As a scanning microscope, STimulated Emission Depletion (STED) nanoscopy needs parallelization for fast wide-field imaging. Using well-designed optical lattices for depletion together with wide-field excitation and a fast camera for detection, we achieve large parallelization of STED nanoscopy. Wide field of view super-resolved images are acquired by scanning over a single unit cell of the optical lattice, which can be as small as 290 nm * 290 nm. Optical Lattice STED (OL-STED) imaging is demonstrated with a resolution down to 70 nm at 12.5 frames per second.
© 2014 Optical Society of America
Recent developments in super-resolution microscopy techniques [1–5] achieved nanometer scale resolution and showed great potential in live cell imaging. Techniques based on single molecule localization [3,4,6–10] and structured illumination (SIM) [5,11,12] are intrinsically parallelized because they use wide-field illumination and cameras for detection. The first techniques require a low density of simultaneously emitting single molecules and therefore need a large number of frames for super-resolved image reconstruction, which limits their imaging speed. The second rely on sophisticated data post-processing, and consequently require high signal to noise camera frames accumulated over long integration times. Because it requires fewer raw data images, linear SIM has showed its potential for video rate imaging but it can only improve the resolution by a factor two of the diffraction limit . Nonlinear SIM can achieve better resolutions but at the expense of imaging speed since it requires a larger number of raw images per time frame .
STED [2,15–18] nanoscopy is a scanning technique requiring only a limited number of fluorescence photons, which can be acquired at short pixel exposure times, to locate the position of the emission. Since it is a coordinate-targeted method, it does not need sophisticated data processing. However, STED and more generally RESOLFT [19–21] (REversible Saturable OpticaL Fluorescence Transitions) need parallelization in order to fully benefit from this temporal resolution for fast wide-field imaging.
A straightforward approach to STED parallelization is based on focused beam multiplication. A configuration using four pairs of scanning excitation and doughnut-STED beams, together with four avalanche photodiodes has been reported . However, scaling the parallelization further up with this approach is mainly limited by experimental complexity and by the power of the available laser sources. Another approach for parallelization is based on structured illumination pattern. RESOLF parallelization has been proposed using 1D interference pattern, but the resolution improvement is only obtained along one direction . Recently methods for massive parallelization of RESOLFT with photo-switchable proteins  and STED nanoscopy  based on the use of 2D structured illumination have been reported. Larger field of view could be achieved for parallelized RESOLFT using photo-switchable fluorescent proteins, because it requires less intensity to switch. However, protein switching is a relatively slow on-off process (~10 ms), which sets a limit to the imaging acquisition rate. Moreover RESOLFT with photo-switchable fluorescent proteins is constrained in its versatility by the need for genetic modification and transfection.
In this article, we show how well-designed optical lattices [25–28] created by multi-beam interference can provide efficient depletion patterns with moderate laser power and can be used for large parallelization of STED, so far the most important and widely used RESOLFT technique. The stimulated emission depletion being an ultrafast on-off switching process (~1 ns), its imaging speed is therefore only limited by the number of detected photons and fast large field of view super-resolution imaging can be achieved.
2. Experimental setup and the generation of optical lattices
Our OL-STED microscopy setup, sketched in Fig. 1, is based on two synchronized laser sources delivering excitation and depletion pulses, akin to a standard STED microscope . The fluorescence excitation beam (wavelength 571 nm and pulse duration 2 ps) is delivered by a frequency doubled optical parametric oscillator (Mira-OPO, Coherent) pumped by picosecond Ti-sapphire laser (Mira 900, Coherent) at a repetition rate of 76 MHz. A second Ti-sapphire active mode-locked laser (Tsunami, Spectra Physics) emitting at 760 nm, provides the depletion beams. It is synchronized with the excitation laser and delivers Fourier transform limited pulses of ~100 ps duration. A Spatial Light Modulator (SLM LCOS, Hamamatsu) or a set of two Wollaston prisms, conjugated with the sample plane, is used to generate multiple depletion beams. The laser beams are focused on the back focal plane of a high numerical aperture objective (Nikon Apo TIRF 60x NA = 1.49) and illuminate a wide-field region of the sample (~3 µm). The total magnification of the system is 225x, with extra 1.5x and 2,5x added after the tube lens of the microscope. An optical lattice is produced by the interference of the depletion beams at the sample plane and is overlaid with a uniform excitation beam. Two dichroic filters are used for illumination (Chroma 605 DCXR and Chroma T700DCSPXXR-UV). The fluorescence emission is collected with the same objective, filtered from excitation and depletion photons using filters (Semrock, Bandpass 641/75 and Shortpass 720) and sent to a fast CMOS camera (ORCA-Flash4.0, HAMAMATSU). A fast piezo-scanner (PI P-733.3DD) is used for sample scanning.
