Open Access
Add to BibSonomyAbstract
Most structured illumination microscopes use a physical or syn–thetic grating that is projected into the sample plane to generate a periodic illumination pattern. Albeit simple and cost-effective, this arrangement hampers fast or multi-color acquisition, which is a critical requirement for time-lapse imaging of cellular and sub-cellular dynamics. In this study, we designed and implemented an interferometric approach allowing large-field, fast, dual-color imaging at an isotropic 100-nm resolution based on a sub-diffraction fringe pattern generated by the interference of two colliding evanescent waves. Our all-mirror-based system generates illumination pat-terns of arbitrary orientation and period, limited only by the illumination aperture (NA = 1.45), the response time of a fast, piezo-driven tip-tilt mirror (10 ms) and the available fluorescence signal. At low µW laser powers suitable for long-period observation of life cells and with a camera exposure time of 20 ms, our system permits the acquisition of super-resolved 50 µm by 50 µm images at 3.3 Hz. The possibility it offers for rapidly adjusting the pattern between images is particularly advantageous for experiments that require multi-scale and multi-color information. We demonstrate the performance of our instrument by imaging mitochondrial dynamics in cultured cortical astrocytes. As an illustration of dual-color excitation dual-color detection, we also resolve interaction sites between near-membrane mitochondria and the endoplasmic reticulum. Our TIRF-SIM microscope provides a versatile, compact and cost-effective arrangement for super-resolution imaging, allowing the investigation of co-localization and dynamic interactions between organelles – important questions in both cell biology and neurophysiology
© 2013 Optical Society of America
1. Introduction
Fluorescence microscopy is an essential tool for studying sub-cellular dynamics, but image resolution is diffraction-limited. Super-resolution microscopies (see [1–3]) for review) over-come this deficiency, either by point-spread function engineering (as for STED [4],) or by localizing individual fluorophores and accumulating their positions over time (in the case of PALM [5] and STORM [6]). These techniques attain lateral resolutions of a few tens of nm, and when combined with astigmatic detection optics and/or TIRF, also high axial resolution. But they either impose important constraints on the fluorophores that can be used, cannot easily provide multi-color co-localization information, or are too slow for imaging large fields. Restrictive for live-cell imaging, STED augments photobleaching and damage, making long-period observation of live cells impractical. STED, PALM and STORM all achieve spatial resolution gains at the expense of temporal resolution (for STED due to its point scanning nature; for the localization microscopies due to the need to accumulate individual-fluorophore positions from thousands of images). Many biological applications, however, require simultaneously a large field-of-view and a high frame rate because signaling events, organelle dynamics and/or membrane fusion events can happen at any place, anytime.
For such live-cell time-lapse imaging structured illumination microscopy (SIM) [7–10] appears ideal because it offers a compromise between spatial and temporal resolution [11]. Using wide-field excitation with the projection of a diffraction-limited fringe pattern, SIM doubles lateral spatial resolution over the entire field-of-view, obtaining an even greater enhancement in its saturated versions. Further resolution gains can be obtained when using total internal reflection fluorescence (TIRF) [12–18] because sub-diffraction fringe periods can be generated in the near-field with standing evanescent waves (EWs) [19, 20]. In 2-D SIM, the final super-resolved image is typically computationally reconstructed from nine wide-field images acquired at three grating orientations and three phase shifts each. Successful reconstructions using fewer images have been reported [21]. TIRF-SIM has been implemented mostly in the prismless TIRF geometry [22]. Here, two focused spots are positioned at opposite eccentric positions in the back-focal plane (BFP) of a high-numerical aperture (NA) objective to set up a standing EW-field. In order to turn the pattern orientation, the spots are rotated with scanning optics, for which spatial light modulators (SLMs), acousto-optic or electro-optic elements have been used. These devices increase cost, constrain both excitation wavelength and polarization and (at least for SLMs) limit temporal resolution. Surprisingly, beyond the initial proof-of-principle experiments, there have been only a few follow-up studies with TIRF-SIM and only a handful imaged live biological samples.
With routine biological applications and a use in core imaging facilities in mind, we present here a simple, compact and inexpensive implementation of TIRF-SIM built from optical bench components on a 50 cm by 50 cm breadboard. This system allows multi-color wide-field imaging with 100-nm isotropic resolution over the entire 50 by 50-µm field of view. An improved phase retrieval and image processing algorithm [23] removes the earlier requirement for laborious real-time feedback control for fringe stabilization [15]. In this report, we first characterize our set-up by imaging static and moving sub-resolution fluorescent microspheres. Under conditions realistic for biological imaging (µW incident power, 20-ms exposure time) we acquired super-resolved images at 3.3 Hz temporal resolution for each color channel. Dynamics and organization of near-membrane mito-chondrial and endoplasmic reticulum (ER) networks in cortical astrocytes are reported and sub-diffraction details of their interaction sites are visualized.
