Correlative Light Electron Microscopy
A Way of Finding "The Needle in The Haystack"
- Correlative Light Electron Microscopy - A Way of Finding « The Needle in The Haystack »
- Fig. 1: Diagram of the sample preparation for correlative workflow. A. The chemically fixed brain slice is imaged by fluorescence microscopy to map the fluorescently labelled cells; here we clearly identified the neuron. The cell coordinates are then recorded and will serve as a map for the cryo-sectioning step. B After sucrose infiltration step and plunging into liquid nitrogen, the tissue block is installed inside the cryo-chamber of the microtome. Using the stereomicroscope, the fluorescent cell (black arrow) can be imaged during this step. C After cryo-sectioning, the section is warmed up ot room temperature and imaged at high resolution (high NA) by fluorescence microscopy. The cell body of the fluorescent cell is easily seen on the section. This image will be used as a guide to drive the electron microscope to the area of interest e.g. the fluorescent cell. D The cell of interest (white arrow) is then imaged at high resolution by HAADF STEM SEM imaging. Bars represent: A = 100 µm, B = 350 µm, C = 200 µm, D = 2 µm
- Fig. 2: Setup of the cryo-microtome with a fluorescence microscope. To trim and cut the area of interest accurately, the fluorescent cells need to be visible all the time during mircotomy (A). Therefore the original stereomicroscope is removed and replaced by a fluorescence stereomicroscope (B). In addition the microtome is placed in a climate chamber to adjust humidity to prevent frosting of the sample and cryo-chamber during sectioning (C).
- Fig. 3: Transmission electron micrographs of brain prepared by the Tokuyasu cryo-sectioning method. A Overview of an astrocyte cells body (As) close to myelinated (My) axons (Ax) recognized by the gold label (black dots) of the GFP, which is constitutively expressed in these astrocytes. The cytoplasm contains mitochondria (M), numerous vesicles and a cell nucleus (N) with the typical pattern of hetero- (H) and euchromatin (E). B Here the thin protrusions of the astrocyte (As) and a classical synapses (Sy) are visible as well as the extracellular space (ECS). At higher magnification (inset) the synaptic membranes and the post-synaptic density is clearly apparent and numerous synaptic vesicles (arrows) are detected. Bars represent: A = 500 nm, B = 50 nm.
Biological organisms are complex and often we tackle the problem of understanding the role of a specific cell in the organism. Combining light/fluorescent and electron microscopy, i.e., correlative light electron microscopy is the appropriate way to dissect this complexity. Here we propose a workflow based on the cryo-sectioning method to characterize, at high-resolution, a small biological feature, a “needle”, spatially localized inside a large biological volume, the “haystack”.
An image tells more than a 1000 words, this is the reason why, imaging techniques, predominantly fluorescence microscopy, have been intensively employed to dissect the complexity of 3D samples, especially since the introduction of super-resolution microscopy techniques . However, this set of techniques has the disadvantage that only labeled structures can be studied in relation to each other and nm-sized organelles cannot be resolved . On the other hand, electron microscopy has been the method of choice to observe the molecular organization of a sample with a higher resolving power. Thus specific proteins can be localized and can subsequently be related within the biological context. However, the field of view in the transmission electron microscope is limited to few hundred μm2.
Combining a low magnification overview and localization capabilities of light/fluorescent microscopy with the high resolution of electron microscopy, i.e., correlative light electron microscopy, appears to be the most appropriate way to dissect the complexity of biological samples. There are many different ways of doing correlative light and electron microscopy. Here, we describe our approach to follow a few fluorescently labeled cells from the excision of the tissue, until the high resolution imaging by electron microscopy . To transfer a precisely located biological region, from the light to the electron microscope, we developed a workflow based on the cryo-sectioning method of Tokuyasu . The first step is based on the 3D localization of the structures of interest in the entire sample by fluorescence microscopy.
The position of the fluorescence labeled structures is recorded and then used during the cryo-microtomy step. In order to better control the cutting area, we modified our cryo-ultramicrotome by replacing the classical binocular by a stereo-fluorescent binocular microscope. Thus we are able to visualize the fluorescence in the tissue bloc during trimming and sectioning. The second step is based on the acquisition of a precise 2D map of the fluorescence on the cryo-section sitting on the TEM grid. This map will be reused to guide the electron microscope during acquisition. Finally, we image the area of interest, at high resolution, with a scanning electron microscope using a STEM detector to acquire large tiles of a this area (e.g., 3600 μm2) down to 1 nm pixel resolution and with the transmission electron microscope to acquire tomograms of specific subareas.
Sample Preparation while Following the Fluorescently Labeled Cells
Vibratome sections of mouse brain with either eGFP-labeled astrocytes  or individual neurons filled, by microinjection, with Lucifer yellow are immediately fixed in a mixture of formaldehyde and glutaraldehyde. While in fixative, the whole slice is imaged at low magnification in a fluorescence microscope to map the fluorescent areas and cells (fig 1A). After fixation, small blocks of about 1 mm3 from the zone of interest are cut and infiltrated with sucrose . Each tissue cube was mounted on an aluminum pin used for microtomy, and then imaged by fluorescence microscopy in three dimensions (x, y and z) to precisely localize the labeled cell(s). The Z-imaging can be done with any (wide-field) fluorescence light microscope, and there is no need for confocal microscopy. Imaging was performed from top to bottom of the tissue cube using low magnification air objectives. This provided the 3D coordinates of the cell of interest inside the brain tissue. The Z-position of the cell is estimated at the position where it is in focus.