Figure 2 shows the optical lattice generated by three beams. The beams issued from the SLM with the same polarization are parallel to the optical axis of the objective, and intersect its back focal plane at the vertices of a centered equilateral triangle. The optical lattice depends on the beams’ polarization, and on θ, the angle formed by the optical axis and the beams emerging from the objective . If the beams’ polarization is parallel to one of the three sides of the triangle, their interference produces a hexagonal lattice with a periodicity 2λ/(3n sinθ), where λ is the depletion laser wavelength and n~1.5 the sample refractive index. The intensity profile of the optical lattice, calculated for θ = 60° and displayed in Fig. 2(a), shows a suitable depletion pattern for STED parallelization: an array of zero intensity minima, each of which being surrounded by a nearly uniform high intensity region.
To probe the depletion pattern, we scan a big fluorescent bead (>100 nm) with a piezo-stage over the field of illumination in presence of both depletion and excitation beams, while recording its fluorescence with the CMOS camera. For each scanning step, an image is acquired and the fluorescence intensity, integrated over the Point Spread Function (PSF) of the bead image, is plotted in Fig. 2(b). The signal maxima of the image of Fig. 2(b) correspond to the regions of minimal depletion, which occur at the zero-intensity positions of the optical lattice. While this image provides information about the position of the intensity maxima and minima of the depletion pattern, the depth of the minimum can be obtained by scanning a gold nanoparticle  in the depletion pattern and recording its intensity scattering with an avalanche photodiode (data not shown). Intensity ratios between the minima and maxima less than 3% can be obtained.
As expected, we obtain a hexagonal lattice with periodicity of 390 nm. Our calculations show that for θ = 71°, the interference contrast is maximum (~1) and the intensity distribution around the minima is nearly isotropic in the sample plane . However, at this incidence angle, the transmittance of our objective is low (few percent only). We choose θ = 60°, a compromise maximizing the transmittance of the objective (~30%) and the contrast of optical lattice (~0.98). The elongated shape observed on the bright spots of Fig. 2(b) can be explained by anisotropy of the intensity distribution around the minima when the pattern is constructed at θ deviating from the optimal value of θ = 71°.
A key point for faster OL-STED image acquisition is to generate lattices with the smallest possible unit cell. For this purpose, one can use a four-beam configuration where a smaller unit cell can be obtained with the same angle θ. In Fig. 2, two pairs of beams (one pair of beams polarized along the x axis and the other pair along the y axis) interfere independently and create a square optical lattice (Fig. 2(c)) with a periodicity p = λ/(2n sinθ). These beams with the desired polarization can be prepared by using the SLM and half-wave plates or by a combination of two Wollaston prisms and half-wave plates in order to avoid the diffraction losses inherent to the SLM. In Fig. 2(d) we probe the square depletion pattern obtained for θ~60° by scanning a large fluorescent bead akin to Fig. 2(b). This pattern has a periodicity of 290 nm, smaller than the one obtained in the previous configuration. Better parallelization also requires the largest depletion pattern possible. The size of this pattern is set as a compromise between the desired OL-STED resolution and the available depletion power. Figure 2(d) shows that a 2.9 µm * 2.9 µm field of view can be obtained with efficient depletion using an average laser power of 400 mW (measured before the objective). It corresponds to 100 unit cells of optical lattice and therefore a two order of magnitude gain in scanning area.
