D2.1 Operating 2D STED Microscope

D2.1 Operating 2D STED Microscope Nature: Report Dissemination Level: Public Lead Beneficiary: UNIVDUN Author(s): Piotr Zdankowski Work Package: WP2 Task: ESR5 Version: 0.02 Last modified: 24/04/2017 Status: Complete Approved by: Kees Weijer Date: 24/04/2017 <delete the text below if the Dissemination Level is not Public> Disclaimer: The material contained in this document is provided for information purposes only. No warranty is given in relation to use that may be made of it and neither the copyright owners or the European Commission accept any liability for loss or damage to a third party arising from such use. Copyright Notice: Copyright PHOQUS Consortium 2016. All rights reserved. Page 1 of 16 Version: 0.02 Status: Released

Table of Contents 1 Introduction... 3 1.1 STED Microscopy... 3 2 STED Microscope design... 4 2.1 Microscope optical design... 4 2.1.1 Depletion laser beam path... 4 2.1.2 Excitation laser beam path:... 5 2.1.3 Pinhole size calculations.... 5 2.2 Microscope control software... 6 3 Results... 7 3.1 Evaluation of the microscope performance.... 7 3.1.1 Gold beads measurements.... 7 3.1.2 Laser pulse measurements... 10 3.1.3 Fluorescent beads measurements... 11 4 Conclusions.... 15 Page 2 of 16 Version: 0.02 Status: Released

1 Introduction Fluorescence microscopy has been a well-established and extremely useful technique to image biological structures. However, its resolution has always been restricted by the diffraction limit. The introduction of super-resolution microscopy techniques [1] that emerged at the turn of the last century allowed us to see into nanostructures and step into the world of nanoscopy. Among the various super-resolution techniques, Stimulated Emission Depletion (STED) Microscopy [2] has been especially interesting as it relies solely on optical phenomena and does not require any additional computational processing. STED has proven to be able to resolve structures as small as 20nm [3]. To date, there are a few companies that offer a commercial STED microscopes, but their price tag is extremely high and they do not offer the flexibility of a home-built system. What is more, imaging through thick tissues is a very challenging task as every biological sample introduces aberrations and scatters propagating light. This problem becomes notably more significant in the STED microscopy as even small amount of aberrations can degrade the depletion beam shape substantially [4]. Lastly, STED microscopy relies on very high depletion beam powers which can cause significant photodamaging of samples and it is important to minimise the exposition of the sample to the STED beam [5]. This can be achieved using fast resonant scanning over the sample. Our approach was to develop a custom-built system with adaptive optics (AO) correction of the depletion beam to overcome the issues with commercial systems and which can be tailored to the end applications. 1.1 STED Microscopy STED microscope is a type of confocal scanning fluorescence microscope which allows the acquisition of images with a resolution well beyond the Abbe s resolution limit [6], which is dependent on the wavelength of the imaging light and the numerical aperture (NA) of the imaging optics. The principle of operation of the fluorescent microscope is that first, we have to label the chosen structures of the sample with fluorescent markers. Then, by illuminating them with an appropriate wavelength, we cause the molecules of illuminated structure to change their energetic state from ground to the excited state (increasing their energy). Since the molecules prefer to stay in the ground state, they will start to release their energy in the form of fluorescent photons. This phenomenon is called spontaneous emission. Those photons are later separated from the excitation photons and collected with the photodetector. In order to build an image, we need to scan the sample point by point, collecting fluorescent photons from each point of the structure. The problem appears, when we want to image structures smaller than the resolution limit. Due to the diffraction and limited size of imaging optics, the microscope objective has a confined NA. The light that passes through the objective can never be focused into an infinitely small focal spot meaning that we will collect photons from all the structures that are in the area of the illuminating focal spot. This has the consequence that we will not be able to distinguish objects smaller than the size of that spot. STED microscopy is one solution to this problem. If we shine the light onto the excited molecules we can turn them off (deplete them) by making them release the photons in the form of stimulated emission. Now, if we shape the depleting beam in the form of the donut and superimpose it with the excitation beam, we will effectively be exciting the structures that are in the area of the donut hole, which can be significantly smaller than the area of the focal spot. Scanning the sample with those two superimposed beams enables the acquisition of images with the resolution much higher than classical diffraction limited microscopes [7][8]. Page 3 of 16 Version: 0.02 Status: Released

