Super-resolution Microscopy and the Third Dimension

November 04, 2013

Optical imaging devices have a finite depth of field and diffraction-limited resolution. The depth of field problem was tackled first with confocal microscopes; diffraction-unlimited resolution has been available for a few years now with super-resolution microscopes. Super-resolution microscopes with a solved depth of field problem are now available. Localization super-resolution uses astigmatism-based z-coding. STED super-resolution employs a mixed phase mask concept for tunable super-resolution in lateral and axial dimensions.

Optical sectioning

Visual perception is often regarded as the most impressive of our senses. And many attempts have been made to preserve optical impressions – or enhance them to augment the fun. Optical imaging devices have one annoying disadvantage, though: the range of sharp imaging is limited. This limited range is called "depth of field". Although even our eyes feature the same limitation, our brain has means to cleverly filter only sharp parts of the whole image into our consciousness. We realize that fraud at the latest if we look at a photograph and wonder why there are unsharp regions.

In optical microscopes, the depth of field is comparably shallow, especially if we aim to record images at high resolution. And there have always been attempts to increase the depth of field – or get rid of the unsharp contributions.

The problem was solved in the past by sectioning the sample into rather thin slices – thinner than the depth of field. These samples do not show any out-of-focus contributions. But they also lack the structural integrity of the objects one is trying to understand. To reconstruct a three-dimensional view of the object, e.g. a cell or a small organism, one had to cut a large series of thin sections and then combine the recorded images. This procedure is obviously quite cumbersome, and cutting often causes severe deformations that thwart correct reconstruction.

A better method is what is called “confocal microscopy” (for review see [1]). A confocal microscope removes any contributions that do not belong to the depth of field by an optical arrangement whose most prominent part is the "detection pinhole". In order to generate a three-dimensional reconstruction, it is sufficient to increment the focus position by about half the depth of focus and merge the recorded images by appropriate software. In the last 20 years, this method has become a standard procedure in biomedical imaging [2].

A second approach that serves the same purpose is multiphoton microscopy (for a review see [3]). Here,  optical sectioning occurs by a non-linear effect which excites fluorescent molecules only in a thin layer.

Fig. 1: Beam path of a confocal scanning microscope. Although light is radiating from all layers in z direction, only one single focal plane is transferred to the detector. Light from above and from below the focal plane is efficiently removed by the pinhole, which acts as a spatial filter. OL, objective lens; BS, beam splitter; PH, pinhole.
Beam path of a true confocal scanning microscope (TCS). The detection pinhole acts as a spatial filter and  only lets light from the focal plane pass through. Realization e.g. in Leica TCS SP8 systems.

Better z-resolution: 4PI

A natural limitation of a conventional microscope is the fact that light from the sample is collected  from one side only. From a single spot, only a half-sphere can contribute to the image reconstruction. This half-sphere corresponds to a solid angle of 2p. If it were possible to operate two opposing lenses on both sides of the specimen, the theoretically coverable solid angle should be 4p, hence the 4PI microscope [4]. In practice, the maximum one can cover with a single lens is ca. 1.3 p, and the 4PI microscope is in fact a 2.5 p microscope. In a 4PI microscope, the sample has to be sandwiched between two coverslips, and it must be thin enough to leave space for focusing. Therefore, a sample thicker than 50 µm is usually unsuitable for 4PI microscopy. Coherent illumination from both sides creates a confined intensity layer that can allow a resolution of some 100 nm in the axial direction. This is 5–10 times better as compared to the axial resolution in a classical confocal microscope. Although the term 4PI microscopy suggests a doubling of the resolution, the lateral performance is not significantly better than in a confocal microscope. Such instruments were the first commercially available microscopes that improved imaging with focused light (Leica TCS 4PI [5]).

