Chemically induced photoswitching of fluorescent probes–a general concept for super-resolution microscopy.

3. Experimental 3.1.

Sample preparation

HeLa or U2OS cells were plated and grown in Labtek 8-well chambered cover glass (Nunc, Germany) up to 80% confluency. The cells were transiently transfected with a plasmid encoding for one of the fluorescent proteins (mEos2, PAmCherry1, bsDronpa, psCFP2, Dendra2) and actin using the transfection reagent Turbofect (Fermentas, Germany). After overnight expression, the cells were fixed using 4% PFA for 10 minutes and stored in phosphate buffered saline (PBS). Labeling with organic fluorophores was done using standard immunofluorescence protocols. The microtubulin network of COS-7 cells was labeled using a rabbit anti-β-tubulin primary antibody (Abcam, Cambridge, UK). The secondary antibody was a goat F(ab’) 2 anti rabbit (Invitrogen, USA) that was custom-labeled with ATTO 655 following protocols published by the manufacturer (Atto-Tec, Germany). The degree of labeling was determined to 1.3. Microtubulin of HeLa cells was labeled with monoclonal anti-α-tubulin antibody (MS-581-P1; Thermo Scientific, Waltham, MA) and Alexa Fluor 532 labeled secondary antibody (A11002, Invitrogen, Carlsbad, CA). Staining of the endoplasmatic reticulum of HeLa cells was done using monoclonal antibody against calnexin (ab31290; Abcam, Cambridge, UK) and Alexa Fluor 647 labeled secondary antibody (A21235; Invitrogen, USA). 3.2. Microscopic configuration A standard epifluorescence microscope operated in total internal reflection fluorescence (TIRF) and equipped with suitable laser sources was used for PALM and d STORM imaging, essentially configured as published elsewhere [ 6 , 7 , 8 ]. Briefly, a multi-line argon-krypton laser (Innova 70C; Coherent, USA) was coupled into an inverted microscope (IX71, Olympus, Japan) equipped with a 60x oil immersion objective (PlanApo 60x, NA 1.45, Olympus). Excitation and fluorescent light was separated using a dichroic mirror. Here either FF560/659-Di01 (AHF, Germany) or FF410/504/588/669-Di01 (AHF, Germany) or HC576/661 (AHF, Germany) were used. The fluorescence signal was read out with an electron-multiplying CCD camera (EMCCD; Andor Ixon DU 897, Belfast, Ireland). Raw data was processed using the rapidSTORM algorithm [ 30 ]. 3.3.

PALM imaging

Fluorescent proteins were imaged with a suitable combination wavelengths provided by various laser sources: 378 nm or 405 nm and 488 nm (psCFP2, bsDronpa), 378 nm or 405 nm and 568 nm (mEos2, PAmCherry1, Dendra2). Typically, irradiation intensities of 20 to 40 mW (1–2 kW cm -2 ) for the read-out wavelength (488 nm or 568 nm) were used. A lower intensity of about 10 mW was used for imaging bsDronpa. Filter sets for imaging mEos2, PAmCherry1 and Dendra2 included a 568/10 bandpass filter (AHF, Germany) as clean-up filter in the excitation path, and a combination of a 568LP and a 610/75 (AHF) in the detection path. Filter sets for imaging psCFP2 and bsDronpa included a 488/10 clean-up filter (AHF) in the excitation path, and a combination of a 488LP and a 550/88 (AHF) in the detection path. Typically, 3000–6000 frames were recorded at frame rates of 100 Hz, and the experiment was terminated when all fluorescent proteins were irreversibly photobleached. In experiments that showed reversible photoswitching of the fluorescent proteins, a similar number of frames was recorded and the experiment finished prior to all fluorescent proteins photobleached. 3.4. dSTORM imaging d STORM imaging was performed in suitable imaging buffers as described elsewhere for carbocyanine [ 7 ] and rhodamine or oxazine fluorophores [ 8 ]. Briefly, photoswitching of the carbocyanine Alexa 647 was performed with 514 nm and 647 nm radiation in oxygen-depleted PBS buffer and 100 mM MEA added. Photoswitching of ATTO655 was performed in 10 mM MEA and with irradiation at 647 nm. Photoswitching of Alexa Fluor 532 was performed in 100 mM MEA and with irradiation at 514 nm. 3.5.

Dual color imaging

Dual color imaging of ATTO655 and mEos2 was performed by sequentially imaging first ATTO655 and then mEos2. As imaging buffer, 250 mM NaHCO 3 (pH8) with 10 mM MEA was used. The filter sets used were a 488/568/647-10 bandpass filter (AHF, Germany) for the excitation path. In the detection paths, a combination of a LP647 and 705/75(AHF) (ATTO655 channel) and a combination of LP568 SP630 and 610/75 (AHF) (mEos2 channel) was used. The two channels were aligned by imaging multi-labeled beads (TetraSpeck, Invritrogen) and calculating the local weighted mean using the local weighted mean function of Matlab (Mathworks, Natick, MA) [ 31 ]. 3.6.

Data analysis

Reversible switching was assessed through calculating the distribution of all localization distances within the population of distinct regions of interest (ROI) with the size of 1.65 × 1.65 µm². In order to have comparable input data, we chose intracellular regions with a high occurrence of globular actin as ROI. Regions with filaments were avoided, because of a variable and dense structure that is typical for actin multimers. The spatial distribution of all localizations was calculated by Ripley’s K-function [ 28 , 29 ]: (1) where n is the total number of localizations within the ROI, dij is the distance between the localization i and j, Nr is the number of localization around localization i within the distance j and λ is a weighting factor correcting for the area of the ROI. The function is linearized: (2) and then normalized: (3) so that a normal distribution would give an expected value of H(r) = 0 for all r [ 32 ]. In order to account for edge effects we chose the torroidal edge correction [ 33 ]. To achieve statistical significance we employed 95% confidence envelopes produced with Monte Carlo methods [ 32 ].