{"paper_id":"1071471b-9fe7-4006-a10b-e9c93057e0b5","body_text":"Page 1 of 23 \n \nBiocompatible sulfonium-based covalent probes for endogenous tubulin \nfluorescence nanoscopy in live and fixed cells \nMarie Auvraya, Tanja Koenenb, Olexandr Dybkovc, Henning Urlaubc,d and Gražvydas Lukinavičiusa* \naChromatin Labeling and Imaging group, Department of NanoBiophotonics, Max Planck Institute for \nMultidisciplinary Sciences, Am Fassberg 11, 37077 Göttingen, Germany. \nbDepartment of NanoBiophotonics, Max Planck Institute for Mu ltidisciplinary Sciences, Am Fassberg \n11, 37077 Göttingen, Germany. \ncBioanalytical Mass Spectrometry, Max Planck Institute for Multidisciplinary Sciences, Am Fassberg 11, \n37077 Göttingen, Germany. \ndBioanalytics Group, Institute for Clinical Chemistry, Univ ersity Medical Center Göttingen,  Robert-\nKoch-Str. 40, 37075 Göttingen, Germany. \n \n*Email: grazvydas.lukinavicius@mpinat.mpg.de  \n \nAbstract \nFluorescent probes enable the visualization of dynamic cellular processes with high precision, \nparticularly when coupled with super-resolution imaging techniques that surpass the diffraction limit. \nTraditional methods include fluorescent protein fusion (e. g., GFP) or organic fluorophores linked to \nligands targeting the protein of interest. However, these approaches often introduce functional \ndisruptions or ligand-associated biological effects. Herein, we address these challenges by developing \ncovalent fluor escent probes for endogenous tubulin, a critical cytoskeletal protein involved in \nprocesses such as cell movement, division, and biomolecule trafficking. Using well -known tubulin \nbinder cabazitaxel and cell permeable fluorophore silicon-rhodamine—as a basis, we introduce a novel \nbiocompatible cleavable linker containing a sulfonium center.  This allowed the construction of t he \noptimized probe, 6 -SiR-o-C9-CTX, demonstrating excellent cell permeability, fluorogenic properties, \nand the ability to covalently lab el tubulin across various  human cell lines. Importantly, the targeting \nmoiety could be washed out while preserving tubulin staining, ensuring minimal disruption of tubulin \nfunction. This labeling technique is compatible with STED nanoscopy in both live and  fixed cells, \noffering a powerful high-resolution imaging tool. \n  \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 27, 2025. ; https://doi.org/10.1101/2025.01.27.635008doi: bioRxiv preprint \n\nPage 2 of 23 \n \nIntroduction \nFluorescence microscopy offers a possibility to visualize dynamic processes in living cells. In the \nrecent decades, super -resolution microscopy has become an essential tool to understand these \nbiological processes with an unrivalled precision by overcoming the diffraction limit .1 These methods \nheavily relies on labeling techniques , which show a minimal influence on the function of the studied \nbiomolecules2,3. In living cells, two strategies are mostly used for labeling of a protein-of-interest (POI). \nFirst, the fusion with a fluorescent protein (for example, the well-known Green Fluorescent Protein – \nGFP) or with a self-labelling tag that hosts a synthetic fluorophore.4 Nevertheless, these strategies add \na large extra domain to the POI: for example, GFP (27 kDa ) has a length of 4.2 nm and a diameter of \n2.4 nm.5 This large size mainly causes two problems : mislocalization and function disruption of POI.6 \nThe second  approach involves using fluorescent probes, which consist of an organic fluorophore   \nconjugated via a linker to a ligand that targets and non-covalently binds the POI. This method ensures \nthe fluorescent tag is as small as possible. However, ligands can o ften exhibit biological effects and \nmay occupy active sites on the protein, potentially altering its function.  In this study, we aimed to \nexplore an alternative labeling strategy for endogenous proteins that minimizes impact on their native \nfunction. In 2009, Hamachi introduced the ligand -directed labeling strategy.7 This approach relies on \nproximity induced reaction between a cleavable linker and POI. Once the probe binds to the POI, the \nnucleophilic side chain of an amino acid nearby the binding site can react with the cleavable linker to \ncreate a covalent bond between the tag and the protein, while releasing the ligand.  Over the last \nfifteen years, cleavable linkers that can react wi th various amino acids have been developed. 8,9,10,11,12 \nDespite the tremendous potential of this approach for fluorescence microscopy, it has not been \nextensively applied. In particular, its application in the field of f luorescence microscopy is mostly \nrestricted to extracellular targets, like membrane receptors. 13-16 The main reason is that it is \nchallenging to find a compromise between reactivity and stability of the cleavable linker  inside cells. \nIn addition, making these probes cell-permeable can be an issue.  \nIn this study, we developed a s eries of covalent fluorescent probes for endogenous tubulin. This \nprotein is highly abundant in mammalian cells, in which it represents 3-4% of the total proteins and up \nto 10 % in brain.17 Two subunits α- and β-tubulin form heterodimers and polymerize into microtubules, \none of the main components of the cytoskeleton.  Both proteins exist in numerous isotypes encoded \nby different genes. In human, there are 9 genes for α-tubulin and 10 for β-tubulin.18 In addition, tubulin \ncan undergo post-translational modifications, leading to a huge variety of forms in cells.19 It is therefore \ninvolved in many different cellular processes like cell movement, division or trafficking of biomolecules. \nWe developed fluorescent probes based on widely used tubulin binders —taxanes, which are well -\n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 27, 2025. ; https://doi.org/10.1101/2025.01.27.635008doi: bioRxiv preprint \n\nPage 3 of 23 \n \nknown anticancer drugs. To enhance their utility, we introduced a novel biocompatible cleavable linker \ncontaining a sulfonium center. Through in vitro  and in cellulo  evaluations, we identified  silicon-\nrhodamine (SiR) probe  6-SiR-o-C9-CTX as an outstanding fluorogenic, cell -membrane-permeable \ncandidate. Importantly, we demonstrated that the ligand c an be washed out while retaining robust \ntubulin staining, thus providing a labeling technique that minimally impacts protein structure and \nfunction. This probe enables covalent labeling of endogenous tubulin across various cell lines, making \nit suitable for stimulated emission depletion (STED) nanoscopy in both live and fixed cells.  \nResults  \n▪ Design and synthesis of the probes \nTubulin-covalent probes are made of three parts: a dye, a ligand to direct the probe to the POI and a \ncleavable electrophile to create a covalent bond between the POI and the fluorophore, while releasing \nthe ligand (Figure 1A).  \nMany cleavable electrophiles have been developed in the past ten years, highlighting the \nimportance of finding a compromise between the reactivity and the stability of probe. We focused our \nattention on sulfonium, a biocompatible electrophile, that is endogenously present at high \nconcentrations in cells across multiple species. 