2. Material and methods
2.1 Dyes and reagents
TransFluoSpheres of 93-nm diameter (λex,max/ λem,max = 505/515 nm, 488/645 nm and 540/560 nm, respectively), Mito-EGFP (488/510 nm), Mitotracker Deep Red (660/680 nm), ER-EGFP (488/510 nm) and cell culture media were purchased from Invitrogen (Saint Aubin, France). All other drugs were from SigmaAldrich (Lyon, France).
2.2 Cells and solutions
Experiments followed EU and institutional guidelines for the care and use of laboratory animals (Council directive 86/609EEC). Astrocytes were prepared from the cortices of newborn mice and cultured as described elsewhere [24]. BK-7 coverslips (Thermo Fisher Menzel-Gläser, Braunschweig, Germany) were treated and coated with poly-ornithine as described [25]. During experiments, cells were kept in a static bath of extracellular saline containing (in mM): 140 NaCl, 5.5 KCl, 1.8 CaCl2, 20 glucose, 10 HEPES (pH 7.3, adjusted with NaOH). Isolated astrocytes were imaged after labeling at 37°C with Mitotracker Deep Red (0.1 µM, 3 min), or after overnight transfection with mito-EGFP or ER-EGFP (2 µg/µl), using published protocols. Cells were rinsed five times with dye-free solution and imaged ~15 min later. All experiments were performed at room temperature (20-23°C).
2.3 Excitation optical path
We built an off-axis Michelson interferometer, Fig. 1. The beam of an OPSL (Sapphire 448-20, Coherent, Utrecht, The Netherlands) was filtered to remove the 956-nm pump line (BG18, Schott, Mainz, Germany), optically isolated (FI-488-5SI Faraday rotator, Linos-Qioptiq, Göttingen, Germany), shuttered (LS3, Uniblitz, Vincent Assoc., Rochester, NY) and delivered to the optical table via a mono-mode optical fiber (Qioptiq PointSource, Hamble, UK). Polarization was adjusted with a zero-order half-wave plate (refer to Fig. 1; λ/2, WPH05M-488, Thorlabs, Newton, NJ). A first mirror M1 mounted on a piezoelectric tip/tilt-platform (TT; S-330.8SL, Physik Instrumente (PI), Karlsruhe, Germany) scanned the beam, which was then split up into two with a non-polarizing 50/50 cube (Qioptiq-Linos). The transmitted part was re-directed laterally off-set with a retroreflector (RE; PS971M-A, Thorlabs, Newton, NJ) while the reflected part was back-reflected by a piezo-mounted (PA, P-010.00H, PI) second plane mirror (M2). This arrangement created two beams at symmetric off-axis positions. Varying the azimuthal and polar angle (ϕ, θ) of the M1 changed the distance between the two spots as well as the orientation of the meridional plane containing them. The phase φ between the two beams was controlled by moving M2. A converging achromatic lens (FL; f = 300 mm, Linos) focused the beams in the BFP of a piezo-mounted (not shown; PIFOC P-712.CLQ, PI) high-NA objective (obj; PlanApo × 60/NA1.45oil TIRFM, Olympus, Hamburg, Germany), producing a standing EW-field across the (50 µm)2 field-of-view. A pattern period of 175 nm, three pattern directions (0, 60 and 120°) and three phase shifts of 120° each were used.
Fig. 1 , Schematic layout of the excitation optical path of our compact custom TIRF-SIM. Laser: 488-nm OPSL/561-nm DPPS, λ/2: wave plate, TT: tip/tilt-mirror, BS: 50/50-beam splitter, RE: retro-reflector, PI: piezoelectric actuator, M1, M2: mirrors, FL: f = 300-mm focusing lens, DC: dichroic, TL: tube lens (f = 300mm), obj: × 60/NA1.45 PlanApo objective, BFP: back-focal plane, CS: coverslip. See main text for details. Inset, measured interference patterns created at the sample plane and schematic focused-spot positions in the objective BFP.
Shortening of EW vector in propagation direction (along the reflecting interface) leads to a node spacing s = λ0/(2n2 sinθ) of the evanescent fringes, where λ0 and n2 are the vacuum wavelength and substrate refractive index, respectively. Like s (that determines the attainable lateral resolution) the EW penetration depth δ = λ0/[4π (n22sin2θ-n12)1/2] (that determines the axial optical sectioning) depends linearly on λ0. Thus, any wavelength-dependence can be corrected by adjusting θ. Of note, s does not depend on the sample refractive index n1 and can be smaller than the Rayleigh limit.
A 561-nm DPPS laser (Compass 561-40, Coherent) served for dual-color imaging. After adjusting its polarization with a zero-order λ/2 plate (WPH05M-561, Thorlabs), both laser beams were combined with a 2-mm thick (λ/10-flatness) dichroic mirror (zt491RDCXT, AHF, Tübingen, Germany) and travelled along the same path up to the TT mirror. The beams were alternately shuttered for dual-color SIM.