3. Results and discussions
OL-STED images of various samples are obtained as follows (Fig. 3(a)): First we scan the sample over a unit cell in the presence of the wide field excitation and the depletion pattern, while acquiring with the sCMOS camera a fluorescence image (128 * 128 pixels) for each scanning step. We then overlay a binary mask on these images, the transparent parts corresponding to the minima positions of the depletion pattern. Therefore, the CMOS camera together with the digital mask act as an array of parallelized “point detectors” (100 detectors, each giving the integrated signal of 13 camera pixels and recording an image of the size of the lattice unit cell). The complete OL-STED image is then obtained by assembling all the unit cell images together. We choose the total magnification of the microscope to be 225, so that the PSF spreads over more than 100 pixels (a pixel of the camera corresponds to ~29 nm at the sample plane). This increases the overall detection dynamics (higher detection saturation) and provides a better spatial discretization of the PSF. The number of camera pixels forming the point detector is chosen such that it optimizes the detection efficiency of a point emitter, minimizes the reading noise and ensures a low cross talk between two adjacent detectors. The ratio of the number of photons measured by the central detector to that measured by a neighboring one is less than 2%. This cross talk can be corrected while reconstructing the final STED images.
Figure 3(b) and 3(c) display 2.9 µm * 2.9 µm images of a sample containing 20 nm fluorescent beads spin-coated on a glass coverslip without and with the 4 beam depletion pattern. Figure 3(b) clearly shows that the PSF of the beads are diffraction limited ~290 nm, while the resolution of the OL-STED image of Fig. 3(c) is well below this limit (typically ~70 nm, see Fig. 3(d)). Since this resolution depends on the depletion intensity, it is better in the center of the pattern where the intensity is maximal. The resolution is ~1.5 less at the edges of the images where the intensity maxima are two times lower (the resolution scales as the inverse square root of the STED beam intensity ).
We apply OL-STED microscopy to image microtubules in fixed COS cells. The microtubules were stained using a standard immunofluorescence protocol involving a primary antibody (anti beta-tubulin) and a secondary antibody labeled with fluorescent Dyes (Atto647N). As shown in Figs. 3(e)–3(g), the OL-STED gives super-resolved images of tubulin fibers. The resolution is below 100 nm and microtubules distant by less than the diffraction limit are clearly distinguished. Importantly, the acquisition time of an OL-STED image ~80 ms is only limited by the CMOS camera readout time and can therefore be further shortened using faster cameras.
Photobleaching can be a constraint in STED nanoscopy when recording large number of frames. To reduce photobleaching, we use a structured illumination instead of a homogeneous one to excite the molecules. Similarly to the depletion optical lattice, we use a SLM (Pluto-VIS, Holoeye) to produce four excitation beams which generate an interfere pattern (Fig. 4(a)) complementary to the depletion pattern (Fig. 4(b)) (having maxima where the other has minima, and vice versa). In this case, excitation of molecules occurs mainly at regions of zero depletion intensities and therefore the probability that a molecule get promoted to highly excited and reactive levels is reduced [32, 33]. To compare the photobleaching rates of the homogeneous and the structured excitation configurations, we record successive OL-STED images of a highly concentrated sample of 20 nm fluorescent beads in the two cases. Figure 4(c) shows the corresponding time evolutions of the total signal (integrated over all the pixels of the image, and normalized to that of the first image) together with their exponential decay fits. In the case of structured illumination, the decay time is found to be two times longer than that of homogeneous case. The high-speed capability of our OL-STED microscope is demonstrated by imaging the movement of 20 nm fluorescence particles in a Carbopol gel (concentration 2%). Successive OL-STED images recorded with the structured excitation at 12.5 frames per second are presented in Fig. 4(d). The fastest movement one can track is limited to particle displacements of less than one super-resolved PSF (70 nm) per one OL-STED image acquisition time (80 ms). Interestingly, we clearly show that recording of fast relative movement of two particles separated by a distance well below the diffraction limit is possible with our methodology.
4. Theoretical comparison of the resolutions of OL-STED and multi-doughnut STED
The OL-STED setup provides many advantages compared to the multi-doughnut configuration . First, with few interfering beams and a single chip detector, we can easily obtain a large number of intensity minima and their corresponding “point detectors”. Secondly, for the same number of local minima and the same target resolution, the OL-STED setup requires up to six-fold less depletion power than a multi-doughnut STED setup (see calculations below). This is due to the fact that with interference one can achieve depletion intensities higher than that obtained for isolated donuts, and better-confined zero-intensity regions. Finally, the periodicity of the optical lattice is much smaller than the distance between the two neighbor doughnuts (several microns), therefore small scan regions and shorter acquisition times are required.