2 STED Microscope design 2.1 Microscope optical design Fig. 1. Schematic of the STED microscope. M1 to M6 correspond to broadband mirrors, SLM stands for Spatial Light modulator, f1 to f11 are lenses, GLP is a Glan Laser Polariser, DM1 and DM2 are dichroic mirrors, GM and RM are galvanometric and resonant scanning mirrors respectively, MO is a microscope objective, EF is a fluorescence emission filter, MPPD and SPAD are photon counting detectors. 2.1.1 Depletion laser beam path The 766nm depletion laser is a 500ps pulsed diode laser (VisIR STED PicoQuant), specially designed for the purpose of STED microscopy. It has a beam diameter of 2.2mm. Then it passes through the half-wave plate /2 to rotate its polarisation. Lenses f1 and f2 are responsible for beam expansion. After expansion beam size is 16mm, so that the beam diameter overfills the Spatial Light Modulator (SLM) aperture. After the expansion, the depletion beam passes through the half-wave plate which rotates the polarisation in order to maximise the phase modulation of the SLM. The SLM displays an appropriate phase mask to shape the depletion beam. This phase mask is a circular aperture with a blazed grating, which is necessary, since it separates the diffraction orders after the beam reflects from the SLM. The modulated phase mask is represented in the 1 st diffraction order of the blazed grating. The size of the aperture is set to 557 pixels which equating to a 11mm beam size diameter Page 4 of 16 Version: 0.02 Status: Released

(SLM pixel pitch is equal to 20 m). After the SLM, the depletion beam is decreased to 1.83mm by a 4f system and passes through the Glan-Laser polariser (GLP), that only transmits polarisation corresponding to the SLM polarisation. The fast axis of the GLP is set so that the 1 st diffraction order is maximised. DM1 is a fluorescence dichroic mirror (Semrock FF720- SDi01), which reflects the STED beam and transmits fluorescence. Further on, the STED beam passes through the dichroic mirror DM2 (Chroma ZT647rdc-UF3), where it is combined with the excitation beam. There are three 4f systems in the beam path: 4f1: f6 (f=400mm) and f7 (f=40mm), imaging the SLM plane containing the hologram onto the plane in-between two galvanometric mirrors GM 4f2: f8(f=80mm) and f9(f=100mm) imaging the RM plane onto resonant mirror RM plane 4f3: f10(f=50mm) and f11(400mm) imaging RM plane onto the back focal plane of the microscope objective MO (100x Nikon Plan Apo, NA=1.45). 2.1.2 Excitation laser beam path: The excitation beam laser is a diode pulsed laser with the wavelength of 637nm and is coupled into polarisation maintaining fibre. At the output of the fibre it passes through the half-wave plate /2, which rotates the polarisation to match the STED beam polarisation and then is collimated by the aspheric lens collimator f3 (f=10.99mm). After f3 the beam size is 2.1mm. Lenses f4 (f=60mm) and f5 (f=200mm) expand the excitation beam to the size of 7mm. After the beam expander the excitation beam is reflected from DM2, where it is combined with the STED beam. After DM2 both the STED and excitation beams propagate coaxially. Both beams then pass through two 4f systems (4f1 and 4f2) before passing through the quarter-wave plate /4, which converts the beam polarisation to circular. Then the beams go through the MO and are focused on the sample. The excitation beam that goes into the MO should overfill the pupil diameter, however size of the phase mask image of the depletion beam should be exactly matching the pupil diameter of MO. The pupil diameter can be calculated from the following formula: D = 2NA f For the applied Nikon microscope objective, that has NA = 1.45 and f = 2mm, the pupil diameter is equal D = 5.8mm. Depending on the sample type that is imaged, there are two different possible imaging paths: a reflective path and a fluorescent path. The imaging beam propagates back through the same beam path as the excitation/sted beam path, until it reaches: a) a pellicle beamsplitter P1 which reflects the imaging beam onto the imaging lens FiR (f=80mm), which focuses it through the pinhole PHR and falls onto the photon counting detector MPPC; b) a dichroic mirror DM1, which transmits the fluorescent beam, then focusses it through the imaging lens FiF (f=80mm), passing through the set of 3 filters: EF1 (Semrock FF01-750/SP), EF2 (Semrock FF01-676/37) and EF3 (Semrock FF650/100). After the filters the beam passes through the pinhole PHF and falls on the photon counting detector SPAD. 2.1.3 Pinhole size calculations. An Airy disc has a radius according to the Abbe s formula of: Page 5 of 16 Version: 0.02 Status: Released