Unlimited resolution: STED

Optical approaches alone obviously did not lead to remarkable improvements in resolution. "Remarkable" indicates something that enhances not just by a few percent, but by a factor of at least twofold or better. The first concept was proposed by Stefan Hell in 1994 [6]. This concept includes non-optical physical phenomena, which is the crucial difference as compared to earlier attempts. Here, the phenomenon employed was "stimulated emission", first theoretically described by Einstein [7] and applied practically in laser light sources. The STED concept utilizes stimulated emission to partially deplete excited fluorescence states before emission occurs. This corresponds to switching off excited states in a region of the excitation point spread function. If the depletion can be managed to occur only at the border of a diffraction-limited excitation spot, the residual emitting area will be smaller and hence the resolution is improved. Fortunately, there is a comparably simple means to illuminate only the borders of a spot-diffraction pattern: a donut-shaped focus. Such a focus is generated by inserting an appropriate phase plate into the illuminating beam path. The illumination is then two-fold: first by an appropriate wavelength for fluorescence excitation, with an ordinary circular "Airy focus", then, with an appropriate wavelength by a donut-shaped "depletion focus". Emission is then collected like in any other scanning optical microscope. STED microscopes are thus "easily" derived from confocal laser scanning microscopes. The first commercially available system specified resolution down to 80 nm (Leica TCS STED) in 2007 [8].

A different approach: GSD

Once STED super-resolution had proven its capacity to permit images that feature much better resolution than pure optical (diffraction-limited) imaging, the spell had been broken, and it was safe to discuss resolution beyond the diffraction limit. The second prominent approach was of stochastical nature. Under the constraint that the fluorescence generators are sufficiently diluted to ensure that an observed spot originates only from a single emitter, one can assume that the spot represents the point spread function. It is then a simple task to find the center of that point spread function, which allows the emitter to be localized with very high precision. In order to reconstruct a full image, it is necessary to record a large sequence of images, each containing just a few separated emitters – leading to a coherent image of the structure under research. Here again, a switcher is required to ensure sufficiently sparse emitters per image and also recover the silent emitters for the subsequent images that are necessary to cover the full structure. Early proposals used switchable fluorescent proteins to do that task [9]. However, nearly all fluorochromes also exert switching behavior from excitable states to dark states and back. The best known dark states are triplet states, which are assumed by the fluorochrome stochastically. When illuminated with appropriate intensity, the majority of the fluorescent molecules can be kept in a dark state, and only occasionally a molecule resides for a short while in the excitable state. During this time, the psf is recorded and is ready for subsequent super-resolved localization. This method is called ground state depletion microscopy [10, 11]. A commercial system based on this concept was introduced in 2011, the Leica SR GSD [12].

Localizing the third dimension

Localization microscopy offers improved lateral resolution. Theoretically, one could also localize the intensity maximum of all spots along the z axis, but this would require a prohibitively long acquisition time. In addition, localization microscopy is typically based on the TIRF technique, which is not made for focusing into the sample. Nevertheless, the range covered by TIRF is some 300 nm, and it would be of interest to be able to measure the position of an emitter within this range. To accomplish this task, an optical aberration was turned to advantage: astigmatism. Lenses tend to generate a rod-like structure of a spot source when not in focus. Even if the spot is correctly reconstructed in the focal plane, one can detect a pencil-shaped structure above and below the focal plane. Interestingly, these pencils rotate with the z-axis. In z-resolving localization microscopes, astigmatism is introduced intentionally by insertion of a cylindrical lens [13]. The angle of rotation is then measured for each localized spot, and finally gives an accuracy of some 50 nm in the z axis. This method is applicable to any type of localization and consequently also combined with the above-mentioned ground state depletion technique. A commercial version of such an instrument was introduced recently (Leica SR GSD 3D [14].

Unlimited resolution in all dimensions: STED 3D

The final goal is unlimited resolution both laterally and axially without restrictions in field size and free working distance. The STED approach offers free focusing into the sample – at least theoretically. The obstacle of limited axial resolution can be overcome with special phase masks that deplete excitation at the anterior and posterior ends of the point spread function. Unfortunately, the design of such phase masks does not allow for a zero in the lateral direction, simultaneously. To arrive at both lateral and axial isomorphic point spread functions (which in essence are spheres), the STED technique was combined with the 4PI technique [15]. Although this approach worked well,  cumbersome 4PI microscopy in combination with just another non-trivial technology is not ready to serve as an everyday tool in biomedical laboratories. Even well-educated and trained imaging facilities would question the ultimate benefit of such a device. A much simpler solution is to combine the lateral and axial modes in a 3D STED microscope. Here, the depletion beam is first split by a tunable splitting device. One leg is equipped with a vortex phase mask (that is the best design for improving lateral resolution). The other leg is equipped with a phase mask that is ideal for improving the axial resolution, called “z-donut”. Recombination of both legs creates a point spread function whose shape is close to a hollow sphere (imagine a jelly donut). This hollow sphere will remove excited states around the center of the excitation point spread function in all directions. As the ratio of lateral and axial depletion is tunable, the settings can be optimized to achieve either isomorphic resolution, tiniest detection volume, or maximum lateral versus axial resolution. The technique is now commercially available (Leica TCS SP8 STED 3X [16]).The title picture shows a comparison of confocal microscopy (left) and 3D STED (right). Histone was stained in nuclei of HeLa cells. The dotted line in the horizontal section (upper part) indicates the position of a profile section, which is displayed in the lower part. Note the significantly improved resolution in both lateral and axial direction with 3D STED.