We theoretically compare the resolution obtained in the OL-STED and in the multi-doughnut STED setups. An analytical expression of the doughnut STED resolution (Full Width at Half Maximum) has been derived in .
For high STED intensities this expression reads:
We use the same theoretical approach to establish an analytical expression for 4 beam OL-STED resolution . We first assume that all the beams (excitation and depletion) have uniform intensity profiles. The optical lattice created by the interference of the depletion beams has an intensity distribution at the sample plane:Fig. 2).
The resolution along the x axis (for y = 0) is given by:
To compare the performance of OL-STED to multi-doughnut STED, we calculate the power needed to achieve a given resolution for same number of intensity minima for the two experimental configurations. For this purpose, we first express the intensity maxima function of the total depletion powers andused for each configuration. In the doughnut configuration, using a Laguerre-Gauss mode intensity profile, one obtains , while in the case of an optical lattice of Fig. 2(c) . Equaling Eqs. (1) and (3), we obtain which clearly shows that OL-STED is more efficient than multi-doughnut STED. For ,and , the multi-doughnut STED microscope requires 6.2 times more depletion power than the OL-STED microscope.
We showed that tailored optical lattices allow large parallelization of standard STED microscopy. Super-resolved 2.9 µm x 2.9 µm images with ~70 nm resolution are obtained up to a rate of 12.5 frames per second, limited only by the CMOS camera readout time. A larger field of view can be achieved using a depletion laser with a lower repetition rate , which would provide higher intensity pulses for depletion.
We warmly thank Philippe Tamarat for his valuable help with the laser sources, Olivier Rossier and Grégory Giannone for providing the cell culture and staining, and Laurent Cognet for helpful discussions. We acknowledge financial support from the Agence Nationale de la Recherche, Région Aquitaine, the French Ministry of Education and Research, the European Research Council and FranceBioImaging (Grant N° ANR-10-INSB-04-01).
References and links
2. S. W. Hell and J. Wichmann, “Breaking the diffraction resolution limit by stimulated emission: stimulated-emission-depletion fluorescence microscopy,” Opt. Lett. 19(11), 780–782 (1994). [CrossRef] [PubMed]
4. E. Betzig, G. H. Patterson, R. Sougrat, O. W. Lindwasser, S. Olenych, J. S. Bonifacino, M. W. Davidson, J. Lippincott-Schwartz, and H. F. Hess, “Imaging intracellular fluorescent proteins at nanometer resolution,” Science 313(5793), 1642–1645 (2006). [CrossRef] [PubMed]
7. M. Heilemann, S. van de Linde, M. Schüttpelz, R. Kasper, B. Seefeldt, A. Mukherjee, P. Tinnefeld, and M. Sauer, “Subdiffraction-resolution fluorescence imaging with conventional fluorescent probes,” Angew. Chem. Int. Ed. Engl. 47(33), 6172–6176 (2008). [CrossRef] [PubMed]
9. G. Giannone, E. Hosy, F. Levet, A. Constals, K. Schulze, A. I. Sobolevsky, M. P. Rosconi, E. Gouaux, R. Tampé, D. Choquet, and L. Cognet, “Dynamic superresolution imaging of endogenous proteins on living cells at ultra-high density,” Biophys. J. 99(4), 1303–1310 (2010). [CrossRef] [PubMed]
10. J. Fölling, M. Bossi, H. Bock, R. Medda, C. A. Wurm, B. Hein, S. Jakobs, C. Eggeling, and S. W. Hell, “Fluorescence nanoscopy by ground-state depletion and single-molecule return,” Nat. Methods 5(11), 943–945 (2008). [CrossRef] [PubMed]
