization of the PA-SiR structure as well as HOMO/LUMO calculations were performed at the B3LYP/6-31G(d) level of theory by using the software package Gaussian 09 42 .
X-ray crystal structure determination. Colorless needle shaped crystals were grown from slow evaporation of a CH 2 Cl 2 /MeOH solution at 4°C. Crystal data and details of the structure determinations are compiled in Supplementary Table 11. A full shell of intensity data were collected at low temperature with an Agilent Technologies Supernova-E CCD diffractometer (Mo-K α radiation, microfocus X-ray tube, multilayer mirror optics). Detector frames (typically ω-, occasionally φ-scans, scan width 0.5°) were integrated by profile fitting 43,44 . Data were corrected for air and detector absorption, Lorentz and polarization effects 43 and scaled essentially by application of appropriate spherical harmonic functions 43,45,46 . Absorption by the crystal was treated numerically (Gaussian grid) 45,47 . An illumination correction was performed as part of the numerical absorption correction 45 . The structures were solved by ab initio dual space methods involving difference Fourier syntheses (VLD procedure) 48,49 and refined by full-matrix least squares methods based on F 2 against all unique reflections 50-53 . All nonhydrogen atoms were given anisotropic displacement parameters. The positions of most hydrogen atoms (except those of the methyl groups, which were treated as variable metric rigid groups with local C 3 symmetry) were taken from difference Fourier syntheses and refined.
CCDC 1942173 contains the supplementary crystallographic data for this paper. These data can be obtained free of charge from the Cambridge Crystallographic Data Centre's and FIZ Karlsruhe's joint Access Service via https://www.ccdc.cam. ac.uk/structures/?
LC-MS analysis. Liquid chromatography mass spectrometry (LC-MS) was performed on a Shimadzu MS2020 connected to a Nexera UHPLC system equipped with a Supelco Titan C18 80 Å (1.9 μm, 2.1 × 50 mm) column. Buffer A: 0.05% HCOOH in H 2 O Buffer B: 0.05% HCOOH in ACN. PA-SiR was dissolved in MQ water (~20 μM). UV irradiation was performed for 1 min in a quartz cuvette (Hellma Analytics) and aliquots were taken to measure LC-MS at defined time points using an analytical gradient from 10 to 90% B within 6 min with 0.5 mL min −1 flow.
1 H NMR analysis PA-SiR. PA-SiR (1 mg, 2.0 μmol) was dissolved in PBS/D 2 O (1 mL, 90:10) and NaOH (1 μL, 5 M) was added to achieve better solubility as PA-SiR was isolated as its TFA salt (pH = 7-8, pH paper). 1 H NMR spectra were measured on a Bruker AV 600 spectrometer at 600 MHz and 298 K. Chemical shifts δ are reported in ppm downfield from tetramethylsilane using the DMSO signal (δ H = 2.50 ppm) instead of the residual deuterated solvent signal as an internal reference. Spectra were measured with NS = 128 using a water suppression presaturation sequence. UV irradiation was performed outside of the spectrometer for the indicated times with a transilluminator (Biometra TI 1, 312 nm). After each UV irradiation step the NMR sample was transferred to the NMR spectrometer.
Protein production and purification. Proteins were expressed in Escherichia coli strain BL21(DE3)-pLysS. Luria-Bertani broth cultures were grown at 37°C to optical density at 600 nm (OD 600nm ) of 0.8, induced by the addition of 0.5 mM isopropyl-β-D-thiogalactopyranoside and grown at 17°C overnight in the presence of 1 mM MgCl 2 . The cells were harvested by centrifugation (4500 g, 10 min, 4°C) and lysed by sonication. The cell lysate was cleared by centrifugation (20,000 g, 20 min, 4°C). All proteins were purified using affinity-tag Ni-NTA (Qiagen) leading to higher than 95% pure proteins (verified by sodium dodecyl sulfate polyacrylamide gel electrophoresis (PAGE) coomassie staining). mEos3.2-Halo was purified analogously but using an additional Strep-Tactin (IBA) column purification step to reach higher purity and following the suppliers' instructions. Proteins were finally concentrated using an Ultra-0.5 mL centrifugal filter device (Amicon) with a molecular weight cut-off according to the protein size and then stored in a glycerol 45% (v/v) solution at −20°C. Proteins were used from glycerol stocks and were further diluted. The amino acid sequences can be found in the Supplementary Methods.