2.4 Emission optical path
Fluorescence was collected through the same objective as used for excitation, and was extracted with a 2-mm thick (λ/10-flatness) dual-band dichromatic reflector (U-T51004-v2, Chroma, Bellows Falls, VT). Two stacked 488/568-nm holographic dual-notch filters (Barr Assoc., Westford, MA) blocked the reflected beams. An achromatic lens (f = 300 mm, Linos) focused the collected fluorescence onto a water-cooled sCMOS detector (PCO.edge, PCO, Kelheim, Germany). To maximize the fluorescence collection, no futher emission filters were used. Measured pixel size in the sample plane was 65 nm. A LABVIEW (National Instruments (NI), Austin, TX) code was written to synchronize the beam shutters, pattern orientation, phase shift, and image capture that were all controlled via the output of a NI 9264 DAQ board (NI). The set-up was contained in a light-tight box to shield it from ambient light, dust and air-condition draught. We found that room ventilation and cooling fans had to be cut.
2.5 Data processing, image analysis and statistics
For calculating TIRF-SIM images, we used a reconstruction algorithm that uses a MATLAB/DipImage-based a posteriori phase retrieval based on cross-correlations and allowing the optimization of the phase pattern even when the pattern itself is too fine for detection [23]. Fluorescence image analysis was performed in ImageJ (NIH). Intensity profiles were scaled by first subtracting a constant background intensity B from the TIRF images and normalization to peak fluorescence. Since the reconstruction clips the background of TIRF-SIM images to zero, we obtained B on the raw TIRF images in a local region of interest (ROI) devoid of fluorophore labeling.
3. Results
3.1 Fast, large-field, multi-color imaging at isotropic 100-nm resolution
TIRF-SIM images of sub-diffraction green-fluorescent (505/515 nm) microspheres sprinkled on a glass coverslip showed the expected lateral-resolution increase across the entire 50-µm by 50-µm field-of-view of our microscope, Fig. 2, (A) vs. (B). Zooming in on an individual bead shows the reduced apparent bead size. The increased capacity to image finer detail is also evident from the near-isotropic increase in the observable region in spatial-frequency space. Image Fourier transforms (FTs) are shown in the respective bottom insets. Intensity profiles across the bead shown in the inset, along with examples of other beads taken at different excitation/emission wavelengths, are displayed in Fig. 2(C). Gaussian fits with the beads intensity profiles were used to measure the full-width half-maximal diameter (FWHMmeas), from which the FWHM of the point spread function (PSF) without the effect of the finite bead size was obtained as FWHMPSF = (FWHMmeas2 - FWHMbead2)1/2 [11]. Here, FWHMbead = 2(ln2/2)1/2d, and d = 93 nm is the bead diameter specified by the supplier. We note that the numbers reported in table one are not the true PSFs (for which smaller beads should be used) but rather indicative of the size at which a ~100-nm object will appear.
Fig. 2 TIRF-SIM multi-color imaging at 100-nm resolution. TIRF (A) and TIRF-SIM (B) images of 93-nm diameter green-fluorescent beads (ex./em. maxima 505/515 nm), integration time 20 ms per image. Insets (top), zoom on individual bead, scale bar: 0.5 µm. (bottom), Fourier transforms (FT) of the TIRF and TIRF-SIM images show the increased spatial-frequency band-pass. (C), Representative fluorescence-intensity profiles of beads imaged in TIRF and TIRF-SIM, respectively. Excitation was at 488 for the left (same green-fluorescent bead as depicted in (A),(B), red line), and middle panel (red-emitting bead). Right: 561-nm excitation of a yellow-green emitting bead. Images were taken with a pattern period s of 175 nm at 20-ms integration time (i.e., 306 ms for acquisition of nine images used for the calculation of final super-resolved image). We used low µW laser powers in the sample plane.
Table 1. Measured full-width at half-maximum diameters (FWHMmeas) of 93-nm fluorescent beads excited at 488 or 561 nm. FWHMPSF values are corrected for the finite bead size. Numbers are means ± SD for n = 10 beads for each color.
Axially, the optical sectioning of our microscope is determined by the axial intensity decay of the EW. At a beam angle θ corresponding to a 175-nm pattern period, δ is ~93 nm (see [25]). Thus, with our multi-color TIRF-SIM we obtain, at 20-ms integration time, an isotropic resolution of around 100 nm at 3.3 Hz over a 50-µm × 50-µm field of view.