6-10 In addition, recent proteomic studies by Z. Li \ndemonstrated promising res ults for ligand -directed chemistry with para-benzyl sulfonium. 20,21 \nRhodamine derivatives are among the most popular dyes to develop probes for live-cell imaging. They \nare in equilibrium between two forms: a nonfluorescent hydrophobic spirolactone  (OFF state) and a \nfluorescent hydrophilic zwitterion  (ON state)  (Figure 1A ). In aqueous medi a, aggregation drives \nrhodamine to the spirolactone form, which is cell permeable. Once bound to its target, the interactions \nwith the protein will push equilibrium towards the fluorescent zwitterionic form. One of the most \nprominent examples is SiR , which displays outstanding fluorogenicity and cell permeability. 22 In our \nprevious studies, we showed that the best performing ligand for tubulin in combination with SiR \nisomer-6 derivatives was cabazitaxel.23,24 We thus performed a mini screening using silicon-rhodamine \nas dye and cabazitaxel (CTX) as ligand , and fine-tuned the structure by probing three different \nregioisomers (ortho, meta and para) of the cleavable electrophile (Figure 1A and ESI Figure S1). \n \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 27, 2025. ; https://doi.org/10.1101/2025.01.27.635008doi: bioRxiv preprint \n\nPage 4 of 23 \n \n \nFigure 1. Structure, synthesis and properties of the covalent probes. (A) Structure of tubulin probes \nfeaturing a sulfonium-based cleavable linker and silicon-rhodamine (SiR) which can toggle between ON \nand OFF states. (B) Synthetic path of the covalent probes (C)  Photophysical properties of the probes. \nLeft: Absorption and emission spectra of 6-SiR-o-C9-CTX in presence of tubulin (blue), PBS (green), in \npresence of BSA (black)  and PBS containing 0.1% SDS (red). Spectra are represented as averages of \nthree independently repeated experiments (N=3). Middle: Fluorescence increase upon tubulin binding, \nas compared to PBS. Right: Percent of dye open when bound to tubulin . Data are presented as mean \nof triplicate with standard deviation. \nThe probes were synthesized via a convergent 7-step synthetic path with overall yields between 20 \nand 43 % ( Figure 1B). First, thiols 7 were synthesized from the corresponding bromine derivatives, \nperforming a nucleophilic substitution with thiourea followed by hydrolysis. Afterwards, thiols reacted \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 27, 2025. ; https://doi.org/10.1101/2025.01.27.635008doi: bioRxiv preprint \n\nPage 5 of 23 \n \nwith bromine derivatives of various chain lengths to give thioethers 8-13. Sulfoniums with free amines \nand carboxylic acids 14-19 were obtained with excellent yields via a one pot reaction in a mixture of \nformic and acetic acids in presence of an excess of methyl trifluoromethanesulfonate. The synthesis \nended with a four -step, one pot reaction that links the cleavable electrophile to both the ligand and \nthe dye. This last step allowed to get expected compounds with good yields ranging from 35 to 58%. \nOverall, this synthetic path is highly convergent, which allows a rapid access to structural diversity.  \n▪ Photophysical properties  \nPhotophysical properties of the probes were investigated in phosphate-buffered saline (PBS), in \nPBS containing 0.5 mg/ml bovine serum albumin (BSA), in PBS containing 0.1% sodium dodecyl sulfate \n(SDS) or in presence of tubulin after 4h at 37°C to ensure complete tubulin polymerization (Figure 1C, \nESI Figures S2, S3 and Table S1). As mentioned previously, SiR-based probes are in equilibrium between \ntwo forms: a spirolactone and a zwitterionic forms. Multiple studies demonstrate that SiR equilibrium \nis highly sensitive to the environment.22,25,26 In aqueous media ( e.g. PBS), the SiR is mainly in close d \nspirolactone form and tends to form aggregates, which explains the low absorban ce and emission \nobserved in this media for most of the probes. However, probe 5, containing a PEG linker, exhibits a \nfluorescence at least six times higher than other probes in aqueous media (ESI Figure S3). This linker \nprevents aggregation by increasing w ater solubility, and therefore push the equilibrium towards the \nopen fluorescent form. In the presence of SDS, aggregation is completely inhibited, with SiR adopting \na zwitterionic form. This results in similar photophysical properties across all derivatives. Upon binding \nto tubulin, a shift in the spirocyclization equilibrium causes a 4- to 14-fold fluorescence enhancement \nrelative to PBS for all probes except probe 5. For probe 5, only a modest increase in fluorescence (1.4-\nfold) is observed , due to its r esidual fluorescence in aqueous media . Among the probes, probe 2 \ndemonstrates the highest fluorescence enhancement upon tubulin binding compared to PBS (14.3 -\nfold, Table S1). Notably, the non -target protein bovine serum albumin (BSA) induces only minimal \nchanges in absorbance and fluorescence (ESI Figure S3). \nWe assumed that the probes fluorescence in presence of PBS containing 0.1% SDS corresponds to \n100 % of SiR fluorescent zwitterion. This  allowed to estimate the equilibrium shift upon  tubulin \nbinding. We estimate that probes can reach up to 49 % of fluorescent zwitterion content once bound \nto tubulin (Figure 1C). Overall, probe 2 possesses the best photophysical properties, being the most \nfluorescent (49% of probe in ON state after binding to tubulin) and the most fluorogenic (14.3 -fold \nfluorescent enhancement in presence of tubulin compared to PBS). \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 27, 2025. ; https://doi.org/10.1101/2025.01.27.635008doi: bioRxiv preprint \n\nPage 6 of 23 \n \n• In vitro reaction with purified tubulin  \nNext, we checked the stability of the probes in aqueous media (PBS) at 37°C as some probes \ncontaining cleavable linkers might be stable only for a few hours in solution.12 Only minor degradation \nwas observed, as more than 75% of the probes remain after 48h at 37°C (ESI Figure S4).  Major \ndegradation pathways were hydrolysis (approx. 5-10% after 48h) and intramolecular reaction (approx. \n5% after 48h), as the ligand contains some free nucleophilic groups. Prior to live -cell experiments, \nreactions between purified tubulin from porcine brain and the probes were conducted in vitro at 37°C \n(Figure 2, ESI Figure S5). Probes were used in 4-fold excess compared to tubulin in order to assess the \nselectivity of the reaction.  The reaction was monitored by  SDS–polyacrylamide gel electrophoresis \n(SDS-PAGE) and in-gel fluorescence analysis.  \n \nFigure 2. In vitro labeling experiments of purified  pig (Sus scrofa) tubulin (0.5 mg/mL  ̴ 5 µM) with \nthe covalent probes (20 µM) over 48h at 37°C in General Tubulin Buffer . (A) Reaction between the \nprobes and tubulin . (B) Representative SDS -PAGE analysis of the labeling reaction. The gel was \nanalyzed by in-gel fluorescence imaging (up) and stained with Coomassie Brill iant Blue (down). ( C) \nTime-course of in vitro labeling of tubulin with the covalent probes. The experiments were performed \nin triplicate to obtain mean and standard deviation values (shown as error bars). Table contains mean \nvalues and 95% confidence intervals (Cl) obtained from fitting data points to one phase association \nmodel from GraphPad Prism . (D) Structure of β-tubulin (from pig – PDB: 5SYF) with Taxol. The main \nlabeling site is highlighted in blue (Cys 356), and the minor labeling site in red (Cys 12).  \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 27, 2025. ; https://doi.org/10.1101/2025.01.27.635008doi: bioRxiv preprint \n\nPage 7 of 23 \n \nApparent reaction rate was ranging from 0.12 to 0.28 h-1 and it corresponds to a second order rate \nconstant approx. 2 to 5 L•mol-1•s-1, which is comparable to  previously reported LDAI (ligand-directed \nacyl-imidazole chemistry).27 On average, reactions with ortho-isomers were around 50% faster than \nwith meta or para isomers. Compounds can be divided into t wo groups  based on the degree -of-\nlabeling (DOL) . Probes 1-4 reach DOL between 0.8 and 0.9, which correspo nds to a n excellent \nefficiency, close to ideal DOL of 1. Decent labeling yield (around 50%) can be achieved within 2-3 hours \nfor Probe 1 and 6-SiR-o-C9-CTX (2). Probes 5 and 6 demonstrate DOL > 1 suggesting multiple labeling \nsites and a lower selectivity. \nAfterwards, we also checked that the bond between the dye and t ubulin was stable in aqueous \nconditions (ESI Figure S6). Indeed, no degradation was observed after 48h in PBS (pH=7.4) at 37°C. The \nlabeling site(s) were then determined by proteolytic in-gel digestion followed by mass spectrometry. \nSeven different peptides were observed with a +573 Da modification , corresponding to five different \nsequences labeled with SiR (ESI Figures S7 & S8, Table S2). For all these peptides, the modified amino \nacid was a cysteine, suggesting that sulfonium ligand directed chemistry is selective for cysteine . The \nhigh number of observed peptides can be explained by the numerous tubulin iso types and isoforms \ncontained in the sample. Indeed, most of the identified peptides come from different tubulin isotypes \nbut correspond to the same modified amino acid. Overall, a primary labeling site and three secondary \nlabeling sites were identified. The major labeling site was Cys356, located within the \nTAVCDIPPR/VAVCDIPPR peptide of β-tubulin (ESI Figures S7 & S8, Table S2). Additionally, a secondary \nlabeling site was identified in β-tubulin at Cys12 within the peptide EIVHIQAGQCGNQIGAK, which MS1 \nprecursor abundance was approximately ten times lower (ESI Figure S7). Modeling demonstrates that \nboth labeling site s are located near the taxane binding pocket inside lumen of microtubule , as \nillustrated in Figure 2D, Figure S11 and Movie 1. \nSurprisingly, two minor labeling sites were also identified on α-tubulin (Cys347 and Cys376) despite \ncabazitaxel is known to bind β-tubulin. It might be explained by the close proximity of α and β subunit \nin microtubules. Nevertheless, the precursor abundance for the corresponding peptides was 20 times \nlower than for the major labeling site on β-tubulin, suggesting a really low labeling efficiency at these \ntwo positions on α-tubulin (ESI Figure S7). \nDespite this experiment was done with porcine tubulin, there is high structural homology of tubulin \namong species. 28 The sequences of the different α and β-tubulin isotypes are highly conserved \nbetween human and pig as shown by sequence alignments and crystal structures (ESI Figures S9-11). \nAll the detected labeled peptides can be also found in human tubulin , including  the main labeled \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 27, 2025. ; https://doi.org/10.1101/2025.01.27.635008doi: bioRxiv preprint \n\nPage 8 of 23 \n \npeptide (AVCDIPPRGL), which is highly conserved. This suggests that analogous labeling might occur in \nother organisms  and we can expect similar labeling sites using human cell line for living cells \nexperiments. \nOverall, the in vitro tubulin labeling study highlighted the great potential of probes 1 and 2 for \ntubulin labeling in living cells. We have also shown that Cys356, which is nearby the taxane binding site \non β-tubulin, is the main labeling site of the probe. \n▪ Tubulin labeling in living cells \nProbe cytotoxicity was first evaluated on HeLa CCL cells  (ESI Figure S12, Table S3). After 24h, most of \nthe probes results in toxicity in the nanomolar range, with cytotoxicity thresholds ranging from 31.25 \nto 250 nM. These values are consistent with the toxicity of previously reported tubulin probe SiR-CTX \n(cytotoxicity threshold : 62.5 nM) .24 Taxane derivativ es like cabazitaxel are known to stabilize \nmicrotubules and inhibit cell proliferation by blocking the cell cycle just before mitosis.29 This leads to  \napoptosis and an increase in SubG1 phase population of cells. We hypothesize that this measurement \ncould be useful to assess and compare the cell permeability of the probes. According to the obtained \nvalues, ortho isomers are more permeable compared to meta and para isomers. Once more, probe 5 \nis an exception and seems to be less cell-permeable than other ortho isomers. PEG linker increases the \nsolubility while reducing the cell permeability, by pushing the equilibrium towards the open form. It is \nconsistent with the spectroscopic data showing that for this compound, most of probes are in \nzwitterionic form in aqueous media.  \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 27, 2025. ; https://doi.org/10.1101/2025.01.27.635008doi: bioRxiv preprint \n\nPage 9 of 23 \n \n \nFigure 3. Tubulin labeling in living cells. (A) Live human dermal fibroblasts stained with probes (1 µM \nin OptiMEM) and Hoechst 33342 nucleic acid dye (1 µg/mL) for 4h. Confocal microscopy images were \nacquired using LEICA SP8. Gray channel (λex= 633 nm, λem= 650-710 nm) corresponds to probe staining \nand blue channel (λex= 405 nm, λem= 415-480 nm) corresponds to Hoechst 33342 staining. Scale bar = \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 27, 2025. ; https://doi.org/10.1101/2025.01.27.635008doi: bioRxiv preprint \n\nPage 10 of 23 \n \n20 μm . (B) Live -cell imaging of U -2 OS and HeLa CCL cells stained with 6-SiR-o-C9-CTX (1 µM in \nOptiMEM) and Hoechst 33342 (1 µg/mL) for 4h (in case of U-2 OS, verapamil 10 µM was also added). \nCells were washed three times with HBSS and imaged in DMEM+. Confocal microscopy images were \nacquired using LEICA SP8. Gray channel (λex= 633 nm, λem= 650-710 nm) corresponds to probe staining \nand blue channel (λex= 405 nm, λem= 415-480 nm) corresponds to Hoechst 33342 staining. Scale bar = \n20 μm. (C) Analysis of cell lysates (human fibroblasts (H. Fibro.), U-2 OS and HeLa CCL) using SDS-PAGE \nand Western blot. Predominant single fluorescent band demonstrates selective β-tubulin labeling with \n6-SiR-o-C9-CTX (3 µM for 1h) in living cells. \nTo find out the best probe for fluorescence microscopy, a mini screening was performed on three \ndifferent cancerous and non-cancerous human cell lines (dermal fibroblasts, HeLa CCL and U-2 OS) in \nliving cells (Figure 3 A,B & ESI Figures S13-14). With U-2 OS cells, verapamil was added to inhibit efflux \npumps.30 For probes 4 and 5, the staining was weak, which  is in accordance with the lower cell \npermeability suggested by cytotoxicity experiments.  