Fig. 2: Beam path of the STED 3X illumination module. The depletion laser energy is split into two fractions, the ratio can be adjusted by a tunable splitter (VS). One arm contains a phase mask (φxy) that is designed for diffraction-unlimited lateral resolution. The other arm is equipped with a different phase mask (φz) that allows super-resolution in axial direction. Both illumination patterns are combined in a merger (M). The resulting depletion beam is tunable for resolution improvement in lateral and axial direction.
Beam path of the Leica TCS SP8 STED 3X illumination module. The depletion laser energy is split into two fractions, the ratio can be adjusted by a tunable splitter. The resulting depletion beam is tunable for resolution improvement in lateral and axial di

References

  1. Sheppard CJR and Shotton DM: Confocal Laser Scanning Microscopy. Bios. Scientific Publ. (1997); ISBN: 978-1872748726.
  2. Borlinghaus RT: Microscopy and the Third Dimension. Zeiss Information with Jena Review 5 (1995).
  3. Diaspro A (ed.): Confocal and Two-Photon Microscopy: Foundations, Applications and Advances. Wiley (2001); ISBN: 978-0-471-40920-5.
  4. Hell SW and Stelzer EKH: Properties of a 4Pi confocal fluorescence microscope. In: Journal of the Optical Society of America – A: Optics, Image Science, and Vision 9 (12): 2159–66 (1992); doi: 10.1364/JOSAA.9.002159.
  5. Cremer C: www.kip.uni-heidelberg.de/AG_Cremer/en/content/4pi-mikroskopie – retrieved 20.09.2013.
  6. Hell SW and Wichmann J: Breaking the diffraction resolution limit by stimulated emission: stimulated-emission-depletion fluorescence microscopy. Opt. Lett. 19 (11): 780–82 (1994).
  7. Einstein A: Strahlungsemission und -absorption nach der Quantentheorie. Verhandlungen der Deutschen Physikalischen Gesellschaft 18: 318–23 (1916). Bibcode: 1916DPhyG..18..318E.
  8. Press Release: Breaking the Diffraction Barrier: Light Microscopy Goes Beyond the Limits with the New Superresolution Leica TCS STED. Leica Microsystems (2007).
  9. Betzig E et al: Imaging Intracellular Fluorescent Proteins at Nanometer Resolution. Science 313: 1642 (2006); doi: 10.1126/science.1127344.
  10. Hell SW and Kroug M: Ground-state-depletion fluorescence microscopy: a concept for breaking the diffraction resolution limit. Appl. Phys. B 60: 495–97 (1995).
  11. Foelling J et al: Fluorescence nanoscopy by ground-state depletion and single-molecule return. nature methods 5 (11): 943–45; doi: 10.1038/NMETH.1257 (2008).
  12. Press release: New Widefield Super-Resolution System for Applications in Biomedical Research. Leica Microsystems (2011).
  13. Huang B et al: Three-Dimensional Super-Resolution Imaging by Stochastic Optical Reconstruction Microscopy. Science 319 (5864): 810–13 (2008); doi: 10.1126/science.1153529.
  14. Press release: New Super-Resolution System for 3D Localization Microscopy. Leica Microsystems (2013).
  15. Schmidt R et al: Spherical nanosized focal spot unravels the interior of cells. Nature Meth. 5: 539–44 (2008).
  16. Press release: Leica TCS SP8 STED 3X – fast and direct 3D super-resolution". Leica Microsystems (2013).

Comments

Help us to improve the Leica Science Lab and participate in our survey! Start Survey

Arrow Down

Arrow Up