3.2 Time-lapse imaging of moving samples
We used nine component images to compute the final TIRF-SIM image [23]. We systematically used a set of five criteria to assess reconstruction quality: (i), the reconstructed image should not show a residual hexagonal pattern; (ii), the Fourier transform (FT) of the reconstructed image should be devoid of residual intensity peaks; (iii) the flower-shaped outline of the OTF should not show bright lining; (iv) the individual ‘leaves’ of the image FT should not have discernable bright or dark divisions, neither radially nor angularly; and, (v), information should be contained in an area beyond the standard wide-field FT.
To mimic sample movement with respect to the illumination pattern, we piezo-translated green-fluorescent 93-nm beads at various speeds v while taking super-resolved images at 3.3 Hz. The drift correction function of the SIM reconstruction software [23] was disabled for this experiment. SIM reconstructions were possible up to v = 230 nm/s, Fig. 3(A), (B).
Fig. 3 Resistance to sample movement. Sequence of TIRF (A) and TIRF-SIM (B) images of green-fluorescent beads (505/515nm) moving at 230 nm/s. The whole sample was translated on a piezo-stage and 20-ms images were acquired at 3.3 Hz. Red dashed line serves as a reference for the eye. Note the sample displacement to the right (red arrow). Contrast inverted for better clarity in print. Scale bar: 1 µm.
3.3 Imaging ER/mitochondrial connectivity
The endoplasmic reticulum (ER) and mitochondria are two subcellular organelles present in most eukaryotic cells. Their interplay is a key determinant of cell function and survival via the control of intracellular Ca2+ signaling [26]. Specific sites of physical association between the ER and mitochondria are known as mitochondria-associated membranes and have been postulated as sites where direct Ca2+ exchange occurs, with important implications for energy production, cell fate and apoptosis. A detailed study of interactions between these two heavily intertwined tubular networks has been difficult because the small size and proximity of the organelles results in blur that do not resolve their fine structure on conventional TIRF images.
TIRF-SIM images of mouse cortical astrocytes expressing mito-EGFP showed the typical mitochondrial chains as well as isolated mitochondria, Fig. 4(A). The entire cell could be visualized, allowing the study of rare or sparse events. Time-lapse imaging at 3.3Hz allowed us to track the fission/fusion events and the extension of individual mitochondria, Fig. 4(B). As most mitochondria were fairly static on these images and only a few organelles moved and/or changed their morphology, phase information could be obtained and the SIM reconstruction worked for subcellular dynamics occurring on a time-scale much faster than the speed limit found earlier in the case of the translation of the whole sample.
Fig. 4 TIRF-SIM imaging of near-membrane mitochondrial dynamics. (A), whole-filed image of an astrocyte expressing mito-EGFP, integration time 20 ms per image (306 ms per to obtain the super-resolved full-field image, s = 180 nm). (B), 3.3-Hz time-lapse sequences showing mitochondrial dynamics in the zoom region shown in (A). Scale bar: 1 µm. (C), Distance travelled vs. time of the leading edge of the mitochondrium identified with a red arrow head on panel (B). The speed of tip displacement obtained from a linear fit was υ = 1.24 ± 0.06 µm/s.
For example, the migrating tip of the near-membrane mitochondrion identified in Fig. 4(B) by a red arrowhead advanced with ~1.24 µm/s, indicating that TIRF-SIM is suitable for imaging a broad range of dynamic phenomena occurring in live cells at or near the plasma membrane.
Time-lapse imaging was readily combined with obtaining super-resolution colocaliza-tion information when exciting with a second wavelength. Dual-color TIRF-SIM images of astrocytes expressing (λex,max/ λem,max = 488/510 nm) and labeled with Mito-Tracker Deep Red (644/665 nm) were acquired by toggling between the 488- and 561-nm laser lines. Laser positions in the objective BFP for each wavelength were not changed, meaning that there was a slight (15%) mismatch in pattern period and EW penetration depths. This lead, along with the variation in λem, to a lateral-resolution difference of 19 nm which we considered negligible. Figures 5(A) and 5(B) illustrate the striking improvement in resolution and contrast resulting from SIM compared to TIRF alone. The intricate connectivity between mito-chondrial (red) and ER (green) networks is now revealed and particularly evident when zooming in, Fig. 5(C). The fine ramifications of the ER having widths of 99, 95 and 118 nm (FWHM) wound around and constricted nearby mitochondria. They identify potential interaction sites between both organelles (arrowheads). Dimensions measured on TIRF-SIM images were much smaller than those seen on the conventional TIRF images, which were 247, 227 and 409 nm, respectively. Intensity profiles allowed a detailed study of ER/mitochondrial co-localization, Fig. 5(D). All these images were taken in live cells at frame rates and laser powers allowing long-period observation.