11. R. Heintzmann, T. M. Jovin, and C. Cremer, “Saturated patterned excitation microscopy--a concept for optical resolution improvement,” J. Opt. Soc. Am. A 19(8), 1599–1609 (2002). [CrossRef] [PubMed]
12. M. G. L. Gustafsson, “Nonlinear structured-illumination microscopy: wide-field fluorescence imaging with theoretically unlimited resolution,” Proc. Natl. Acad. Sci. U. S. A. 102(37), 13081–13086 (2005). [CrossRef] [PubMed]
13. P. Kner, B. B. Chhun, E. R. Griffis, L. Winoto, and M. G. L. Gustafsson, “Super-resolution video microscopy of live cells by structured illumination,” Nat. Methods 6(5), 339–342 (2009). [CrossRef] [PubMed]
14. E. H. Rego, L. Shao, J. J. Macklin, L. Winoto, G. A. Johansson, N. Kamps-Hughes, M. W. Davidson, and M. G. L. Gustafsson, “Nonlinear structured-illumination microscopy with a photoswitchable protein reveals cellular structures at 50-nm resolution,” Proc. Natl. Acad. Sci. U. S. A. 109(3), E135–E143 (2012). [CrossRef] [PubMed]
15. E. Rittweger, K. Y. Han, S. E. Irvine, C. Eggeling, and S. W. Hell, “STED microscopy reveals crystal colour centres with nanometric resolution,” Nat. Photonics 3(3), 144–147 (2009). [CrossRef]
16. G. Donnert, J. Keller, R. Medda, M. A. Andrei, S. O. Rizzoli, R. Lührmann, R. Jahn, C. Eggeling, and S. W. Hell, “Macromolecular-scale resolution in biological fluorescence microscopy,” Proc. Natl. Acad. Sci. U. S. A. 103(31), 11440–11445 (2006). [CrossRef] [PubMed]
17. U. V. Nägerl, K. I. Willig, B. Hein, S. W. Hell, and T. Bonhoeffer, “Live-cell imaging of dendritic spines by STED microscopy,” Proc. Natl. Acad. Sci. U.S.A. 105(48), 18982–18987 (2008). [CrossRef] [PubMed]
18. V. Westphal, S. O. Rizzoli, M. A. Lauterbach, D. Kamin, R. Jahn, and S. W. Hell, “Video-rate far-field optical nanoscopy dissects synaptic vesicle movement,” Science 320(5873), 246–249 (2008). [CrossRef] [PubMed]
19. S. W. Hell and M. Kroug, “Ground-state-depletion fluorescence microscopy: A concept for breaking the diffraction resolution limit,” Appl. Phys. B 60(5), 495–497 (1995). [CrossRef]
20. M. Hofmann, C. Eggeling, S. Jakobs, and S. W. Hell, “Breaking the diffraction barrier in fluorescence microscopy at low light intensities by using reversibly photoswitchable proteins,” Proc. Natl. Acad. Sci. U. S. A. 102(49), 17565–17569 (2005). [CrossRef] [PubMed]
21. T. Grotjohann, I. Testa, M. Reuss, T. Brakemann, C. Eggeling, S. W. Hell, and S. Jakobs, “rsEGFP2 enables fast RESOLFT nanoscopy of living cells,” eLife 1, e00248 (2012).
23. A. Chmyrov, J. Keller, T. Grotjohann, M. Ratz, E. d’Este, S. Jakobs, C. Eggeling, and S. W. Hell, “Nanoscopy with more than 100,000 ‘doughnuts’,” Nat. Methods 10(8), 737–740 (2013). [CrossRef] [PubMed]
24. B. Yang, F. Przybilla, M. Mestre, J.-B. Trebbia, and B. Lounis, “Massive parallelization of STED nanoscopy using optical lattices,” arXiv:1307.3833 [physics] (2013).
26. G. Grynberg, B. Lounis, P. Verkerk, J.-Y. Courtois, and C. Salomon, “Quantized motion of cold cesium atoms in two- and three-dimensional optical potentials,” Phys. Rev. Lett. 70(15), 2249–2252 (1993). [CrossRef] [PubMed]
33. J. Vogelsang, R. Kasper, C. Steinhauer, B. Person, M. Heilemann, M. Sauer, and P. Tinnefeld, “A Reducing and oxidizing system minimizes photobleaching and blinking of fluorescent dyes,” Angew. Chem. Int. Ed. Engl. 47(29), 5465–5469 (2008). [CrossRef] [PubMed]