Cryogenic electron microscopy (Cryo-EM) fundamentals

In cryo-ET a vitrified sample is imaged in a TEM as it is tilted from approximately -60⁰ to +60⁰. This results in a series of 2D images that can be combined to reconstruct a 3D image, a so-called tomogram. This technique is powerful enough to resolve 3D structures of intracellular organelles and protein complexes in near-native state in their cellular environment. Resolving structures in their cellular context is a groundbreaking advancement for structural and cellular biology. The location, as well as the structure of a biomolecule, is crucial to the understanding of its cellular function.

Samples are vitrified by plunge freezing or high pressure freezing

To vitrify a sample several techniques are available, the most widely used are plunge freezing and high pressure freezing. The choice of the technique mainly depends on the size of the sample. For samples thinner than 10 µm plunge freezing can be used. In this method the grid is rapidly plunged in a cryogen (e.g. liquid ethane) cooled by liquid nitrogen. For samples thicker than 10 µm, sufficiently high cooling rates cannot be obtained by plunge freezing. A slow freezing rate causes formation of ice crystals, which damages the sample. For this reason, thicker samples are often frozen using high-pressure freezing, which can be used for samples with a thickness of up to 200 µm. Here, the sample is frozen under high pressure (2100 bar), which allows adequate freezing rates for vitrification.

Samples are thinned by cryo-FIB milling or by cryo-ultramicrotomy

The maximum resolution of a TEM scales down with the increased thickness of the sample. Therefore, to obtain a high resolution tomogram samples should ideally be between 100 and 300 nm thick. A vitrified sample can be sectioned using cryo-ultramicrotomy: cutting the sample with diamond knife under cryogenic conditions. This technique, however, has a number drawbacks, the sections tend to have physical defects, get compressed and wrinkled, making them difficult to image. The most successful technique to prepare a thin section within the sample, a so-called lamella, is milling using a focussed ion beam (FIB). With this technique, a focussed beam of gallium ions at high current is used to sputter away material around the region of interest (ROI) to create a lamella. This process is performed in a scanning electron microscope (SEM) fitted with a FIB (FIB/SEM) that is equipped with a cryo-stage to keep the sample under cryogenic temperatures.

Regions of interest are identified by cryogenic fluorescence light microscopy

Identifying the ROI to mill in the right location is a crucial step, since being off target for this process could result in milling away your structure of interest. To overcome this, cryogenic fluorescence light microscopy (cryo-FLM) is often used. In this technique biomolecules of interest are fluorescently labelled and the sample is first imaged with cryo-FLM to identify the ROI. The FIB is then used to create a thin lamella in the sample exactly at the ROI and the sample is transferred to the cryo-TEM for high resolution imaging. This method has become the gold standard for cryo-ET since it produces high resolution EM data that is superbly correlated with FLM while the cryogenic conditions keep samples at a near-native state. However, this method is very laborious and error prone since the sample needs to be transferred between three different microscopes.

Optimization by integration - Integrating a fluorescence light microscope into the FIB/SEM

As was discussed in the previous sections, the optimal method to process a sample for cryo-ET is also the most laborious and includes three different microscopes. The fragile sample needs to be transferred from the cryo-FLM to the cryo-FIB/SEM and then to the cryo-TEM, all while maintaining a temperature below -160°C. To complicate things further, the sample is often transferred back to the cryo-FLM after FIB milling to confirm that the ROI is still present in the lamella. These transfer steps make the workflow an exceedingly challenging task that translates in a low sample preparation success rate of approximately 20%. At Delmic we have designed several products that integrate fluorescence microscopy into a cryo-FIB/SEM system. These designs abolish the need for a separate cryo-FLM system and most importantly, reduce sample handling. Using an integrated FLM in the FIB/SEM means that ROI identification by FLM, SEM imaging, FIB milling and ROI confirmation by FLM can all be done in the same system without the need to handle the sample, thus increasing the success rate of the cryo-ET workflow considerably.

The current cryo-ET workflow is a complicated and long procedure that requires a highly experienced operator. Integrating FLM imaging into the FIB/SEM increases the success rate significantly, but more can be done to optimize the entire workflow.