Fig. 5 Dual-color superresolution imaging. TIRF (A) and TIRF-SIM (B) images of an astrocyte expressing ER-EGFP, a marker of the endoplasmic reticulum, and labeled with Mito-Tracker-Deep-Red, a marker of mitochondria. Images were taken sequentially upon 488- and 561-nm excitation, respectively, with no emission filter. Exposure time was 200 ms per frame (total time: 3.8s for the acquisition of the two colors). Pattern period was 180 nm (206 nm) for the green (red) channel, respectively. (C) Zoom on the region of interest shown in (A), (B). Scale bar: 1 µm. (D) Intensity profiles, along the line region of interest shown in (A) and (B), for the ER (green) and mitochondria (red) reveal the increase in image contrast and resolution.
4. Discussion
4.1 Nanoscopy of astrocyte dynamics
The potential of super-resolution microscopies for studying astrocyte signaling and morpho-logical dynamics has been pointed out [27], but there are only a few published reports. Using near-field scanning probe microscopy Haydon et al. resolved actin filaments in chemically fixed, hydrated glia [28]. More recently, Verkmann and associates imaged supramolecular arrays of aquaporin-4 in astrocyte membranes using PALM/STORM [29]. Our study is, to the best of our knowledge, the first super-resolution study of live astrocyte dynamics.
4.2 TIRF-SIM lends itself to routine near-membrane imaging
Our approach to fast, full-field multi-color super-resolution imaging has, above all, the virtue of simplicity. The set-up is compact, made of few optical elements and very stable. Our image processing algorithm uses a correlation-based a posteriori determination of the pattern position from the acquired data [23]. It allowed us to extract the high-frequency information and tolerates uncontrolled phase shifts [30] without the need for the measurement of the fringe period or complicated active feedback control [15, 16, 18]. This methodology is insensitive to pattern-phase fluctuations induced by thermal drift, sample movement or pattern photobleaching. In some experiments, we noted that the recovered fringe periods varied among pattern orientations. We interpret this finding as a result of coverslip tilt, however minimal, that changes the beam angle of the reflected beams and could be abolished by mounting the sample on a tip-tilt platform [25]. Again, the algorithm takes into account such orientation-dependent variability. Our system is thus suitable for non-expert users and should be of particular interest for shared facilities, imaging core platforms and training students.
Unlike other super-resolution microscopies, SIM is flexible. It does not require specific fluorophores, it allows multi-color excitation and emission [31, 32], and, particularly when combined with EW excitation, exposes the cells to a fairly low photo bleaching and photo damage (9 times that of conventional TIRFM), which is still compatible with long-period observation. As for every technique, TIRF-SIM has its drawbacks. The resolution gain is ~3 times less than with PALM/STORM or STED. Furthermore, with a sinusoidal grating, SIM requires pattern homogeneity across the field-of-view and its stability during camera integration time. For this reason, to let the piezo-actuated mirror positions stabilize, we allowed some milliseconds after each pattern shift before acquiring the next image. SIM using random speckles have been reported [33] but typically requires acquisition of many images.
TIRF-SIM has been mostly used to study fixed samples or the slow reorganization of cytoskeletal elements because mechanically rotating the physical grating which is projected into the sample plane (or laterally moving a star-shaped grating the case of the Nikon N-SIM) is too slow for imaging fast organelle dynamics. The use, either of a dielectric elastomer transmission grating [14] or synthetic gratings produced with a spatial light modulator (SLM) [18, 34, 35], to dynamically control the illumination pattern, offers flexibility with respect to the wavelength and resolution. However, refresh rates of SLMs are slow, leading to dead times, particularly when different wavelengths are interlaced. Crossed AODs [12, 17, 35] allow fast switching and phase shifts but as diffractive optical elements they operate over a limited wavelength range, and finding the fluorophores having well-separated excitation spectra as required for multi-color imaging without appreciable crosstalk is not always straightforward. Also, non-homogeneities of the sound wave across the AOD crystals result in ruffles of the illumination stripes that can produce artifacts in the reconstructed TIRF-SIM images. In this context, our set-up goes back to a simple interferometric approach used in a proof-of-principle study for 1-D TIRF-SIM [12]. Our time-resolution (3.3 Hz) is limited by the available fluorescence and is limited (at a given laser power) by the exposure time (9 times 20 ms), and, to a lesser degree, by the settle time between two positions of the tip-tilt mirror (we allowed a conservative 10-ms lapse). This still allows temporal resolutions much faster than those obtained with commercial instruments (frame rates for the Zeiss Elyra SIM are ~0.59 Hz and ~0.05 Hz for the Leica STED for typical biological sample). The 3.3 Hz per color give access to a broad range of biological phenomena, including ER and mitochondrial dynamics, filopodia extension and lipid-raft diffusion, to name only a few. Even higher frame rates could be obtained when using our SIM in a non-fluorescent scattering excitation darkfield detection mode [36]. Importantly, this time-resolution is achieved over a large field of view, so that rare or sparse events as spontaneous activity or phenomena only loosely coupled to external stimuli can be imaged.