Probes 1,3 and 6 stained microtubules in living \nfibroblasts but only a weak staining was obtained with U-2 OS and HeLa. For all conditions tested, \nprobe 2 (6-SiR-o-C9-CTX) was the best microtubules stain. Next step was to check whether the bond \nbetween the protein and the dye is covalent. After staining various cell lines with 6-SiR-o-C9-CTX, cells \nwere lysed (Figure 3C) and lysates were analyzed by SDS-PAGE. In-gel fluorescence image revealed the \npresence of a strong fluorescent band at 55 kDa for 6-SiR-o-C9-CTX in the three cell lines tested, \nwhereas nothing was observed for the non -covalent probe SiR-CTX. Western blot analysis confirmed \nthat the fluorescent band corresponds to β-tubulin. Additional minor bands with masses exceeding 55 \nkDa were also detected, which we hypothesize could represent different β -tubulin isotypes and  \nisoforms, and post-translational modifications of them.28  \nThe confirmation that the bond is covalent for 6-SiR-o-C9-CTX and not for SiR-CTX was further obtained \nby washing experiments (Figure 4). Living U-2 OS cells were stained with either 6-SiR-o-C9-CTX or SiR-\nCTX at the same concentration in presence of verapamil. After 4h of incubation, extensive washing \nwas performed. Images acquired before the washing process show a nice microtubule staining for both \nprobes. After washing, the staining performed with 6-SiR-o-C9-CTX was still clearly visible without any \nsignificant loss of intensity, where as for cells treated with SiR-CTX, the staining was not visible \nanymore, meaning the probe was fully washed out.  \n \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 27, 2025. ; https://doi.org/10.1101/2025.01.27.635008doi: bioRxiv preprint \n\nPage 11 of 23 \n \n \nFigure 4. Cleavable linker allows covalent tubulin labeling and prevents taxane cytotoxicity. (A) Live-\ncell imaging of U-2 OS stained with probe (6-SiR-o-C9-CTX – first row, SiR-CTX – second row, 1 µM in \nOptiMEM), Hoechst 33342 (1 µg/mL) and verapamil (10 µM) for 4h before and after extensive washing \n(every 10 min over 2h). Note, signal intensity is enchanted 2-folds for 6-SiR-o-C9-CTX (both before and \nafter washing). Confocal microscopy images were acquired using LEICA SP8. Gray channel ( λex= 633 \nnm, λem= 650-710 nm) corresponds to probe staining and blue channel (λex= 405 nm, λem= 415-480 nm) \ncorresponds to Hoechst 33342 staining . Scale bar = 20 μm . The experiment was reproduced three \ntimes with similar results.  (B) Cell cycle perturbation induced by 6-SiR-o-C9-CTX (1 µM in OptiMEM) \nafter 24h with or without extens ive washing after 4h incubation.  Results are averages of four \nindependent experiments (N=4) and presented as means with standard deviations. Cell cycle diagram \ncreated with BioRender.com. \nWe investigated cytotoxicity of 6-SiR-o-C9-CTX probe 24 hours after staining cells, with or without \nwashing after 4h incubation . In non-washed cells, a significant cytotoxic effect was observed, with \napproximately 40% of the SubG1 phase  population. However, for extensively washed cells, no \nsignificant cytotoxicity was observed compared to the control (Figure 4B, ESI Figure S15). These results \ndemonstrate that extensive washing effectively removes the ligand released during proximity-induced \nlabeling, recovering cells from taxane induced phenotype and leaving labeled microtubules inside \nliving cells. \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 27, 2025. ; https://doi.org/10.1101/2025.01.27.635008doi: bioRxiv preprint \n\nPage 12 of 23 \n \n \nFigure 5. Nanoscopy of covalently labeled tubulin  in living and fixed cells . (A) Confocal and STED \nimages of microtubules in HeLa, U-2 OS or human dermal fibroblast (H. fibro) cells stained with 6-SiR-\no-C9-CTX (1 µM in OptiMEM) for 4h. Images show excellent tubulin labeling in live and glutaraldehyde \nfixed cells. Images acquired with Abberior Expert Line. Scale bar = 2 µm. (B) Confocal and STED image \nof microtubules in glutaraldehyde fixed human dermal fibroblasts stained before  fixation with 6-SiR-\no-C9-CTX (1 µM in OptiMEM) for 4h. (C) Line profile plots of fluorescence signal of the area marked in \nimages from ( B). (D) Quantitative analysis of apparent microtubules FWHM in human dermal \nfibroblasts live or fixed. The line corresponds to the mean and each dot corresponds to the result of \none single measurement, N ≥ 12 from three different fields of view. \nFinally, we wanted to investigate if this new covalent probe for tubulin, 6-SiR-o-C9-CTX, was suitable \nfor super-resolution microscopy and in particular stimulated emission depletion (STED) microscopy. \nThe fluorescence of SiR can be inhibited using a 775 nm depletion laser, leading to an enhanced \nresolution. We stained three different cell lines ( human dermal fibroblasts, U-2 OS and HeLa) with 6-\nSiR-o-C9-CTX and successfully acquired STED images of living cells or after fixation with glutaraldehyde \n(Figure 5). We obtained microtubule apparent FWHM 38 ± 9 nm for fixed fibroblasts and 49 ± 13 nm \nliving fibroblasts , which corresponds to a 8-9-fold resolution enhancement compared to confocal  \n(Figure 5D, ESI Figure S16). In addition, the probe was photostable enough to acquire a 500 seconds \ntimelapse (20 frames, Movie 2). \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 27, 2025. ; https://doi.org/10.1101/2025.01.27.635008doi: bioRxiv preprint \n\nPage 13 of 23 \n \nDiscussion \nWe developed a serie s of covalent fluorescent probes for endogenous tubulin based on proximity \ninduced reactivity. To do so, biocompatible benzyl sulfoniums were used as cleavable linkers. Probes \nwere obtained thanks to a highly convergent synthetic path, which will allow this strategy to be easily \nextended to other dyes and other ligand s in the near future.  Ortho isomers showed enhanced \nphotophysical properties, reactivity and cell permability over meta and para isomers. Ortho-benzyl \nsulfonium offers a good balance between stability and reactivity: they posses a similar labeling rate as \nLDAI, but can be used inside living cells because they are stable enough. We demonstrated that 6-SiR-\no-C9-CTX can be used to covalently label endogenous tubulin in living cells after only few hours  and \nthat this probe is suitable for STED nanoscopy. Probe is likely to be useful for labeling of microtubules \nin multiple species because the labeling site is mapped to highly conserved sequence AVCDIPPR of -\ntubulin. Introduction of cleavable linker generated tubulin probe that, after labeling releases cytotoxic \ntaxane, which can be washed out  and has a minimal impact on microtubule structure and function. \nOverall, this work underscores the immense potential of ligand -directed labeling and sulfonium \nchemistry for the covalent labeling of endogenous proteins. Our methodology offers new \nopportunities for advanced imaging techniques like nanoscopy and sets the stage for broader \napplications in cellular biology and beyond.  \nMethods \nGeneral procedure for the final step of probe synthesis \nIn a 1.5 mL Eppendorf, 6-SiR-COOH (1.0 eq.) was solubilized in dry DMSO (0.1 M), and DIPEA (10 eq.) \nwas added. After 5 min, TSTU (1.05 eq.) was added. After 10 min, the formation of activated acid was \ncontrolled by LC-MS and the solution of amine-bearing sulfonium in DMSO (1.3 eq.) – obtained at the \nprevious step - was added. The formation of the amide was controlled by LC-MS (it typically takes 15-\n20 min). Once the amide bond was formed, TSTU (1.5 eq.) was added, followed by CTX-NH3.CF3OCO2 \n(3.0 eq.) and DIPEA (10 eq.). The formation of the final product was controlled by LC-MS. The reaction \nwas quenched by adding trifluoroacetic acid (50 µL). The crude was diluted in ACN (1 mL) and purified \nby reverse -phase HPLC (Device A  (see ESI ), H 2O+0.1%TFA/ACN, linear gradient 70:30 to 0:100 , 40 \nmL/min). The fractions contain ing the product were lyophilized  and the yields were determined as \nfollowed:  \nThe fluorescent probes were dissolved in a precise amount of DMSO-d6 (600 µL) and were transferred \nto an NMR tube to obtain 1H spectra. Afterwards the contents of the NMR tubes were transferred to \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 27, 2025. ; https://doi.org/10.1101/2025.01.27.635008doi: bioRxiv preprint \n\nPage 14 of 23 \n \nan Eppendorf and were considered as stock solution. Two 1 µL samples were taken from stock solution \nand were diluted in Eppendorf with 59 µL of PBS containing 0.1% SDS (60-fold dilution). After 15 min, \nabsorption of 2 µL of the diluted samples were measured on nanodrop (Nanodrop 1000, Peqlab) with \n1 mm optical path. The measured absorption intensity values at the dyes absorption maxima value \nwere averaged and concentration of the stock solution was determined according to the equation \n(considering that the molar absorption coef ficient of the probe is comparable to that of 6 -SiR-COOH \nɛ(λMAX) = 90 000 L.mol-1.cm-1)23: \n𝐶 =\n𝐷𝑖𝑙𝑢𝑡𝑖𝑜𝑛 𝑓𝑎𝑐𝑡𝑜𝑟∗𝐴\nɛ𝑙    (1) \nwhere C is the concentration of stock solution; A the sample absorption, ɛ the extinction coefficient of \nthe dye in PBS containing 0.1% SDS and l the path length. \nOnce the concentration of the stock solution was measured the amount of the obtained fluoresce nt \nconjugate could be calculated by following equation:  \n𝑛 = 𝐶 ∗ 𝑉 (2) \nwhere C is the concentration of stock solution and V the volume of stock solution. \nFinally, the yield could be determined by the classical equation:  \n𝜂 (%) =\n𝑛𝑒𝑥𝑝\n𝑛𝑡ℎ\n∗ 100 (3) \nwhere nexp is the obtained amount of the isolated product (mol) and nth – maximal theoretical amount \nof the product in the reaction (mol). \nNote, the final probes of ortho isomer are not fully stable under the reaction conditions. Therefore, it \nis better to stop the reaction after 30-45 min even if the conversion is not full. \nMeasurement of labeling kinetics \nTubulin from porcine brain (1 mg, Cytoskeleton, #T240, >99% Pure) was solubilized in 100 µL General \nTubulin Buffer (Cytoskeleton, #BST01) supplemented with 1 mM of GTP (Thermo Fisher, #R0461) to \nget a 10 mg/mL solution. This solution was aliquoted, samples were snap-frozen in liquid nitrogen and \nstored at -80°C prior to use. \nBefore the experiment, an aliquot was thawed using a water bath at room temperature. Once liquid, \nthe sample was immediately placed on ice. In parallel, a solution of General Tubulin Buffer containing \n1 mM of GTP was prepared and placed on ice.  \nThe reaction was conducted mixing tubulin (0.5 mg/mL  ̴ 4.5 µM) and the probe (20 µM) in General \nTubulin Buffer containing 1mM of GTP at 37°C for 48h. Aliquots were taken at different timepoints, \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 27, 2025. ; https://doi.org/10.1101/2025.01.27.635008doi: bioRxiv preprint \n\nPage 15 of 23 \n \nimmediately mixed with 1/3 volume of 4x SDS sample buffer (Tris-HCl pH=6.8: 0.2M, SDS: 8.0% (w/v), \nBromophenol Blue: 0.6 mM, glycerol: 5.4 M + 50 µL of β-mercaptoethanol per mL of solution prior to \nuse), and boiled for 5 min at 95°C. \nThe different samples were loaded on 4 -15% Mini-PROTEAN® TGX™ Precast Protein Gels  (Biorad, \n#4561086). After electrophoresis in Mini-PROTEAN® Tetra Cell using SDS-PAGE running buffer (0.25 M \nTris HCl, 1.92 M glycine and 1% (w/v) sodium dodecyl sulfate (SDS) pH=8.3), fluorescence images were \nrecorded using Amersham Imager 600 RGB. Quantitative data analysis was then performed using the \n“gel analysis” function of Fiji7. \nAll experiments were performed in triplicate on different days. A gel was performed with the 48h \ntimepoints (3 for each compound) for all the compounds to calibrate the values accordingly. Finally, \nall data were calibrated with 6-SiR-o-C9-CTX for which the DOL was found to be 0.94 (see below).  \nConsidering one reactant was in excess, data obtained were fitted using a single exponential (one -\nphase association model from GraphPad Prism - Y=Y0 + (Plateau -Y0)*(1-exp(-K*x)), imposing the \nconstraint Y0=0) to obtain the degree-of-labeling and the reaction rate for each compound.  \nSample preparation for determination of DOL (degree-of-labeling) \n6-SiR-o-C9-CTX (20 µM) and Tubulin (1 mg, 0.5 mg/mL  ̴ 4.5 µM, Cytoskeleton, #T240, >99% Pure) were \nmixed together in General Tubulin Buffer (Cytoskeleton, #BST01) supplemented with 1 mM of GTP \n(Thermo Fisher, #R0461) for 24h at 37°C. The solution was cooled at 4°C for 1h to depolymerize tubulin. \nThe probe that did not react was then removed using PD MidiTrap G-25® (Cytivia). Procedure used was \nthe one advised by the supplier, using BRB80 (80 mM PIPES pH=6.8, 1 mM EGTA, 1 mM MgCl 2) as \neluting buffer and conducted at 4°C. The solution containing the protein was concentrated to the \ndesire volume using Vivaspin® Turbo 4 (Sartorius, MWCO = 10 kDa, #VS04T02). \nDOL was determined measuring A 280nm and A 652nm with a NanoDrop ND -1000 spectrophotometer \n(Peqlab) to determine dye and protein concentrations to calculate DOL according to this formula: \n𝐷𝑂𝐿 =\n[𝐷𝑦𝑒]\n[𝑃𝑟𝑜𝑡𝑒𝑖𝑛]  =  \n𝐴652 𝑛𝑚\n𝜀(6−𝑆𝑖𝑅,   652 𝑛𝑚)\n𝐴280 𝑛𝑚−𝐶𝐹280 𝑛𝑚(6−𝑆𝑖𝑅)∗𝐴652 𝑛𝑚\n𝜀(𝑇𝑢𝑏𝑢𝑙𝑖𝑛,   280 𝑛𝑚) \n (4) \nA solution of 2% SDS (0.5 µL) was added to the solution of protein (5 µL) to ensure that the dye was \nfully open (three different solutions were prepared). After 30 min at room temperature, absorbance \nwas measured at 280 nm and 652 nm. Considering ɛ(6-SiR, 652 nm) = 90 000 L.mol -1.cm-1, ɛ(Tubulin, \n280 nm) = 110 000 L.mol-1.cm-1 and CF280(6-SiR) = 0.147, the DOL for this sample was found to be 0.94. \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 27, 2025. ; https://doi.org/10.1101/2025.01.27.635008doi: bioRxiv preprint \n\nPage 16 of 23 \n \nProteomic study \nSample preparation \nOne µg of the 6-SiR-o-C9-CTX-labeled tubulin (as described above) was separated on 4 -15% Mini-\nPROTEAN® TGX™ Precast Protein Gels (Biorad, #4561086) and stained with Coomassie brilliant blue. A \nprotein band corresponding to a SiR-labeled tubulin was excised from