r = 0.61λ 0.61 637nm = = 268nm NA 1.45 This means that for the excitation beam wavelength 637nm, the Airy disc radius size is equal to r=268nm. Taking into account the magnification of the whole setup, the Airy disc radius size in the fluorescent pinhole plane PHF is equal to: r F = 0.268 f11 f MO f9 f10 f7 f8 FiF f6 = 10.72μm Similarly, the Airy disc size for imaging reflective objects PHR is equal to: r R = 0.268 f11 f MO f9 f10 f7 f8 FiR f6 = 10.72μm The chosen pinhole size for fluorescent imaging is 1A.U. (Airy units), which means that the actual pinhole size is 20 m. For reflective objects, the chosen pinhole size is approx. 20A.U, for which the actual pinhole size is 400 m, which allows the system to have quasi-widefield imaging. 2.2 Microscope control software The software used for the STED microscope control has been written in Labview. The acquisition and scanning is carried out using a National Instruments FPGA card. The STED microscope uses a 16kHz resonant mirror for x-axis scanning and a slower, galvanometric mirror for y-axis scanning. 16kHz scanning means that the mirror does a return scan in 62.5 s meaning that one line is scanned in half of the period, 31.25 s. The FPGA can acquire digital input data with 80MHz clock frequency (1 clock tick is equal to 12.5ns), meaning there are 2500 ticks available for the photon counting. The number of pixels per line (ppl) determines the number of photon counts per pixel that can be acquired by the detector in the resonant mirror scan time (e.g. for 250ppl, the maximum number of counts is 10, for 500ppl the maximum number of counts is 5). The acquisition software has the option of multiple line scans, so that the count number in the image can be increased. Page 6 of 16 Version: 0.02 Status: Released

3 Results 3.1 Evaluation of the microscope performance. 3.1.1 Gold beads measurements. In order to measure the performance of both excitation beam quality and STED beam quality it is best to image the beams and calculate their point spread functions (PSF). It is not possible to measure both beams using fluorescent detection, since there are no such beads that would be excited at both wavelengths (637nm and 766nm). Hence the measurements of the PSF for both beams are carried out by imaging gold beads and collecting their reflected signal. Firstly, the PSF of the microscope has been simulated using the following parameters: Pixel size: x = 26nm, y = 26nm, z = 30nm. Number of z stacks: 100 NA = 1.45 Tables 1-2 contain the calculated full width at half maximum (FWHM) of the simulated beads of different pixel sizes and for different wavelengths. a) b) d) e) f) c) Fig. 2. 256x256 pixels simulation of the PSF for 637nm using Born&Wolf algorithm [9]. a) presents xy plane of the PSF, b) yz profile, c) xz profile; d) x profile and Gaussian fitting, e) y profile and Gaussian fitting, f) z profile and Gaussian fitting. Table 1. Calculated FWHM for the simulated PSF shown in fig. 2. Number of Pixels Pixel size FWHM x 256 26 nm 0.219 μm y 256 26 nm 0.219 μm z 100 30 nm 0.763 μm Page 7 of 16 Version: 0.02 Status: Released

a) b) d) e) f) c) Fig. 3. 256x256 pixels simulation of the PSF for 766nm using Born&Wolf algorithm. a) presents xy plane of the PSF, b) yz profile, c) xz profile; d) x profile and Gaussian fitting, e) y profile and Gaussian fitting, f) z profile and Gaussian fitting. Table 2. Calculated FWHM for the simulated PSF shown in fig. 3. Number of Pixels Pixel size FWHM x 256 26 nm 0.262 μm y 256 26 nm 0.262 μm z 100 30 nm 0.938 μm Having the simulated PSF for the experimental microscope configuration the next step is experimental verification carried out by imaging gold beads. 150nm gold nanoparticles from BBI solutions were diluted in ethanol and spread onto the coverslip, which was then mounted with a 97% solution of Thiodiethanol on the microscope slide. Fig. 4-5 shows the obtained PSF for both excitation beam and STED beam. Tables 3 and 4 shown the FWHM values calculated for both beams. Gold beads were recorded for two different pixel sizes and total number of pixels. The field of view was 5 m x 5 m. a) b) d) e) f) c) Fig. 4. 256x256 pixels PSF of 150nm gold bead image recorded with excitation laser. a) presents xy plane of the PSF, b) yz profile, c) xz profile; d) x profile and Gaussian fitting, e) y profile and Gaussian fitting, f) z profile and Gaussian fitting. Page 8 of 16 Version: 0.02 Status: Released