Due to the sinusoidal illumination pattern used here, SIM-image reconstruction limited relative fluorophore and pattern movement to ~230 nm/s. Without much surprise, this speed corresponds to one pixel movement during the acquisition of the nine component images (65 nm/306 ms = 212 nm/s). It turns out, however, that this is not a very severe constraint, for multiple reasons; (i), our reconstruction algorithm retrieves the phase and pattern period from each image individually and hence can correct for whole-image movement. The microscope is hence fairly insensitive to thermal drift or mechanical strain; (ii), the SIM pattern is retrieved from the full field, but organelle dynamics often only occurs in a tiny sub-region of the image as seen, e.g., on Fig. 4(D). Much faster organelle movement can thus be tolerated; (iii), the worst-case scenario for SIM reconstruction would be Brownian movement of fluorescently tagged molecules or small organelles occurring in random directions. Thus, while imaging the rapid movement of single vesicles might be beyond the reach of our TIRF-SIM, many biological processes including the diffusion of whole organelles, lipid rafts, protein complexes and other macromolecules in the plane of the membrane will fall within the available speed range. Shorter exposure times, pixel binning and the use of an image splitter allowing simultaneous dual-color excitation can speed up acquisitions for studying faster dynamics.
Finally, although we demonstrated here TIRF-SIM in the plane of the basal plasma membrane only, 3D-SIM could be achieved by adjusting the beam angle q to subcritical angles (in a variant dubbed ‘HILO’ for highly inclined laminar optical sheet imaging) and adding a third laser on-axis beam for z-modulation. In that case 15 images need to be taken for the image reconstruction, which ceterum paribus will lead to a temporal resolution of ~2.2 Hz (again at 20 ms of integration time).
4.3 Optimizing polarization
Fringe contrast depends on polarization. For a fixed polarization vector, contrast changes with pattern orientation (Fig. 6, leftmost panel) unless the polarization is co-rotated. To assure optimal contrast at all orientations, polarization must stay perpendicular to the meridional plane of the objectives lens (s-polarization), (Fig. 6, middle left).
Fig. 6 Different options for optimizing fringe pattern contrast. Left, unrotated “vertical polarization” provides high contrast vertical fringes but low contrast in other directions. Middle left, ideally, polarization would rotate with the pattern to assure tangential s-polarization to give optimal contrast. Middle right, circular polarization produces constant but low fringe contrast. We used instead an unrotated “horizontal polarization” (right) corresponding to a local maximum in fringe contrast. The precise polarization angle depends on the given objective NA, beam angle and pattern directions. See ref [37]. for details.
Rotating the polarization either compromises temporal resolution (for a linear polarizer in a motorized rotational mount) or else requires the use of additional electro-optical elements that considerably increase the cost and complexity of the set up [18]. Alternatively, one could glue together pieces of polarizing glass in the correct orientations to produce a ‘cake’ polarizer. In our case the triangular shape of the individual facets of such a device would limit the beam diameter to a degree that would degrade the quality of the focus in the objective BFP and hence increase the beam divergence degrade the EW pattern; so we did not implement this solution. Circular polarization is another option, with the drawback of lowering fringe contrast for all pattern orientations (Fig. 6, middle right). We therefore instead used an idea proposed by O’Holleran et al. [37]. For a given objective NA, pattern period and directions there is a unique solution that maximizes contrast. In our setup, the polarization state which optimizes pattern contrast at s = 175 nm is parallel to the meridional plane of the vertical pattern and termed here “horizontal polarization”, (Fig. 6, right).
5. Conclusion
We present a TIRF-SIM which fits on a ~50-cm square breadboard and combines simplicity with greater 〉exibility and higher speed than found in most previous systems. By design achromatic it enables fast fine-tuning of the pattern orientation and fringe period over a broad range of angles. This allows one to conveniently adapt the EW penetration depth and lateral resolution between two image acquisitions without much loss of temporal resolution. This is a requirement for the interlaced acquisition of multi-color or time-lapse data, as illustrated by our dual-color imaging of ER and mitochondrial organization. Even higher spatial resolution could be obtained, in the linear regime, with 1.57 or 1.65-NA objectives, shorter excitation/emission wavelengths or nanostructured grating substrates [38]. The exploitation of excitation non-linarites, such as two-photon TIRF [39], the saturation of the fluorophore excitation rate [32, 40], see, however [41], or the transition rate of a photo-switchable fluorescent protein [11, 32] could further enhance resolution.
Author contributions
M. B. built the apparatus and performed all experiments, M. B., R. H., M. O. and K. W. analyzed the data, M. B., R. H. and K. W. wrote software, K. H. performed cell culture and molecular biology work, M. B. and M. O. wrote the manuscript, M. O. supervised and directed research.
