Background

Fluorescence microscopy has for many years been one of the most important tools for understanding how biological systems are organized and how they function. Over the last decade, fluorescence microscopy has been revolutionized by the development of “super-resolution” methods that extend the limit of optical microscopy beyond the diffraction limit, providing unprecedented levels of information on the organization of molecular networks in cells. These methods include Structured Illumination Microscopy (SIM) (Gustafsson, 2000), Stimulated Emission Depletion Microscopy (STED) (Hell and Wichmann, 1994), and techniques based on the localization of individual fluorescent molecules with much better precision than the diffraction limit. The latter includes Photo-activated Localization Microscopy (PALM) (Betzig et al., 2006) and Stochastic Optical Reconstruction Microscopy (STORM) (Rust et al., 2006), referred to collectively in this protocol as Single Molecule Localization Microscopy (SMLM) techniques. Because of a requirement to stop motion within the cell when aiming for high resolution, super-resolution methods usually involve the use of chemical fixation, which has the potential to introduce artefacts that can be visible at the resolutions achieved (Whelan and Bell, 2015). This problem has long been recognized in the field of electron microscopy, where a different method is now routinely used to fix cells and preserve their structure: vitrification at cryogenic temperatures. In this method, samples are frozen using a cryogen, cooling rapidly to temperatures of -150 °C or below, resulting in the formation of amorphous ice, with the high rate of cooling preventing the formation of ice crystals (Dubochet and McDowall, 1981). SMLM techniques in particular should benefit from the use of cryogenic conditions. This is because the localization precision of fluorescent molecules is dependent on the number of photons emitted by the fluorophores (Thompson et al., 2002), and at cryogenic temperatures significantly more photons can be collected. Conventional widefield fluorescence microscopy at cryogenic temperatures is already used in Correlative Light and Electron Microscopy (CLEM) (de Boer et al., 2015) workflows, but the mismatch in resolution between optical microscopy and EM limits the value of the fluorescence data, which is usually restricted to locating general areas of interest. Hence, enabling cryogenic super-resolution fluorescence microscopy would be a significant advance for cryogenic CLEM workflows.

Given the advantages of vitrification over chemical fixation, it would be advantageous to apply it to super-resolution optical microscopy methods. However, its use has been severely limited for one major reason: super-resolution methods require the use of high numerical aperture (NA) objective lenses (NA > 1), and achieving this NA typically requires the use of oil immersion objectives, incompatible with cryogenic microscopy. A number of approaches have been taken to avoid this problem, including using non-immersion objectives with extremely bright fluorophores (Liu et al., 2015) and using special cryogenic immersion fluids and custom sample stages (Nahmani et al., 2017). However, these approaches have not been generally applied to the imaging of biological samples.

We have developed a low-cost, generally applicable method for super-resolution cryo-fluorescence microscopy that uses commercially available optical components and cryo-stage (Wang et al., 2019). For this, we use super-hemispherical solid-immersion lenses (superSILs). These lenses are made of high refractive index materials, and take the form of truncated spheres. In our application, their role is to take the place of immersion oil, and fill the gap between the sample and a conventional non-immersion objective. The combination of superSIL and air objective gives a high numerical aperture, dependent on the refractive index of the lens material (Terris et al., 1994; Zhang et al., 2007; Chen et al., 2013).

We have characterized the performance of superSILs for cryo-SMLM, showing that we are able to achieve a point spread function (PSF) size of around 150 nm, compared with around 330 nm for a cryogenic microscope with a conventional low NA air objective. This is because the combination of superSIL and objective achieves a NA of 2.17, much higher than the maximum NA of ~1.4 of oil-immersion objectives. The superSIL microscope achieves single molecule localization precisions of < 10 nm, and a lateral resolution of 12 nm (Wang et al., 2019). A major advantage of the superSIL method is that it can be achieved using readily available off-the-shelf components, with the possibility of easily adding the cryostage to microscopes commonly found in cell biology research laboratories. Moreover, superSIL assemblies can be cleaned and reused multiple times, making them an inexpensive resource. In this protocol, we focus on use of the superSIL system for super-resolution microscopy of bacteria, but also indicate how it might be applied to imaging of mammalian cells. A similar protocol could also be used to image other samples, such as purified protein complexes or vesicles. Used as described, the protocol provides a straightforward method for super-resolution imaging of biological samples at cryogenic temperatures, with the advantage of avoiding potential fixation artefacts. The resolution and localization precision achieved are significantly better than those routinely obtained in conventional room temperature SMLM microscopes. These advantages alone should make superSIL SMLM microscopy a valuable addition to the toolbox of cell biology researchers.

It is also possible to consider other potential advantages of the superSIL microscopy protocol. Firstly, because of the ultra-high NA, the resolution that can be achieved even at room temperature is significantly better than that obtained in conventional SMLM microscopes. The superSIL protocol therefore provides a simple low-cost method by which a basic fluorescence microscope could be converted to a high-performance super-resolution system. Our protocol describes use of the superSIL in an upright microscope, but the method could also be used in an inverted microscope by using a modified version of the superSIL assembly.

We believe that superSIL microscopy could have an important role to play in CLEM, helping to bridge the resolution gap between cryo-fluorescence microscopy and EM. We do not describe any CLEM approaches in this protocol, but possible methodologies could involve the transferring and imaging of thin lamellae produced by focused ion beam (FIB) milling under both superSIL and transmission EM modes, or the correlation of superSIL SMLM images with serial block-face scanning EM. Given the small size of the lens assembly, it may also be possible to incorporate a superSIL into the stage of an electron microscope, potentially permitting simultaneous or near-simultaneous imaging of the same sample with both light and electrons.