Protein labeling for live cell fluorescence microscopy with a highly photostable renewable signal† †Electronic supplementary information (ESI) available: Supplementary methods, figures, movies, and data. See DOI: 10.1039/c7sc01628j

A novel method of protein labeling uses the highly dynamic reversible association of a cell-permeable fluorogenic dye and lipocalin Blc mutants.


Methods
Synthesis of the chromophore library S11 Molecular docking S15 Molecular cloning Protein expression and purification Cell culture and transient transfection S16 Fluorescence microscopy Chromophore titration S17 Analysis of fluorescence titration data S18 Quantum yield measurements S19    "Talon" in the protein list stands for empty Talon beads. "K" in the list of chromophore stands for PBS buffer without chromophore (exposure is 100x times longer in this case). Only chromophores that showed detectable fluorescence signal are shown. Each vertical bar consists of two columns, corresponding to GFP (marked with green) and TxRed (marked with red) filter sets. The last column is the photograph of protein solutions (at 2 mg/ml) co-purified with unknown bacterial colored compounds.       NMR spectra were recorded on a 700 MHz Bruker Avance III NMR at 293 K. Chemical shifts are reported relative to residue peaks of CDCl 3 (7.27 ppm for 1 H and 77.0 ppm for 13 C) or DMSO-d 6 (2.51 ppm for 1 H and 39.5 ppm for 13 C). Melting points were measured on a SMP 30 apparatus. High-resolution mass spectra (HRMS) spectra were recorded on an Agilent 6224 TOF LC/MS System (Agilent Technologies, Santa Clara, CA, USA) equipped with a dual-nebulizer ESI source and on a Bruker micrOTOF II instrument.

Synthesis of the compounds GA and GC
Compounds GA and GC were synthesized using typical procedures by general methods I-III (Scheme). The synthetic procedures for all novel compounds are presented directly. Scheme S1 . General methods used in synthesis General method I Corresponding aldehyde (50 mmol), amidoacetic acid (60 mmol) and sodium acetate (100 mmol) were dissolved in corresponding anhydride (50 mL). The mixture was stirred at 110 o C for 5 hours. The mixture was dissolved by EtOAc (400 mL), washed with water (2x150 mL) and brine (2x150 mL) and dried over Na 2 SO 4 . The solvent was evaporated and the product was purified by flash chromatography (Hexane-EtOAc). The residue was dissolved in ethanol (100 mL) and the methylamine solution (40% aq) was added. The mixture was stirred for 4 hours and the solvent was evaporated. The dimethylformamide (50 mL) and Cs 2 CO 3 7.0 g (21 mmol) were added and the mixture was refluxed for 10 minutes. The solvent was evaporated and the mixture was dissolved by EtOAc (300 mL), washed with water (2x100 mL) and brine (2x100 mL) and dried over Na 2 SO 4 . The solvent was evaporated and the product was purified by column chromatography (CHCl 3 -EtOH).