Acknowledgements
We thank M. van ’t Hoff for initial contributions, P. Jegouzo for fine mechanics, D. Li for help with transfections and live-cell imaging, V. Emiliani (CNRS UMR8154) and L. Moog (Coherent) for the loan of equipment, J. Hirrlinger (Leipzig) for plasmids, A. Jost (Jena) for help with image reconstructions, and J. Estaquier, M. Laforge, N. Ropert (Paris) and R. Uhl (Munich) for discussions. JS. Kehoe is acknowledged for careful proofreading. Financed by the French Agence National de la Recherche (ANR P3N nanoFRET2), the European Union (FP7 ERA-NET nanoSYN), and the National Research Infrastructure FranceBioImaging (FBI). The Oheim lab is part of the École des Neurosciences de Paris Ile-de-France (ENP) network and the C’nano Ile-de-France nanoscience and nanotechnology excellence cluster.
References and links
1. R. Heintzmann and M. G. L. Gustafsson, “Subdiffraction resolution in continuous samples,” Nat. Photonics 3(7), 362–364 (2009). [CrossRef]
2. B. Huang, “Super-resolution optical microscopy: multiple choices,” Curr. Opin. Chem. Biol. 14(1), 10–14 (2010). [CrossRef] [PubMed]
3. B. O. Leung and K. C. Chou, “Review of super-resolution fluorescence microscopy for biology,” Appl. Spectrosc. 65(9), 967–980 (2011). [CrossRef] [PubMed]
4. 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]
5. 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]
6. M. J. Rust, M. Bates, and X. Zhuang, “Sub-diffraction-limit imaging by stochastic optical reconstruction microscopy (STORM),” Nat. Methods 3(10), 793–796 (2006). [CrossRef] [PubMed]
7. W. Lukosz, “Optical systems with resolving powers exceeding the classical limit,” J. Opt. Soc. Am. 56(11), 1463–1471 (1966). [CrossRef]
8. R. Heintzmann and C. Cremer, “Laterally modulated excitation microscopy: improvement of resolution by using a diffraction grating,” Proc. SPIE 3568, 185–196 (1999). [CrossRef]
9. M. G. L. Gustafsson, “Surpassing the lateral resolution limit by a factor of two using structured illumination microscopy,” J. Microsc. 198(2), 82–87 (2000). [CrossRef] [PubMed]
10. J. T. Frohn, “Super-resolution fluorescence microscopy by structured light illumination,” (Ph.D. thesis, Eidgenössische Technische Hochschule (ETH), Zürich, 2000).
11. L. M. Hirvonen, K. Wicker, O. Mandula, and R. Heintzmann, “Structured illumination microscopy of a living cell,” Eur. Biophys. J. 38(6), 807–812 (2009). [CrossRef] [PubMed]
12. O. Gliko, G. D. Reddy, B. Anvari, W. E. Brownell, and P. Saggau, “Standing wave total internal reflection fluorescence microscopy to measure the size of nanostructures in living cells,” J. Biomed. Opt. 11(6), 064013 (2006). [CrossRef] [PubMed]
13. E. Chung, D. Kim, and P. T. C. So, “Extended resolution wide-field optical imaging: objective-launched standing-wave total internal reflection fluorescence microscopy,” Opt. Lett. 31(7), 945–947 (2006). [CrossRef] [PubMed]
14. M. Beck, M. Aschwanden, and A. Stemmer, “Sub-100-nanometre resolution in total internal reflection fluorescence microscopy,” J. Microsc. 232(1), 99–105 (2008). [CrossRef] [PubMed]
15. E. Chung, D. Kim, Y. Cui, Y. H. Kim, and P. T. C. So, “Two-dimensional standing wave total internal reflection fluorescence microscopy: superresolution imaging of single molecular and biological specimens,” Biophys. J. 93(5), 1747–1757 (2007). [CrossRef] [PubMed]
16. R. Fiolka, M. Beck, and A. Stemmer, “Structured illumination in total internal reflection fluorescence microscopy using a spatial light modulator,” Opt. Lett. 33(14), 1629–1631 (2008). [CrossRef] [PubMed]
17. O. Gliko, W. E. Brownell, and P. Saggau, “Fast two-dimensional standing-wave total-internal-reflection fluorescence microscopy using acousto-optic deflectors,” Opt. Lett. 34(6), 836–838 (2009). [CrossRef] [PubMed]
18. 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]
19. Z. Huang and N. L. Thompson, “Theory for two-photon excitation in pattern photobleaching with evanescent illumination,” Biophys. Chem. 47(3), 241–249 (1993). [CrossRef] [PubMed]
20. G. E. Cragg and P. T. C. So, “Lateral resolution enhancement with standing evanescent waves,” Opt. Lett. 25(1), 46–48 (2000). [CrossRef] [PubMed]
21. A. L. Stout and D. Axelrod, “Evanescent field excitation of fluorescence by epi-illumination microscopy,” Appl. Opt. 28(24), 5237–5242 (1989). [CrossRef] [PubMed]
22. F. Orieux, E. Sepulveda, V. Loriette, B. Dubertret, and J. C. Olivo-Marin, “Bayesian estimation for optimized structured illumination microscopy,” IEEE Trans. Image Process. 21(2), 601–614 (2012). [CrossRef] [PubMed]
23. K. Wicker, O. Mandula, G. Best, R. Fiolka, and R. Heintzmann, “Phase optimisation for structured illumination microscopy,” Opt. Express 21(2), 2032–2049 (2013). [CrossRef] [PubMed]
24. D. Li, K. Hérault, M. Oheim, and N. Ropert, “FM dyes enter via a store-operated calcium channel and modify calcium signaling of cultured astrocytes,” Proc. Natl. Acad. Sci. U.S.A. 106(51), 21960–21965 (2009). [CrossRef] [PubMed]
25. M. Brunstein, M. Teremetz, C. Tourain, and M. Oheim, “Combined evanescent-wave excitation and supercritical-angle fluorescence detection improves optical sectioning,” (arXiv.org>physics> http://arxiv.org/abs/1302.1615, 2013).