Table 3. Calculated FWHM for the recorded PSF of 256x256 pixels 150nm gold bead recorded with excitation laser. Number of pixels/stacks Pixel size FWHM x 256 20 nm 0.247 μm y 256 20 nm 0.233 μm z 100 20 nm 0.708 μm a) b) d) e) f) c) Fig. 5. 256x256pixels PSF of 150nm gold bead image recorded with STED laser. a) presents xy plane of the PSF, b) yz profile, c) xz profile; d) x profile and Gaussian fitting, e) y profile and Gaussian fitting, f) z profile and Gaussian fitting. Table 4. Calculated FWHM for the recorded PSF of 256x256 pixels 150nm gold bead recorded with STED laser. Number of pixels/stacks Pixel size FWHM x 256 20 nm 296.163 y 256 20 nm 243.132 z 100 20 nm 880.927 Imaging gold beads also helps to image and evaluate the quality of the phase mask applied on the SLM as well as spatial alignment of both STED and excitation beams. Imaging the phase mask is very important, as aberrations can significantly distort the Laguerre-Gaussian beam. Corrected and aligned beams with different phase masks are shown on fig. 6-7. Page 9 of 16 Version: 0.02 Status: Released

a) b) Fig. 6. STED beam phase mask imaged with the 150nm gold beads. a) vortex phase mask creating Laguerre-Gaussian beam creating 2D STED beam (donut beam); b) π-step phase mask creating a 3D STED beam a) b) Fig. 7. STED beam phase mask imaged with the 150nm gold beads merged with the gold bead recorded with the excitation beam. This shows spatial alignment of the excitation and STED beams. a) vortex phase mask creating Laguerre-Gaussian beam creating 2D STED beam (donut beam); b) π-step phase mask creating a 3D STED beam 3.1.2 Laser pulse measurements With good spatial alignment of the STED and excitation beams, the next step is to check the temporal alignment of the pulses of both lasers. In order to do that, pulses were measured with a 2GHz photodiode (DET025A/M) and a fast 2GHz oscilloscope. Pulses can be finely tuned using the electronic pulse delay unit (MPD Picosecond Delayer). Results of the pulse measurements are shown in fig. 8. Page 10 of 16 Version: 0.02 Status: Released

Fig. 8. Pulse measurements for excitation beam and STED beam. Pulses had to be measured separately for each laser, since the time resolution of the diode was not good enough to be able to distinguish both pulses simultaneously. The STED pulse should arrive approx. 150ps after the excitation pulse. Measurements shown in fig. 8 show both pulses measured for the delay set to 2020ps, which makes the STED beam arrive ~160ps after the excitation pulse. The pulses are set to 40MHz repetition rate. 3.1.3 Fluorescent beads measurements Fig. 9 shows images of fluorescent beads acquired in the confocal mode and for both 2D and 3D STED phase mask. There is a significant increase in the resolution of the STED images well below the diffraction limit. The beads are 23nm GattaBeads filled with Atto647n dye. FWHM was calculated and is shown in tables 5-7 for all 3 cases: no STED (table 5) phase mask, 2D phase mask (table 6) and 3D phase mask (table 7). Fig. 9. Comparison of the images with a) no STED beam, b) 2D STED phase mask, c) 3D STED phase mask. 512x512 images, pixel size 19nmx19nm, 40MHz laser repetition rate Page 11 of 16 Version: 0.02 Status: Released

250nm 620nm 55nm 600nm 100nm 165nm Fig. 10. Comparison of the confocal (a), 2D STED (b) and 3D STED (c) images and their orthogonal views. Images on the left show acquired bead images, images in the right show zoomed bead shown in the blue frame and its orthogonal views YZ (right) and XZ (bottom). Page 12 of 16 Version: 0.02 Status: Released

