Limits of SBF imaging

As discussed previously, serial block face imaging can introduce several artifacts that can limit high-resolution imaging of the specimen. Artifacts can be formed during the preliminary preparation steps or during the observation under the effect of the electron beam hitting the sample. The specimen is embedded in a hard resin that maintains the sample stable under the electron beam but with also a minimum of softness to allow diamond knife ultrathin sectioning. Longer exposition time (dwell time) increases the number of electrons interacting with the sample at the surface softening and melting the resin. As a consequence, the surface of the resin block becomes cracked and irregular as shown in figure 4.6A. Changing the setting such as the accelerating voltage, aperture, dwell time and pressure could limit this artifact but reduce the signal to noise ratio and thus the amount of details seen on the image. Knife marks are a common issue in TEM imaging and are generally inevitable. These marks might be due to damages of the diamond knife or due to the presence of resin particles remaining on the cutting edge of the knife. The same issues are encountered with SBF imaging and the resulting image presents “wheel marks” artifacts as seen in figure 4.5B. Serial block face imaging involves serial sectioning and imaging of the surface of the resin block. Normally, sections shed off the surface of the specimen after sectioning. In some cases, however, sections remain attached at the cutting edge of the diamond knife and redeposit at the surface of the block when the knife retracts. When the specimen is imaged, a folded section appears at the surface of block as shown in figure 4.6C. This artifact will disappear with the next sectioning cycle. One of the major issues encountered with scanning electron

microscopy imaging is termed the charging effect. This artifact is the consequence of the accumulation of electrons at the surface of the specimen. As shown in table 4.1, the charging effect increases with the acceleration voltage, low pressure and a large aperture. Despite the greater generation of back- scattered electrons and thus a higher signal to noise ratio, charging of the specimen is typically manifested by a bright spot artifact on the image as shown in figure 4.6D. The charging effect can be limited by reducing the number of electrons hitting the sample (table 4.1) but also by increasing conductivity of the sample during the preparation by sputter coating the surface of the specimen with gold-palladium and using silver epoxy glue for binding the resin block on the pin.

Figure 4.6 Artifacts commonly observed with serial block-face

imaging.

A. Resin softening and melting (black arrows) occurs when the

number of electron hitting and interacting with the sample is

too high.

B. Knife marks (white arrows) are also a common artifact

generally observed with TEM imaging and are the result of

either damages, or particles remaining at the cutting edge of the

diamond knife.

C. A folded section cut off the resin block in the previous cycle is

sometimes deposited at the surface of the block (white arrow).

D. Accumulation of electrons at the surface of the sample is at

the origin of the charging effect manifested by a bright spot

artifact on the image (black arrows).

4.3 Discussion

Electron microscopy is currently undergoing a revival with the emergence of new volume EM techniques that enable the collection of large amounts of data and the imaging of tissues, cells and sub-cellular structures with unprecedented detail. Serial block-face imaging is still a recent innovation but the number of publications referring to this emerging imaging technique is constantly increasing and the technology has now been applied to a wide type of organisms, tissues and cells (Peddie & Collinson, 2014).

SBFSEM records serial images of the surface of a specimen in a process that is completely automated. The process generates a large data stack of serial images with a resolution approaching that of transmission electron microscopy for the imaging of biological samples. Initially developed for imaging and volume

reconstruction of the neuronal network of the central nervous system, SBFSEM has now been applied to image numerous tissues, organs and cell types (Peddie & Collinson, 2014). Because of the large volume of tissue that can be imaged in one run, SBF imaging appears as a powerful tool for the observation of multicellular structures such as LESCs and LCs. However, segmentation of the data stack by hand is time-consuming and tedious as there is no reliable software capable of automatic segmentation of membranes and volume rendering. For this reason, manual segmentation is often focused on a very specific type of cell or organelle and requires a precise preliminary analysis of the collected data stack. In the present chapter a protocol has been developed to image for the first time the limbal basal epithelial layer and the limbal stroma by SBFSEM. The main issue encountered in the preparation of human limbal biopsies for SBF imaging was the poor preservation of the specimen prior to fixation. This was due to the post mortem degenerative and release of intracellular enzymes as organelles breakdown, which irreversibly affects the quality of corneal biopsies. However, even if the lateral resolution reached in this study was lower than has been reported in other tissues and model organisms, it was sufficient to characterize cell-to-cell interactions that are present in the human limbal stem cell niche. This is the subject of the following chapter.

Chapter 5: High-resolution imaging