Many cryo-TEMs are capable of processing multiple samples using an autoloader system, but in the rest of the cryo-ET workflow, sample preparation in batch is not possible. To tackle this issue, we introduce an automated sample transfer system for the FIB/SEM. This system uses cassettes that hold up to 12 grids and is compatible with autoloader systems on most existing cryo-TEM systems. The automated transfer system works under vacuum and cools the samples without submerging them in liquid nitrogen, which reduces contamination risk. This results in a workflow that abolishes the need for manual handling of grids after sample vitrification and loading into the cassette. Reducing the handling steps and necessary transfers translates into a major leap forward in efficiency and success rate.

Optimization by sample processing automation – Automatic FIB milling and ROI validation

A third optimization point in the workflow is the milling of the lamellae. This process is usually performed by manually selecting a ROI and then designing and milling the lamella. Here, specialized software can assist in automating the milling process. After an overview of a grid is captured in fluorescence mode, ROIs are selected and detailed Z-stacks of the regions are captured in FML mode. In this detailed fluorescence map, the exact location of the point of interest is selected and the lamellae are automatically milled and FLM images are taken afterwards to check whether the lamellae still contain the region of interest. This process can be performed in batch with many grids to have the system perform the milling of the lamellae unassisted overnight, freeing up time on the system.

Together, our solutions reduce the exceedingly challenging cryo-ET sample preparation workflow to a few simple tasks, paving the way to making cryo-ET a routine technique.

Obtaining a tilt series in a TEM currently takes 20-60 minutes which slows down the cryo-ET workflow considerably. Fortunately, next-generation direct detectors, hardware for faster data transfer as well as methods for rapid tilt series are currently being developed [11]. Combined, a tilt series could be obtained in only a fraction of the time it currently takes, speeding up the workflow considerably. This development also increases the need for easy and automatic sample preparation methods to optimally make use of the cryo-TEM. Automatic sample preparation combined with high-speed TEM acquisition will result in the possibility to resolve large amounts of biological structures.

As more and more structures become available data processing could be optimized by automatic mapping. A tomogram that contains several biological structures could be automatically screened against a database. Known structures would be fitted to the structures visible in the tomogram, showing which known structures are present. Routinely obtaining biological structures could also result in the rise of quantitative structural biology: doing statistical analysis to gain a deeper understanding of structure-to-structure variations.

Cryo-ET will reveal structural information of many cellular processes and pathways

A major step forward has been made possible by cryo-ET by moving from obtaining structural data in purified samples to getting the same high resolution data in their biological context, or in situ. At the moment, many studies are focusing on large protein complexes that “stand out of the crowd” due to their known size or location, making them relatively easier to identify and image. As the cryo-ET technology develops further, it will become easier to identify increasingly rare cellular structures and events. Obtaining high resolution structural data of rare cellular events will have a great impact on fields such as drug development and cellular immunology, but will also likely reshape our knowledge on many of the textbook signalling pathways and trafficking routes of the cell.

Even though proteins often assemble in stable multiprotein complexes, most protein-protein interactions are transient in nature [12]. These interactions, often accompanied by changes in protein fold or shape, hold crucial information on how and where important processes take place. Capturing the structural detail of such transient interactions in their cellular context seems like an unobtainable feat at the moment, but with the current advancements in cryo-ET technology, it might not be that far away.

Likewise in the field of nucleic acid research, cryo-ET can play a fundamental role unravelling processes such as the post-transcriptional regulation by (non-coding) RNA. The change in secondary structure of RNA can directly translate into a specific function, such as regulation of gene expression, protein interaction or RNA-RNA interaction [13]. Since the first discoveries of non-coding RNA molecules, it has become apparent that RNA molecules regulate a host of complex cellular processes that are tied strongly to their confirmation [14]. Here, cryo-ET could shed further light on the structure-function relation of RNA in its cellular context. Here the challenge lies in the vast amount of content in the cytoplasm and the relatively small size of signalling RNA molecules. For this, increased precision in finding molecules of interest with cryo-FLM and increased signal-to-noise ratio to resolve structures in the crowded cytoplasm are needed.

At Delmic we strongly believe that cryo-ET will revolutionize the field of structural biology and become the method of choice to obtain the 3D structure of a biomolecule. We aim to make cryo-ET more accessible to researchers by providing solutions that simplify the workflow. We do so by consolidating the sample preparation workflow with an automated approach.