Polyacrylamide gel electrophoresis. HaloTag protein (4 µM) was labeled using PA-SiR-Halo (0, 1, 2, 3, 4, or 6 µM) in activity buffer (50 mM HEPES, 50 mM NaCl, pH 7.3) for 2 h at room temperature. After labeling, the proteins were separated by PAGE (4-20% 10-well Mini-Protean TGX, BioRad) as recommended by the manufacturer and revealed by in gel fluorescence using a ChemiDoc MD Imaging System (BioRad). PA-SiR-Halo labeled proteins were imaged using red epi illumination (695/55 nm). PA-SiR-Halo was activated using the UV-transilluminator of the ChemiDoc MD Imaging System. The solutions were incubated for 5 min at room temperature, then mixed and incubated for additional 20 min at room temperature. The prepared DNA-Lipofectamine complex was added to a glass bottom 35 mm dish with cells at 50-70% confluency. After 12 h incubation in a humidified 5% CO 2 incubator at 37°C the medium was changed to fresh medium. The cells were incubated for 24-48 h before imaging.
Stable cell line establishment. The Flp-In ™ T-REx ™ System (ThermoFisher Scientific) was used to generate stable cell lines exhibiting tetracycline-inducible expression of the gene of interest (GOI). Briefly, pcDNA5-FRT-TO-GOI and pOG44 were co-transfected into the host cell line U-2 OS FlpIn TREx 59 . Homologous recombination between the FRT sites in pcDNA5-FRT-TO-GOI and on the host cell chromosome, catalyzed by the Flp recombinase expressed from pOG44, produced the U-2 OS FlpIn TREx cells expressing stable and inducible the GOI. Selection was performed using 100 μg mL −1 hygromycin B (ThermoFisher Scientific) and 15 μg mL −1 blasticidine (ThermoFisher Scientific). Stable cell lines were seeded on glass bottom dishes as described in the previous section, and induced using 100 μg mL −1 doxycycline (Sigma-Aldrich) for 24-48 h previous to imaging.
Staining. Cells were stained with 0.2-1 μM PA-SiR (1-2 h, 37°C) in phenol-red free DMEM medium supplemented with GlutaMAX, sodium pyruvate and 10% FBS (all Life Technologies), washed with the same medium or PBS (once for 3 min, 37°C) and imaged in the same medium.
Sample preparation for super-resolution microscopy. U-2 OS-CEP41-Halo cells were seeded on 24 mm glass coverslips and stained with PA-SiR as described above. Methanol fixation was performed as follows: growth medium was removed, cells were incubated for 7 min in −20°C cold methanol and washed twice with PBS. Fixed-cell samples were mounted in PBS on cavity slides (VWR ™ ) sealed with twinsil ® 22 (Picodent) and imaged therein. COS-7 cells were seeded on 24 mm glass coverslips and stained with PA-SiR-Actin as described above. The cells were fixed as previously described 60 . Briefly, they were fixed and extracted for 1 min using a solution of 0.3% [w/v] glutaraldehyde and 0.25% [v/v] Triton X-100 in CB buffer (CB: 10 mM MES, pH 6.1, 150 mM NaCl, 5 mM EGTA, 5 mM glucose and 5 mM MgCl 2 ), and then postfixed for 10 min in 2% [w/v] glutaraldehyde in CB. They were treated with freshly prepared 0.1% sodium borohydride for 7 min. Short additional poststaining was performed with 0.5 μM PA-SiR- Actin (1 h, 25°C). Fixed-cell samples were mounted in PBS on cavity slides (VWR ™ ) sealed with twinsil ® 22 (Picodent) and imaged therein.