26. R. Rizzuto, M. R. Duchen, and T. Pozzan, “Flirting in little space: the ER/mitochondria Ca2+ liaison,” Sci. STKE 2004(215), re1 (2004). [PubMed]
27. D. A. Rusakov, K. Zheng, and C. Henneberger, “Astrocytes as regulators of synaptic function: A quest for the Ca2+ master key,” Neuroscientist 17(5), 513–523 (2011). [CrossRef] [PubMed]
28. R. T. Doyle, M. J. Szulzcewski, and P. G. Haydon, “Extraction of near-field fluorescence from composite signals to provide high resolution images of glial cells,” Biophys. J. 80(5), 2477–2482 (2001). [CrossRef] [PubMed]
29. A. Rossi, T. J. Moritz, J. Ratelade, and A. S. Verkman, “Super-resolution imaging of aquaporin-4 orthogonal arrays of particles in cell membranes,” J. Cell Sci. 125(18), 4405–4412 (2012). [CrossRef] [PubMed]
30. S. Takahashi, S. Okada, H. Nishioka, S. Usuki, and K. Takamasu, “Theoretical and numerical analysis of lateral resolution improvement characteristics for fluorescence microscopy using standing evanescent light with image retrieval,” Meas. Sci. Technol. 19(8), 084006 (2008). [CrossRef]
31. G. Best, R. Amberger, D. Baddeley, T. Ach, S. Dithmar, R. Heintzmann, and C. Cremer, “Structured illumination microscopy of autofluorescent aggregations in human tissue,” Micron 42(4), 330–335 (2011). [CrossRef] [PubMed]
32. 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]
33. E. Mudry, K. Belkebir, J. Girard, E. Le Moal, C. Nicoletti, M. Allain, and A. Sentenac, “Structured illumination microscopy using unknown speckle patterns,” Nat. Photonics 6(5), 312–315 (2012). [CrossRef]
34. F. Spira, N. S. Mueller, G. Beck, P. von Olshausen, J. Beig, and R. Wedlich-Söldner, “Patchwork organization of the yeast plasma membrane into numerous coexisting domains,” Nat. Cell Biol. 14(6), 640–648 (2012). [CrossRef] [PubMed]
35. M. van ’t Hoff, V. de Sars, and M. Oheim, “A programmable light engine for quantitative single molecule TIRF and HILO imaging,” Opt. Express 16(22), 18495–18504 (2008). [CrossRef] [PubMed]
36. S. Chowdhury, A.-H. Dhalla, and J. Izatt, “Structured oblique illumination microscopy for enhanced resolution imaging of non-fluorescent, coherently scattering samples,” Biomed. Opt. Express 3(8), 1841–1854 (2012). [CrossRef] [PubMed]
37. K. O’Holleran and M. Shaw, “Polarization effects on contrast in structured illumination microscopy,” Opt. Lett. 37(22), 4603–4605 (2012). [CrossRef] [PubMed]
38. A. Sentenac, K. Belkebir, H. Giovannini, and P. C. Chaumet, “Subdiffraction resolution in total internal reflection fluorescence microscopy with a grating substrate,” Opt. Lett. 33(3), 255–257 (2008). [CrossRef] [PubMed]
39. F. Schapper, J. T. Gonçalves, and M. Oheim, “Fluorescence imaging with two-photon evanescent wave excitation,” Eur. Biophys. J. 32(7), 635–643 (2003). [CrossRef] [PubMed]
40. 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]
41. A. Gur, Z. Zalevsky, V. Micó, J. García, and D. Fixler, “The limitations of nonlinear fluorescence effect in super resolution saturated structured illumination microscopy system,” J. Fluoresc. 21(3), 1075–1082 (2011). [CrossRef] [PubMed]
