Tim Grotjohann; Ilaria Testa; Matthias Reuss; Tanja Brakemann; Christian Eggeling; Stefan W Hell; Stefan Jakobs

“ Since the fluorescent proteins are confined to individual peroxisomes and do not diffuse over large distances, labeled peroxisomes are well suited for a direct comparison of rsEGFP2 and rsEGFP bleaching. The comparison showed that the overall photobleaching of rsEGFP2 was substantially lower than that of rsEGFP when the above fast RESOLFT imaging conditions were applied (Figure 3—figure supplement 3) .
Phototoxicity is a major concern in (superresolution) fluorescence microscopy of living cells. ”

“ However, since the wavelengths (560–800 nm) used for STED are in the comparatively benign long-wavelength part of the spectrum ( [2] ) , STED microscopy can image living cells and tissues, including neurons in the cerebral cortex of a living mouse ( [22] ) . In the RSFP-based RESOLFT microscopy demonstrated here, the employed light intensities (1–80 kW/cm2) are several orders of magnitude lower than in STED-microscopy and comparable to those used in live-cell confocal fluorescence microscopy. ”

“ RESOLFT imaging was performed with a home-built setup maintaining the cells at 35°C. Because the filaments formed by Vimentin-rsEGFP2 are relatively immobile, we selected an irradiation scheme encompassing low switching light intensities in combination with extended irradiation times. The living cells were imaged pixel-by-pixel, by first irradiating with 405 nm light (2 kW/cm2) for 40 μs to switch most proteins into their fluorescent on-state. ”

“ Now, Grotjohann et al. have modified this protein to make a new fluorophore called rsEGFP2 with a shorter switching time, and have used it to image various structures—including Vimentin, a protein that forms part of the cytoskeleton in many cells, and organelles called peroxisomes—inside live mammalian cells. They were able to record these images some 25–250 times faster than would have been possible with previous RESOLFT approaches. ”

“ Unfortunately, in these demonstrations, the still relatively slow switching kinetics of rsEGFP entailed pixel dwell times of 10–20 ms at switching light intensities of ~1 kW/cm2. Hence images of 10 μm × 10 μm in size and a pixel step size of 20 nm required recording times of about an hour. In this study we describe the generation, characterization and application of a novel RSFP, namely rsEGFP2, which facilitates faster switching and, together with modifications of the initial switching scheme, enables 25–250 faster RESOLFT recordings in living cells. ”

“ Parallelization of the scanning procedure with an array of doughnuts or lines (so-called ‘structured illumination’) ( [31] ; [7] ; [10] ) overcomes this dependence on the field of view and cuts down the recording time by the degree of parallelization. Owing to the low-light level operation, the degree of parallelization can easily amount up to a factor of 100–1000. For this reason, given the short pixel dwell times attained herein, parallelized RESOLFT versions should enable video-rate nanoscopy across the whole field of view of the objective lens. ”

“ PALM live-cell imaging reportedly requires light doses of 1,000–100,000 J/cm2 for on-switching with UV (405 nm) light and 25,000–300,000 J/cm2 for fluorescence excitation ( [23] ; [24] ) . Live-cell GSDIM experiments using a yellow fluorescent protein required even larger irradiation doses (100,000–900,000 J/cm2) ( [27] ; [28] ) . Although RESOLFT microscopy may induce phototoxicity after extended exposure, similarly to live-cell confocal microscopy, it currently is the superresolution method entailing the lowest light dose. ”

“ The slight difference between brightness values determined at low and high light intensities may be attributed to the increased population of unidentified dark states at higher intensities. rsEGFP2 exhibits a ‘negative’ switching mode, that is irradiation with light of around 480 nm induces fluorescence and, in a competing process, switches rsEGFP2 off. ”

“ The single molecule brightness, that is the number of emitted photons per time unit, of rsEGFP and rsEGFP2 are similar. However, because a single switching cycle of rsEGFP2 is on average shorter than a single switching cycle of rsEGFP, rsEGFP2 emits fewer photons in a single cycle than rsEGFP. The pKa of the chromophore in the thermal equilibrium state is ~5.8 (Figure 1D) . ”

“ The fluorescence quantum yields and the molar extinction coefficients at the respective absorption maximum were determined relative to the reported values of EGFP (quantum yield ΦFL = 0.60, molar extinction coefficient at 489 nm ε = 53,000 M−1 cm−1) ( [11] ) . Irradiation-dependent changes in the absorption were quantified by illuminating the protein solution in a cuvette with a fiber coupled mercury lamp (Lecia Microsystems, Wetzlar, Germany) equipped with a (488 ± 5) nm excitation filter. For each measurement of the spectrum the irradiation was briefly interrupted. ”

“ Fluorescence was excited by simultaneous irradiation with light of 488 nm and 405 nm. (A–E) : PtK2 cells; (F) : Vero cell. Scale bars: 10 μm.DOI: [10.7554/eLife.00248.008 http://dx.doi.org/10.7554/eLife.00248.008] The possibility to switch rsEGFP2 faster than rsEGFP, its higher switching stamina, and its faster maturation kinetics, suggested that this protein would be superior for fast live-cell RESOLFT imaging. We first expressed rsEGFP2 fused to the C-terminus of Vimentin in mammalian PtK2 (Potorous tridactylis) cells. ”

“ Switching was performed on living PtK2 cells expressing Vimentin-rsEGFP or Vimentin-rsEGFP2 by alternate irradiation with 405 nm (2 kW/cm2) and 491 nm (5.7 kW/cm2) light. Illumination times were chosen so that the fluorescence was fully switched to the minimum or maximum, respectively, in each cycle. Each plotted data point is the average (over 100 cycles) of the maximum fluorescence intensity in each cycle. The data points were fitted by a mono-exponential function, and the resulting curve was baseline corrected and normalized to 1. ”

“ The latter also benefits through IP secured by the Max Planck Society.The remaining have declared that no competing interests exist.Author contributionsTG, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article.IT, Acquisition of data, Analysis and interpretation of data.MR and TB, Acquisition of data.CE, Conception and design.SWH, Conception and design, Drafting or revising the article.SJ, Conception and design, Analysis and interpretation of data, Drafting or revising the article. ”