U-2 OS cells were seeded on 24 mm glass coverslips and transiently transfected (TOMM20-Halo or β-2-adrenergic-receptor-Halo). The next day the cells were stained with PA-SiR-Halo as described above and the coverslips were mounted into attofluor cell chambers (Life technologies) and the imaging medium was supplemented with HEPES (20 mM). Cells were directly imaged after mounting.
Genome-edited U-2 OS cells with Halo-tagged NUP96 32 were seeded on 24 mm round coverslips (No. 1.5 H; 117640; Marienfeld). Cells were cultured under adherent conditions at 37°C, 5% CO 2 and 100% humidity in DMEM (high glucose, without phenol red) supplemented with 10% [v/v] FBS, 2 mML-glutamine, nonessential amino acids, and ZellShield. Before sample preparation, the respective dye was added to the medium to a final concentration of 1 µM and incubated for 2 h. All following incubations were carried out at room temperature and all incubations longer than 1 min were performed on an orbital shaker in the dark to prevent preactivation of the dye. Cells were prefixed in 2. Single-molecule assay. Sample preparation was adapted from two literature procedures 61,62 . Briefly, 18 × 18 mm high-precision coverslips (Carl Roth) were sonicated for 10 min in MQ water, 10 min in acetone, 10 min in MeOH, 10 min in KOH (1 M, prepared from 99.98% purity Carl Roth) and rinsed with MQ water after each step. The coverslips were cleaned with piranha solution (1:3, H 2 O 2 / H 2 SO 4 ) twice for 30 min. After extensive rinsing with MQ water they were dried under a N 2 stream. A solution of 2% [v/v] N- [3-(trimethoxysilyl)propyl]ethylendiamine (Sigma-Aldrich) in dry acetone was prepared and the clean coverslips were immersed in the dark for 1 h. The coverslips were rinsed with acetone, MQ water and then dried with N 2 . A solution of 1 mg biotin-PEG-SVA (MW 5000, Laysan Bio) and 54 mg mPEG-SVA (MW 5000, Laysan Bio) was prepared in 230 μL sodium bicarbonate buffer (10 mM freshly prepared) and applied to three coverslip pairs. After 3 h in the dark the coverslips were washed with MQ water, blow dried with N 2 and stored under N 2 at −20°C. Flow chambers were assembled at need from one glass slide (Carl Roth) and one coated coverslip separated by double sided tape and fixed with epoxy glue.
In total, 100 μL of a 0.2 mg mL −1 solution of streptavidin (Life Technologies) in PBS was applied to the flow chamber and incubated for 10 min. The channel was washed with 400 μL PBS. A solution of SNAP-tag:EGFP:HaloTag (5 μM), fluorophore (2.5 μM), biotin-ligand (5 μM; SNAP-Biotin™ (NEB), HaloTag Biotin (Promega)), in PBS was prepared and incubated for 1 h. In total, 100 μL of a 1:1000-1:500 dilution thereof was applied to the flow chamber and incubated for 10 min. The channel was washed with 400 μL PBS and filled with PBS. They were imaged in TIRF mode using a Leica SR GSD (Supplementary Table 10).
For activation experiment images were taken in the Cy5 (500 ms, ex: 10%), transmission (100 ms) and the GFP channel (100 ms, ex: 5%) consecutively every 9 s. Activation was performed for 50 ms after each acquisition cycle.
Confocal microscopy. Confocal imaging was performed on a Leica DMi8 microscope (Leica Microsystems) equipped with a Leica TCS SP8 X scanhead; a SuperK white light laser, a 355 nm CW laser (Coherent), a HC PL APO 63 ×/1.47 oil objective or a HC PL APO 40.0 ×/1.10 water objective; emission was collected as indicated in Supplementary Table 10. Photoactivation was performed for one frame by using a 355 nm laser. The microscope was equipped with a CO 2 and temperature controllable incubator (Life Imaging Services, 37°C).