After imaging, the tissue sitting on the aluminum pin was plunged into liquid nitrogen and the sample is mounted in the cryo-ultramicrotome to be sectioned.
Cryo-Ultramicrotomy and Fluorescence
The robustness of our correlative light electron microscopy workflow is based on the possibility to image the sample using fluorescence excitation, and then track the area of interest at each processing step prior to electron microscopy. Numerous devices are available to image the fluorescence but the imaging during the cryo-trimming and the cryo-sectioning was not possible. Conventional cryo-microtomes are equipped with a binocular that does not allow fluorescence imaging therefore, we removed the optical part, and replaced it with a fluorescence stereomicroscope (fig. 2). We chose a system combining large depth of field and good resolution (fig. 2A). This detail is important for the spatial rendering indispensable for the cryo-sectioning. Indeed, the position of the sample relative to the edge of the knife must be known for trimming and sectioning. For this, we selected an objective with a long working distance (112 mm). To optimize the sample visibility, we inclined the microscope at an angle of approximately 20°C from the vertical and installed it on a swing arm stand with vibration-damping foot to stabilize the microscope during sectioning (fig. 2B). To limit the ice contamination during sectioning and to maximize the fluorescence visibility, the set up was installed in an enclosed dark room with a system controlling the humidity (fig. 2C). This “homemade” setup can be easily reproduced in laboratories performing cryo-sectioning.
When the area of interest is reached, e.g., the labeled cell (fig. 1B), serial thin cryo-sections are cut, picked-up and installed on electron microscopy grids. The grids are temporarily mounted in buffer between a microscope slide and a coverslip. In the bright field and fluorescence mode, the whole grid is imaged at low resolution, to relate the position of the sections on the grid. Afterwards maps at higher magnification are taken including the use of high NA oil objectives for the fine correlation (fig. 1C). After the imaging step, the grid is removed from the glass slide/coverslip sandwich. Additional preparation steps, like immunogold labeling and heavy metal contrasting, can be done before the sections are embedded in a thin layer of methylcellulose  to be observe under electron beam. This step will irreversibly quench all fluorescent signals.
The grids are mounted on a grid holder for a scanning electron microscope equipped with a STEM detector and image with Maps software (FEI Company) dedicated to correlative microscopy. Using this software, 2 distinct points are imaged either in the SE mode or the STEM modes then aligned with the bright field image of the whole grid. This alignment is correct within about 5 µm. Then the high-resolution micrograph is aligned by using 2 conspicuous points of one section that are visible in the fluorescence and the STEM image. Together with the fine alignment options of the software both images can be aligned to high precision. The high-resolution light micrograph serves as a guide to find the cells of interest. Individual images or large images of several thousands of square micrometers (at nanometer resolution) can be taken (fig. 1D). For STEM imaging we prefer the high-annular dark-field mode because of its higher contrast and better signal-to-noise ratio . After STEM imaging, the grid sitting on the holder can be introduced in the TEM to acquire tomogram(s) of some specifics areas of the cell of interest.
After having focused on the technical aspect of this method, it is also important to mention the sample ultrastructure. Indeed, careful inspection of the micrograph enlightens good preservation of the ultrastructure of the sample without any obvious extraction artifacts (fig. 3). The cryo-section method is particularly suitable for visualizing membranes, vesicles and mitochondria. We clearly identify synapse (fig. 3B) with post and pre densities. Moreover, this approach is optimal to preserve antigenicity and allows immunolabeling to specifically localize antigen in the tissue. For figure 3, immuno-gold localization for the GFP was performed.
Here we demonstrated a workflow to keep an eye on the structure of interest from the moment of sampling until detailed analysis at high resolution. An important feature of this approach is that the fluorescence of the cell of interest could be preserved throughout the preparation procedure, allowing full control of the area of interest at any time. We strongly believe that this workflow can be used to study a various range of samples and be adapted to the special needs of the investigators, e.g., to follow a fluorescently labeled protein in time and then arresting the process when the protein has reached its destination to go for high-resolution analysis of that location.
The authors acknowledge financial support from the Faculty of Biology and Medicine of the University of Lausanne and the Swiss National Science Foundation, R’Equip grant 316030_128692. We thank Dr. Matthias Langhorst for fruitful discussions and help with the fluorescence microscopy, Dr. José Maria Mateos, Dr. Andrea Volterra and Dr. David Bouvier for providing samples and last but not least Willy Blanchard for the photographs and the art work.
 Huang B. et al.: Cell, 143, 1047-1058 (2010)
 Loussert C. et al.: Methods in Cell Biology, 111, 59-72 (2012)
 Loussert C. et al.: Journal of Structural Biology, 189, 53-61 (2015)
 Tokuyasu K.: Journal of Cell Biology, 57, 551-565 (1973)
 Nolte C. et al.: Glia, 33, 72-86 (2001)
 Yakushevska A.E. et al.: J. Struct. Biol. 159 (3) 381-391 (2007)
Dr. Céline Loussert Fonta (corresponding author via e-mail request)
Nestle Research Center
Bruno M. Humbel PhD
Electron Microscopy Facility
University of Lausanne