General method II
Corresponding aldehyde (20 mmol) was mixed with 5 mL of 40% aqueous methylamine solution, anhydrous sodium sulfate (20 g) and chloroform (100 mL). The mixture was stirred for 48 hours at room temperature, filtered and dried over the additional Na 2 SO 4 . The solvent was evaporated and ethyl ((1-methoxy)amino)acetate (3.5 g, 22 mmol) and ethanol (10 mL) were added. The mixture was stirred for 24 hours at room temperature, solvents were removed in vacuum and the product was purified by column chromatography (CHCl 3 -EtOH).
4-(diethylamino)benzaldehyde 8.85 g (50 mmol), 2-propionamidoacetic acid 7.86 g (60 mmol) and sodium acetate 6.6 g (100 mmol) were dissolved in propionic anhydride (50 mL)/ The mixture was stirred at 110 o C for 5 hours. The mixture was dissolved by EtOAc (400 mL), washed with water (2x150 mL) and brine (2x150 mL) and dried over Na 2 SO 4 . The solvent was evaporated and the product was purified by flash chromatography (Hexane-EtOAc 4:1). The residue was dissolved in ethanol (100 mL) and the methylamine solution (40% aq) was added. The mixture was stirred for 4 hours and the solvent was evaporated. The dimethylformamide (50 mL) and Cs 2 CO 3 7.0 g (21 mmol) were added and the mixture was refluxed for 10 minutes. The solvent was evaporated and the mixture was dissolved by EtOAc (300 mL), washed with water (2x100 mL) and brine (2x100 mL) and dried over Na 2 SO 4 . The solvent was evaporated and the product was purified by column chromatography (CHCl 3 -EtOH 10:1).
Orange solid ( Synthesis of the conformationally-locked compounds ABDI-BF 2 ABDI-BF 2 compounds were synthesized as reported previously 5 by the direct borylation of the corresponding imidazolones GC (Scheme S2). The synthetic procedures for all novel compounds are presented directly. Scheme S2 . General procedure of the borylation.

Molecular docking
For preliminary molecular docking we followed previously published protocol 6 . Briefly, a library of lipocalin structures with two or one substitutions in amino acids facing potential binding site was generated with MODELLER software 7 guided by published lipocalin structure (PDB ID: 1QWD). Files for the ligand (GFP chromophore) and receptor (lipocalin mutants) were prepared for docking in AutoDockTools 8 version 1.5.6. Bounding box for docking was calculated in PyMol. Docking with rigid protein geometry was performed with AutoDock Vina 9 .

Molecular cloning
The full copy of the crystallized fragment of the Blc protein 10  Coding sequence of Blc-pBad was checked by sequencing and this vector was used for further work.
All 19 selected mutants differ from Blc in one or two amino acids. They were constructed either by one or by two rounds of self-assembling cloning 11 .

Protein expression and purification
Blc and Blc-mutant proteins were expressed in XJb(DE3) Autolysis (Zymo Research) E. coli strain. One bacterial colony was inoculated into 200 ml of LB broth and grown overnight (~ S16 another 24 hours at 37°C cells were harvested and resuspended in 5 ml of PBS buffer (pH 7.4). Suspension was frozen at -70°C and thawed at 37°C three times. DNA was destroyed by short sonication, the lysate was centrifuged to obtain cell-free extract. The protein was purified using TALON metal affinity resin (Clontech). A fraction of beads with immobilized protein were aliquoted into a different tube while the rest was washed with PBS / 0.1M EDTA buffer (pH 7.4) to elute the protein into the solution. Finally, all purified samples were dialysed against PBS (pH 7.4).