the gel, washed, reduced with \ndithiothreitol (DTT), alkylated with iodoacetamide and digested with trypsin (sequencing grade, \nPromega) overnight. The resulting peptides were extracted, dried in a SpeedVac vacuum concentrators \n(Thermo Scientific) and dissolved in 2% acetonitrile/0.05% trifluoroacetic acid (v:v). \nData acquisition \nPeptides were analyzed by electrospray ionization mass spectrometry in a Thermo Orbitrap Exploris \n480 mass spectrometer coupled to an U ltiMate3000 ultrahigh performance liquid chromatography \nsystem (Thermo Scientific). Chromatographic separation was performed with an in-house packed C18 \nreverse-phase column (75 µm ID × 300 mm, Reprosil -Pur 120 C18-AQ, 3 μm, Dr. Maisch) using 0.1% \nformic acid as solvent A and 80% acetonitrile / 0.08% formic acid as solvent B. Separating part of the \nHPLC method included 3 steps of linear gradients: (1) 12-42%B over 40 min, (2) 42-65%B over 27 min \nand (3) 65-95%B over 7.1 min. Mass spectrometer was equipped with a Nanospray Flex Ion source and \ncontrolled by Thermo Scientific Xcalibur 4.4 and Thermo Exploris 480 3.0 software. Data were acquired \nusing an 88-min Top30 data-dependent acquisition method. One full MS scan across the 350–1600 m/z \nrange was acquired at a resolution of 120000, with an AGC target of 300% and a maximum fill time of \n25 ms. Precursors with charge states 2 –6 above a 1e4 intensity threshold were then sequentially \nselected using isolation window of 1.6 m/z, fragmented with nitrogen at a normalized collision energy \nsetting of 28%, and the resulting MS2 spectra recorded at a resolution of 30000, AGC targets of 75% \nand a maximum fill time of 50 ms. Dynamic exclusion of precursors was set to 22 s. \nData processing  \nProteins and sites of SiR-labeling were identified with Proteome Discoverer 3.1.1.93 using SequestHT \nas a search engine. For this, Thermo raw files were searched against a database that included \nsequences of Sus scrofa UniProt proteome (release 24-01-2024) and common contaminants observed \nin MS experiments. Two missed cleavages were allowed. Protein N-terminal acetylation, M-oxidation, \nC-carbamidomethylation as well as C/K+573.2442 were set as variable modifications. \nCell cycle analysis by imaging cytometry \nThe probes were dissolved in DMSO  (Sigma Aldrich, #900645 -4x2mL) at 500 – 2000-fold stock \nconcentration and added to culture media of cells at 500 – 2000-fold dilution accordingly. In parallel, \nthe appropriate DMSO control samples were prepared by adding corresponding amount of DMSO \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 27, 2025. ; https://doi.org/10.1101/2025.01.27.635008doi: bioRxiv preprint \n\nPage 17 of 23 \n \nvolume to the separate well.  HeLa cells were grown in 6 -well plates (~250,000 cells per well) in \npresence of the fluorescent probe in variable concentrations for 24 h at 37°C in humidified incubator \nwith 5% CO 2. Cells were processed according to the NucleoCount er® NC-3000™ two-step cell cycle \nanalysis protocol for cells attached to T -flasks, cell culture plates or micro -carriers. In particular, the \n250 µL lysis solution (Solution 10, Chemometec Cat. No. 910-3010) supplemented with 10 µg/ml DAPI \n(Solution 12, Chemometec Cat. No. 910-3012) was used per well, incubated at 37 °C for 5 min. Then \n250 µL of stabilization solution (Solution 11, Chemometec Cat. No. 910 -3011) was added. Cells were \ncounted on a NucleoCounter® NC -3000™ in NC -Slide A2™ slides (Chemometec, Cat. No. 942 -0001) \nloaded with ∼30 µL of each of the cell suspensions into the chambers of the slide. Each time, ~10,000 \ncells in total were measured, and the obtained cell cycle histograms were analyzed with ChemoMetec \nNucleoView NC-3000 software, version 2.1.25.8. All experiments were repeated three times (with cells \nfrom different passages) and the results are presented as means with standard deviations. \nCytotoxicity experiment after washing  \nThis protocol refers to Figure 4B. U-2 OS cells were seeded in 12-well plates (~300,000 cells per well) \n24h to 48h prior to the experiment. Cells were incubated in a humidified 5% CO 2 incubator at 37 °C. \nTwo different conditions were tested: first cells were incubated for 24h in presence of OptiMEM \ncontaining 1 µM of Probe 2 and 10 µM of verapamil. For the second condition, cells were incubated \n4h in presence of OptiMEM containing 1 µM of Probe 2 and 10 µM of verapamil, washed briefly 4 \ntimes with HBSS and then washed 10 times over 2h with DMEM+, and then further incu bate with \nDMEM+ for 18h. Cytotoxicity experiments were then proceed as described above. This experiment \nwas reproduced 4 times (N=4) on different days with cells from different passages. T he results are \npresented as means with standard deviations. \nWestern-Blot \nConfluent cells in a 6-well plate were incubated in presence of OptiMEM (Thermo Fisher, #11058021) \nsupplemented with probe (3 µM) for 1h at 37°C in humidified incubator with 5% CO 2. The media was \nremoved and cells were washed twice with HBSS. CelLytic™ M (300 µL per well, Sigma-Aldrich #C2978) \nwas added and the plate was put on a shaker for 30 min. Cell lysates were collected in Eppendorf and \nwere centrifugated at 15 000 g and 4°C for 20 min. Supernatants were collected and mixed with 1/3 \nvolume of 4x SDS sample buffer (Tris-HCl pH=6.8: 0.2M, SDS: 8.0% (w/v), Bromophenol blue: 0.6 mM, \nGlycerol: 5.4 M + 50 µL of β-mercaptoethanol per mL of solution prior to use), and boiled for 5 min at \n95°C. The different samples were loaded on 4-15% Mini-PROTEAN® TGX™ Precast Protein Gels (Biorad, \n#4561086). After electrophoresis in Mini-PROTEAN® Tetra Cell using SDS-Page running buffer (0.25 M \nTris HCl, 1.92 M Glycine and 1% (w/v) Sodium Dodecyl Sulfate (SDS) pH=8.3), proteins were transferred \nfrom the gel to a PVDF-membrane (iBlot™ 2 Transfer Stacks, PVDF, regular size, # IB24001) using iBlot \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 27, 2025. ; https://doi.org/10.1101/2025.01.27.635008doi: bioRxiv preprint \n\nPage 18 of 23 \n \n2 Gel Transfer Device (P0 method from the supplier was used: 20V for 1min, 23V for 4 min, 25 V for 2 \nmin). \nMembrane was then blocked with 1% BSA in PBS containing 0.1% of Tween 20  (Blocking buffer) \novernight at 4°C. Primary antibody (Rabbit Recombinant Monoclonal Beta-3-tubulin antibody, Abcam, \n#ab52623) was added (1/2000, v:v). After 1h at room temperature, the membrane was washed three \ntimes 10 min with PBS +0.1% Tween 20. Membra ne was incubated for 1h at room temperature in \npresence of a solution of secondary antibody ( Donkey anti-Rabbit IgG (H+L) Highly Cross -Adsorbed \nSecondary Antibody, Alexa Fluor™ Plus 488, Thermo Fisher, #A32790) in blocking buffer (1/1000, v:v). \nAfter a bri ef washing step with PBS+0.1% Tween 20, fluorescence images were recorded using \nAmersham Imager 600 RGB.  \nSample preparation for live-cell imaging \nCells were seeded on µ -Slide 8 Well Glass Bottom dishes (Ibidi, #80827)  24 to 48h prior to imaging . \nCells were washed four times with HBSS (Gibco, #14025) to remove FBS and incubated with OptiMEM \n(Thermo Fisher, #11058021) containing 1 µM of probe in a humidified 5% CO2 incubator at 37 °C. After \n2h or 4h (depending on the experiment), the media supplemented with probe was removed, cells were \nwashed three times with HBSS and imaged in DMEM+ [ high-glucose DMEM (Thermo Fisher, \n#31053044) with 10% FBS (Thermo Fisher, #10082147) supplemented with 1 mM Sodium pyruvate \n(Sigma, #S8636), 1% GlutaMax (Thermo Fisher, #350500 38) and 1% Penicillin -Streptomycin (Sigma, \n#P0781)]. \nSample preparation for fixed cells imaging \nProtocol adapted from R. Gerasimaite et al., 2021.31 Cells were seeded on µ-Slide 8 Well Glass Bottom \ndishes (Ibidi, #80827) 24 to 48h prior to imaging. Cells were washed 4 times with HBSS (Gibco, #14025) \nto remove FBS and incubated with OptiMEM (Thermo Fisher, #11058021) containing 1 µM of probe in \na humidified 5% CO 2 incubator at 37 °C. After 2h or 4h (depending on the experiment), the media \nsupplemented with prob e was removed, cells were washed 4 times with 200 µL of PEMP (100 mM \nPIPES pH 6.8, 1 mM EGTA, 1 mM MgCl 2 + 4% PEG8000), permeabilized 90s with 0.5% Triton X -100 in \nPEM (PEMP without PEG8000), and washed again 4 times with 200 µL of PEMP. Then, cells were \nincubated with 200 μL of 0.2% glutaraldehyde in PEM for 15 min, followed by 200 μL of 2 mg/mL NaBH4 \nin PEM (dissolved immediately before use) for another 15 min. The samples were washed 4× with 200 \nμL of PEM and imaged in glycerol buffer (GB, 10 mM Na-PO4, pH 6.8, 1mM EGTA, 6 mM MgCl2, 3.4 M \nglycerol). \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 27, 2025. ; https://doi.org/10.1101/2025.01.27.635008doi: bioRxiv preprint \n\nPage 19 of 23 \n \nLive cell imaging with or without extensive washing \nThis protocol refers to Figure 4A. U-2 OS were seeded on µ -Slide 8 Well Glass Bottom dishes (Ibidi, \n#80827) 24 to 48h prior to imaging. Cells were washed 4 times with HBSS (Gibco, #14025) to remove \nFBS and incubated with OptiMEM (Thermo Fisher, #11058021) containing 10 µM of verapamil and 1 \nµM of 6-SiR-o-C9-CTX or 1 µM of SiR-CTX in a humidified 5% CO2 incubator at 37 °C for 4h. Cells were \nwashed briefly 4 times with HBSS, media was replaced by DMEM+ and cells were imaged without \nfurther washing (it corresponds to the conditions “Before extensive washing”). The other condition \n“After extensive washing” refers to cells that were additionally washed 10 times over 2h with DMEM+ \nafter the brief washing with HBSS, and then imaged in DMEM+. This experiment was reproduced three \ntimes on different days with cells from different passages with similar results.  \nConfocal/STED microscopes and imaging parameters \nConfocal and STED images were acquired using Abberior STED  Facility Line scanning (Abberior \nInstruments GmbH) or TCS SP8 (Leica) microscopes. STED images were acquired using Abberior STED \nFacility Line scanning (Abberior Instruments GmbH) microscopes. Imaging parameters are summarized \nin Table S4. \nTCS SP8 confocal microscope is equipped with 405, 458, 476, 488, 496, 514, 561 and 633 nm excitation \nlasers as well as HC PL APO CS2 63x/1.40 Oil objective (Leica). M icroscope has three Hybrid and two \nPMT detectors which can be tuned to any detection window in the range 400 – 800 nm. Probes were \nexcited at 633 nm (Laser power: 1%) and emission was collected between 650 and 710 nm.  \nAbberior STED Facility Line is equipped with 488, 515, 561, 640 and 700 nm 40 MHz pulsed excitation \nlasers, a pulsed 775 nm 40 MHz 3W STED laser, and an UPlanSApo 60x/1.40 Oil objective. Microscope \nhas two APD and two MATRIX detectors which can be tuned to any detection window in the range 400 \n– 800 nm. Pixel size was 30 nm in the xy plane was used for 2D STED images and 80 nm in the xy plane \nfor large field of view images. Laser powers were optimized for each sample. \nAbberior STED Expert Line equipped with 561 nm and 640 nm 40 MHz pulsed excitation lasers, a pulsed \n775 nm 40 MHz 3W STED laser, and an UPlanSApo 100x/1.40 Oil objective. The following detection \nwindows were used: for the SiR channel 685 / 70 nm. Pixel size was 20 nm in the xy plane was used for \n2D STED images. Laser powers were optimized for each sample. \nVisualization and modeling of the tubulin cryo-EM structures and models. \nCryo-EM structures of pig tubulin complex with Taxol (PDB: 5SYF)32 and human (PDB: 6E7C)33 tubulin \nwere downloaded from Protein Data Bank repository and visualized using Swiss-PdbViewer (v 4.1) or \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 27, 2025. ; https://doi.org/10.1101/2025.01.27.635008doi: bioRxiv preprint \n\nPage 20 of 23 \n \nUCSF ChimeraX (v  1.6)34. To identify Taxol binding pocket in human protein, both structures were \nsuperimposed using iterative fit function (backbone atoms only) of the program. \nProcessing and visualization of the acquired images \nAll acquired images were processed and visualized using Fiji 35. Line profiles were measured using the \n“straight line” tool with the line width set to 3 pixels. To define  apparent microtubule FWHM, line \nprofiles were fitted with Gaussian and Lorentzian distributions for confocal and STED images \nrespectively.  \nStatistical tests \nComparisons were performed using u npaired t-test (two-tailed) in GraphPad Prism 10.4 (GraphPad \nSoftware, Inc., San Diego, CA, USA)  and p-values ≤ 0.05 were considered statistically significant.  All \nmicroscopy imaging experiments were repeated at least two times on different non-consecutive days \n(n ≥ 2). Multiple fields of view (n ≥ 3) were acquired during each imaging session and representative \nimages are shown in the figures. \nAcknowledgements \nThe authors thank the Max Planck Society for supporting this work. The authors are grateful to Dr. \nVladimir Belov, Jan Seikowski, Jens Schimpfhauser and Jürgen Bienert for the NMR measurements of \nnumerous probes and the central analytics’ team (Institute f or Organic and Biomolecular Chemistry, \nGeorg-August University, Göttingen) for acquiring HRMS. Figures 2D and S11A were performed with \nUCSF ChimeraX, developed by the Resource for Biocomputing, Visualization, and Informatics at the \nUniversity of California, San Francisco, with support from National Institutes of Health R01-GM129325 \nand the Office of Cyber Infrastructure and Computational Biology, National Institute of Allergy and \nInfectious Diseases. \nAuthor contributions \nM.A. and G.L. conceived and planned the study. M.A, T.K., O.D. and G.L. performed the experiments. \nM.A, T.K., O.D., H.U. and G.L. performed the data analysis. M.A. and G.L. wrote the initial draft; all \nauthors contributed to the final version of the manuscript. \nFunding Sources \nThe Max Planck Society. \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 27, 2025. ; https://doi.org/10.1101/2025.01.27.635008doi: bioRxiv preprint \n\nPage 21 of 23 \n \nCompeting Interests \nG.L. is a co -inventor o n the patent (EP2748173B1 and US9346957B2, applicant EPFL) \ndescribing SiR and its derivatives. \n  \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 27, 2025. ; https://doi.org/10.1101/2025.01.27.635008doi: bioRxiv preprint \n\nPage 22 of 23 \n \nReferences  \n \n1 Sahl, S. J., Hell, S. W. & Jakobs, S. Fluorescence nanoscopy in cell biology. Nat Rev Mol Cell \nBiol 18, 685-701, doi:10.1038/nrm.2017.71 (2017). \n2 Lukinavičius, G. et al. Stimulated emission depletion microscopy. Nature Reviews Methods \nPrimers 4, 56, doi:10.1038/s43586-024-00335-1 (2024). \n3 D'Este, E., Lukinavičius, G., Lincoln, R., Opazo, F. & Fornasiero, E. F. Advancing cell biology \nwith nanoscale fluorescence imaging: essential practical considerations. Trends Cell Biol 34, \n671-684, doi:10.1016/j.tcb.2023.12.001 (2024). \n4 Wilhelm, J. et al. Kinetic and Structural Characterization of the Self-Labeling Protein Tags \nHaloTag7, SNAP-tag, and CLIP-tag. Biochemistry 60, 2560-2575, \ndoi:10.1021/acs.biochem.1c00258 (2021). \n5 Hink, M. A. et al. Structural Dynamics of Green Fluorescent Protein Alone and Fused with a \nSingle Chain Fv Protein*. Journal of Biological Chemistry 275, 17556-17560, \ndoi:https://doi.org/10.1074/jbc.M001348200 (2000). \n6 Fatti, E., Khawaja, S. & Weis, K. The dark side of fluorescent protein tagging – the impact of \nprotein tags on biomolecular condensation. bioRxiv, 2024.2011.2023.624970, \ndoi:10.1101/2024.11.23.624970 (2024). \n7 Tsukiji, S., Miyagawa, M., Takaoka, Y., Tamura, T. & Hamachi, I. Ligand-directed tosyl \nchemistry for protein labeling in vivo. Nat Chem Biol 5, 341-343, doi:10.1038/nchembio.157 \n(2009). \n8 Fujishima, S. H., Yasui, R., Miki, T., Ojida, A. & Hamachi, I. Ligand-directed acyl imidazole \nchemistry for labeling of membrane-bound proteins on live cells. J Am Chem Soc 134, 3961-\n3964, doi:10.1021/ja2108855 (2012). \n9 Takaoka, Y., Nishikawa, Y., Hashimoto, Y., Sasaki, K. & Hamachi, I. Ligand-directed \ndibromophenyl benzoate chemistry for rapid and selective acylation of intracellular natural \nproteins. Chem Sci 6, 3217-3224, doi:10.1039/c5sc00190k (2015). \n10 Matsuo, K., Nishikawa, Y., Masuda, M. & Hamachi, I. Live-Cell Protein Sulfonylation Based on \nProximity-driven N-Sulfonyl Pyridone Chemistry. Angew Chem Int Ed Engl 57, 659-662, \ndoi:10.1002/anie.201707972 (2018). \n11 Tamura, T. et al. Rapid labelling and covalent inhibition of intracellular native proteins using \nligand-directed N-acyl-N-alkyl sulfonamide. Nat Commun 9, 1870, doi:10.1038/s41467-018-\n04343-0 (2018). \n12 Xin, X. et al. Ultrafast and selective labeling of endogenous proteins using affinity-based \nbenzotriazole chemistry. Chem Sci 13, 7240-7246, doi:10.1039/d1sc05974b (2022). \n13 Stoddart, L. A. et al. Ligand-directed covalent labelling of a GPCR with a fluorescent tag in live \ncells. Commun Biol 3, 722, doi:10.1038/s42003-020-01451-w (2020). \n14 Comeo, E. et al. Ligand-Directed Labeling of the Adenosine A1 Receptor in Living Cells. \nJournal of Medicinal Chemistry 67, 12099-12117, doi:10.1021/acs.jmedchem.4c00835 (2024). \n15 Yamaura, K., Kiyonaka, S., Numata, T., Inoue, R. & Hamachi, I. Discovery of allosteric \nmodulators for GABAA receptors by ligand-directed chemistry. Nature Chemical Biology 12, \n822-830, doi:10.1038/nchembio.2150 (2016). \n16 Wakayama, S. et al. Chemical labelling for visualizing native AMPA receptors in live neurons. \nNature Communications 8, 14850, doi:10.1038/ncomms14850 (2017). \n17 Verdier-Pinard, P. et al. Tubulin proteomics: towards breaking the code. Anal Biochem 384, \n197-206, doi:10.1016/j.ab.2008.09.020 (2009). \n18 Fourel, G. & Boscheron, C. Tubulin mutations in neurodevelopmental disorders as a tool to \ndecipher microtubule function. FEBS Letters 594, 3409-3438, \ndoi:https://doi.org/10.1002/1873-3468.13958 (2020). \n19 Song, Y. & Brady, S. T. Post-translational modifications of tubulin: pathways to functional \ndiversity of microtubules. Trends Cell Biol 25, 125-136, doi:10.1016/j.tcb.2014.10.004 (2015). \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 27, 2025. ; https://doi.org/10.1101/2025.01.27.635008doi: bioRxiv preprint \n\nPage 23 of 23 \n \n20 Wang, R. et al. Low-Toxicity Sulfonium-Based Probes for Cysteine-Specific Profiling in Live \nCells. Anal Chem 94, 4366-4372, doi:10.1021/acs.analchem.1c05129 (2022). \n21 Wang, Y. et al. A Peptide-Based Ligand-Directed Chemistry Enables Protein Functionalization. \nOrg Lett 24, 7205-7209, doi:10.1021/acs.orglett.2c02974 (2022). \n22 Lukinavičius, G. et al. A near-infrared fluorophore for live-cell super-resolution microscopy of \ncellular proteins. Nature Chemistry 5, 132-139, doi:10.1038/nchem.1546 (2013). \n23 Bucevicius, J., Kostiuk, G., Gerasimaite, R., Gilat, T. & Lukinavicius, G. Enhancing the \nbiocompatibility of rhodamine fluorescent probes by a neighbouring group effect. Chem Sci \n11, 7313-7323, doi:10.1039/d0sc02154g (2020). \n24 Lukinavičius, G. et al. Fluorescent dyes and probes for super-resolution microscopy of \nmicrotubules and tracheoles in living cells and tissues. Chem Sci 9, 3324-3334, \ndoi:10.1039/c7sc05334g (2018). \n25 Deng, F. et al. Multiple Factors Regulate the Spirocyclization Equilibrium of Si-Rhodamines. \nThe Journal of Physical Chemistry B 124, 7467-7474, doi:10.1021/acs.jpcb.0c05642 (2020). \n26 Grimm, J. B. et al. Optimized Red-Absorbing Dyes for Imaging and Sensing. J Am Chem Soc \n145, 23000-23013, doi:10.1021/jacs.3c05273 (2023). \n27 Tamura, T. & Hamachi, I. Chemistry for Covalent Modification of Endogenous/Native \nProteins: From Test Tubes to Complex Biological Systems. Journal of the American Chemical \nSociety 141, 2782-2799, doi:10.1021/jacs.8b11747 (2019). \n28 Nsamba, E. T. & Gupta, M. L., Jr. Tubulin isotypes – functional insights from model organisms. \nJournal of Cell Science 135, jcs259539, doi:10.1242/jcs.259539 (2022). \n29 Abal, M., Andreu, J. M. & Barasoain, I. Taxanes: Microtubule and Centrosome Targets, and \nCell Cycle Dependent Mechanisms of Action. Current Cancer Drug Targets 3, 193-203, \ndoi:http://dx.doi.org/10.2174/1568009033481967 (2003). \n30 Gerasimaitė, R. et al. Efflux pump insensitive rhodamine–jasplakinolide conjugates for G- and \nF-actin imaging in living cells. Organic & Biomolecular Chemistry 18, 2929-2937, \ndoi:10.1039/D0OB00369G (2020). \n31 Gerasimaitė, R. et al. Blinking Fluorescent Probes for Tubulin Nanoscopy in Living and Fixed \nCells. ACS Chem Biol 16, 2130-2136, doi:10.1021/acschembio.1c00538 (2021). \n32 Kellogg, E. H. et al. Insights into the Distinct Mechanisms of Action of Taxane and Non-Taxane \nMicrotubule Stabilizers from Cryo-EM Structures. J Mol Biol 429, 633-646, \ndoi:10.1016/j.jmb.2017.01.001 (2017). \n33 Ti, S. C., Alushin, G. M. & Kapoor, T. M. Human beta-Tubulin Isotypes Can Regulate \nMicrotubule Protofilament Number and Stability. Dev Cell 47, 175-190 e175, \ndoi:10.1016/j.devcel.2018.08.014 (2018). \n34 Goddard, T. D. et al. UCSF ChimeraX: Meeting modern challenges in visualization and \nanalysis. Protein Sci 27, 14-25, doi:10.1002/pro.3235 (2018). \n35 Schindelin, J. et al. Fiji: an open-source platform for biological-image analysis. Nature \nMethods 9, 676-682, doi:10.1038/nmeth.2019 (2012). \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 27, 2025. ; https://doi.org/10.1101/2025.01.27.635008doi: bioRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}