For signal to background measurement cells were focused in the transmission channel and z-stacks were recorded with 0.4 μm step size before and after activation. The summed stacks were analyzed as follows: the mean of a rectangular ROI within the nucleus was divided by the mean of a rectangular ROI adjacent to the nucleus.
NUP96-Halo samples were imaged on a custom-built epi-fluorescence microscope with homogenous high-power illumination 63 . The output of a commercial LightHub laser box (Omicron-Laserage Laserprodukte) with 405, 488, 561, and 640 nm laser lines and an additional 640 nm booster laser (Toptica) were focused on a speckle reducer (LSR-3005-17S-VIS; Optotune) and coupled into a multi-mode fiber (M105L02S-A; Thorlabs). The output of this fiber is magnified by an achromatic lens, cleaned up by a quadband filter (390/482/563/640 HC Quad; AHF) and focused into the sample. Fluorescence was collected through a highnumerical aperture (NA) oil-immersion objective (160×/1.43 NA; Leica), filtered by a 700/100 bandpass filter (AHF) and focused onto an Evolve512D EMCCD camera (Photometrics). The focus was stabilized by a total internally reflected IR laser that was focused onto a quadrant photodiode, which was coupled into a closed-loop with the piezo objective positioner. Typically, we acquire 15,000-30,000 frames with 50 ms exposure time and laser power densities of about 13 kW cm −2 . Data were acquired until no more activated fluorophores were observed. The pulselength of the 405 nm laser was adjusted during the acquisition to maintain a similar number of localizations per frame. The different components of the microscope are managed by a field-programmable gate array (Mojo; Embedded Micro) which is controlled using a custom-written plugin for µManager 64 .
Software and image processing. Statistical analysis as well as curve fitting was performed using OriginLab 65 . All images except the NUP96-Halo images were processed with ImageJ/Fiji 66,67 . Super-resolution images and TIRF data from the single-molecule assay were processed with the ImageJ plugin ThunderSTORM 68 . Tracking data were analyzed using the TrackMate plugin 69 . Single-molecule assay data were further processed by a costume written MatLab script provided by Dr. Christian Sieben (EPFL) based on the Crocker, Weeks, and Grier Algorithm 70 . For the quantification of microtubule width, we constructed a perpendicular line profile from a 250 nm long section of the microtubule. We then fitted a Gaussian distribution (bin width 2 nm) to the profile and plotted its FWHM in a boxplot ( Supplementary Fig. 14g, N = 20 line segments per dye).
Live-cell SMLM data were additionally processes using the HAWK plugin using three levels and time grouping, followed by multi-emitter fit in ThunderSTORM allowing for five emitters per fitting region 71 . Costume written MatLab code was used to produce the rolling frame video. The movie presented was convoluted with a Gaussian function (sigma = 12 nm).
The reconstruction of super-resolved images of NUP96-Halo was done using the custom-written software SMAP (Super-resolution Microscopy Analysis Platform, https://github.com/jries/SMAP). First, localizations were detected using a difference of Gaussians algorithm and a dynamic threshold to exclude random signal fluctuations. Then the localizations were fit by a pixelated Gaussian function. Dim localizations (localization precision > 30 nm) and out-of-focus localizations (fitted size of the Gaussian > 160 nm) were filtered out. Localizations that were found within 75 nm of each other in consecutive frames with maximum one frame dark time were grouped into one localization. Images were reconstructed by plotting all localized emitters at the fitted positions as Gaussians with a width proportional to their localization precision.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
The data supporting the findings of this study are available within the paper and its Supplementary Information and are available from the corresponding author upon reasonable request. The source data underlying Figs. 1e and 2c, Supplementary Figs. 9, 10c, 12d-f, 13b, 14g, and 15a-c and Supplementary Tables 1, 2, 5, 7, and 9 are provided as a Source Data file.