Fluorescence microscopy
Primary protein-fluorophore interaction screening was performed with TALON beads. Dialysed Blc-mutant protein containing beads (or free TALON beads as a control) were placed in wells of a 96-well plate with 100 μl of PBS buffer. Then the solutions of fluorophores were added to obtain 10 μM concentration. After 10-min incubation at RT the beads were imaged by using BZ-9000 fluorescence microscope (Keyence) with a 2x objective, OP-79301 SB GFP-BP and OP-79302 SB TexasRed filters, and equal exposure time.
Single-molecule localization super-resolution imaging of living cells was carried out on a Nikon Eclipse Ti N-STORM microscope (Nikon, Japan) controlled by NIS-Elements S17 Software and/or the Micro-Manager software 12 . The excitation light from 488 nm (4.5 W/cm2) or 561 nm (120 W/cm2) laser lines was focused on the sample with a 100X oil-immersion objective (Apo TIRF/1.49, Nikon) and PFS (perfect focus system). Optical path included C-NSTORM QUAD filter cube (Nikon) and a 1.5x magnifier lens. The images (a pixel size of 107 nm) were captured with the EM-CCD camera (iXon3 DU-897, Andor, UK) at 10 MHz readout rate, 14-bit, with an EM gain and a pre-amplifier gain set to 200 and 5.1 (12.15 electrons per A/D count) respectively, with an exposure time of 16-50 ms. The TIRF illuminator mirror was adjusted to allow for total internal reflection (objective-based TIRF). A typical series of 5000 frames was taken using ND Acquisition mode. The running median filtering was applied to the time series prior to localization step 13 . Sparse images of individual fluorophores were fitted with the ThunderStorm 2 plugin for FIJI 14 . Localization uncertainties reported here are calculated by ThunderStorm 2 following EM-CCD optimized formula 1 .
Single-molecule localization with high-power illumination was carried out on Leica SR GSDIM system, equipped with the EM-CCD camera (Andor iXon 3 897), SuMo stabilized stage, HCX PL APO 100x/1.47 OIL objective with additional 1.6x magnification (96 nm effective pixel size). The sample was illuminated with 488 laser line operating in "epi" mode (straight out of the objective") at the power density of 5 kW/cm 2 . The camera exposure time was set to 10 ms. Camera settings: 10 MHz readout rate, 14-bit, pre-amplifier gain 5 (12.15 electrons per A/D count), EM gain 296.
STED microscopy was performed on Leica TCS SP8 STED 3X in a Gated-STED mode with Hybrid detectors and HC PL APO CS2 93x/1.30 glycerine-immersion objective. The 514 nm line of the tunable White Light Laser was used for fluorophore excitation at 40% output power. Fluorescence depletion by stimulated emission was achieved with the 40% output power of 1.44W 592 nm continuous-wave laser.
Cell perfusion experiments (Figure 2a) were performed on the Leica AF6000B imaging system equipped with ZYLA-5.5-CL10 camera (Andor, UK), controlled by the Micro-Manager software 12

Chromophore titration
To investigate the binding of fluorophores to Blc-mutant proteins, fluorescence titration was performed. The data were obtained using newly purificated and dialysed protein, and newly diluted fluorophore solutions.
Samples of dry fluorophore were diluted in DMSO. These solutions were used for subsequent preparation of the 1000X stock solutions in 96% EtOH. The concentration of the fluorophores in the stock solutions was measured using Cary 100 UV/VIS spectrophotometer, and the 100X working solutions in PBS buffer were made before each titration. The concentration of the proteins was evaluated using sample absorption at 280 nm and extinction coefficients computed by ProtParam tool (http://web.expasy.org/protparam/). To reduce dilution-induced artifacts during titration, series of diluted protein solutions were also prepared.
The fluorophore solution (10 μl) and required amount of protein solution and PBS up to 1 ml were added into cuvette, gently stirred and immediately measured using Varian Cary Eclipse fluorescence spectrophotometer. For each fluorophore-protein pair points with at least two different fluorophore and fifteen protein concentrations were measured.

Analysis of fluorescence titration data
The fluorescence titration data were fitted to the 1:1 binding model. Supposing Blc-mutant proteins as monomers with single binding site for fluorophore, the binding of fluorophore to Blc-mutant protein can be described as P + F ⇌ PF, where P is free protein, F is free fluorophore, and PF is protein-fluorophore complex. Then the dissociation constant (K d ) equation is: , where [P] eq is free protein, [F] eq is free fluorophore, and [PF] eq is protein-fluorophore concentrations at equilibrium. As we knew only initial concentration of the protein and the fluorophore, and had not any tool for the free protein and the free fluorophore concentration at equilibrium measurement, the K d equation was rewritten using mass balance equations: [ Thus the fluorescence emission intensity (F) of the protein-fluorophore solution can be expressed as: [PF] and Φ [F] are the relative fluorescence quantum yields of the protein-fluorophore complex and the pure fluorophore, respectively. Therefore to determine apparent K d , the experimentally obtained data were fitted to equation 2 by leastsq method from Scipy optimize package.

Quantum yield measurement
Quantum yield of the protein-fluorophore complexes was determined by direct comparison with purified EGFP protein (quantum yield 0.6). S19