Table 5. Full width at half maximum of the confocal image obtained for the beads shown in fig. 9. Bead Number 1 2 3 4 5 6 Average FWHM X [nm] 227 245 241 236 228 248 238 FWHM Y [nm] 278 279 292 277 275 302 284 FWHM XZ [nm] 606 584 627 596 662 630 618 FWHM YZ [nm] 599 568 588 586 591 678 602 Table 6. Full width at half maximum of the 2D STED image obtained for the beads shown in fig. 9. Bead Number 1 2 3 4 5 6 Average FWHM X [nm] 54 61 46 53 52 80 57 FWHM Y [nm] 37 57 50 59 45 52 50 Table 7. Full width at half maximum of the 3D STED image obtained for the beads shown in fig. 9. Bead Number 1 2 3 4 5 6 Average FWHM X [nm] 123 96 126 125 107 119 116 FWHM Y [nm] 100 127 125 102 138 127 120 FWHM XZ [nm] 165 204 167 144 119 198 166 FWHM YZ [nm] 177 154 161 136 148 198 162 Experiments to demonstrate the STED microscope usability on the biological samples were carried out. To show the increased resolution achieved by STED microscopy microtubule networks in He-La and RPE-1 cells were investigated. In both samples, α-tubulin was stained with the Atto647n anti-mouse secondary antibody. Resulting images are shown in fig. 11 and 12 (microtubules) and fig. 13 (mitotic cell). Fig. 11. He-La cell microtubules stained with ATTO647N dye. Left confocal image, right 2D STED image. Page 13 of 16 Version: 0.02 Status: Released

Fig. 12. RPE-1 cell microtubules stained with ATTO647N dye. Left confocal image, right 2D STED image. Fig. 13. RPE-1 mitotic cell stained with ATTO647N dye. Left confocal image, right 2D STED image. Page 14 of 16 Version: 0.02 Status: Released

4 Conclusions. In the previous sections, we presented clear evidence that we have designed and built a fully functional custom built STED microscope. The performance of the microscope in terms of 2 and 3D resolution is comparable to that of available commercial systems, but it has the huge advantage of the customisation possibilities. The microscope can be used with widely available fluorescent dyes. By selecting an appropriate phase mask on the SLM it can acquire 2D and 3D super-resolution images. The achievable lateral resolution is 50nm and the axial resolution is 165nm. The microscope has been used to acquire super-resolution images of microtubule networks in fixed interphase and mitotic cells. Page 15 of 16 Version: 0.02 Status: Released

5 References [1] E. Betzig, S. W. Hell, and W. E. Moerner, How the optical microscope became a nanoscope, pp. 1 7, 2014. [2] S. W. Hell and J. Wichmann, Breaking the diffraction resolution limit by stimulated emission: stimulated-emission-depletion fluorescence microscopy., Opt. Lett., vol. 19, no. 11, pp. 780 2, Jun. 1994. [3] F. Göttfert, C. a Wurm, V. Mueller, S. Berning, V. C. Cordes, A. Honigmann, and S. W. Hell, Coaligned dual-channel STED nanoscopy and molecular diffusion analysis at 20 nm resolution., Biophys. J., vol. 105, no. 1, pp. L01-3, Jul. 2013. [4] B. R. Patton, D. Burke, R. Vrees, and M. J. Booth, Is phase-mask alignment aberrating your STED microscope?, Methods Appl. Fluoresc., vol. 3, no. 2, p. 24002, 2015. [5] Y. Wu, Reducing photobleaching in STED microscopy with higher scanning speed, arxiv Prepr. arxiv1408.1507, 2014. [6] E. Abbe, Beiträge zur Theorie des Mikroskops und der mikroskopischen Wahrnehmung, Arch. für Mikroskopische Anat., vol. 9, no. 1, pp. 413 418, Dec. 1873. [7] J. B. Pawley, Handbook Of Biological Confocal Microscopy. Boston, MA: Springer US, 2006. [8] a Cornea and P. Conn, Fluorescence Microscopy: Super-resolution and Other Novel Techniques. Academic Press, 2014. [9] B. Richards and E. Wolf, Electromagnetic Diffraction in Optical Systems. II. Structure of the Image Field in an Aplanatic System, Proc. R. Soc. London. Ser. A. Math. Phys. Sci., vol. 253, no. 1274, p. 358 LP-379, Dec. 1959. Page 16 of 16 Version: 0.02 Status: Released