Background

Individual cells and organized systems of cells within a tissue form a hierarchy of ultrastructural details going from complex organelle networks to single protein molecules. Understanding this architecture is a key link to get insights into structure and function. Various electron microscopy (EM)-based tools enable visualization of cellular structures with nanometre resolution. While transmission electron microscopy (TEM) and scanning electron microscopy (SEM) images have been key to our understanding of the ultrastructure of organelles as they generate 2D images which can reveal 3D structures, these methods are limited in that they only allow a single slice or a small volume of a cell to be captured. Moreover, the fact that the sampled fraction is often viewed at a specific sectioning angle through the cell or tissue, limits visibility of many structural aspects, unless many sections are examined. Stitching together multiple sections can extend the effective volume size, but is not very practical if a large number of sections needs to be stitched. Altogether, high resolution isotropic 3D whole cell data remains unachievable with current methods. A technique that overcomes the limitations of diamond-cutting thickness is the use of a focused ion beam, typically of gallium ions. This allows fine increments in cuts of a block surface as thin as 1nm. This technique combined with SEM, is called Focus Ion Beam Scanning Electron Microscopy (FIB-SEM). In their work, Xu and colleagues (1) aimed to advance FIB-SEM by joining its resolution advantages with the possibility of performing long-term imaging capable of months-to-years continuous imaging, without defects in the final image stack. In their work, the authors also optimized ion milling and electron imaging components to achieve 4nm sampling intervals in all dimensions, enabling high resolution imaging by rendering greater details with volumes approaching 100 µm³. In their work, the authors generated reference 3D image datasets at 4nm isotropic voxels for 10 different samples, including cultured cells and tissues. They made this data available as open access, in an interactive web platform called OpenOrganelle. The aim is that this data serves as a reference library for multiple purposes (including quantitative and qualitative studies of cell identity, cell morphology, cell-cell interactions, intracellular organelle organization and structure) and subsequent studies within the scientific community.

Key findings and developments

Key development: finer resolution

The isotropic resolution limits are convoluted by a) the waist size of the incoming primary electron beam, and b) the size of the scattering volume explored by the penetrating primary electrons. To optimize “a”, the authors found that a beam current of 200 to 300 pA, and a landing energy of 700 to 900 eV are optimal for isotropic 4nm imaging. They then characterized the resolution in x-y and z by analysing the step transitions at the edges of gold nanoparticles on a carbon substrate.

Key development: larger volumes

Upon acquisition of the largest possible volumes with the finest resolution, 6,000-10,000 electrons are allocated per voxel to achieve a target contrast to shot noise ratio of 3 or higher. This translates to a sampling rate of <200 kHz (dwell time > 5µs) needed with a 200-pA primary beam. Using the sampling voxel size of 4nm, the volumetric imaging rate was of 10µm³/day. To access volumes of typical whole cells, reliable long-term imaging (approximately one month), with nanometer stability, is required. The enhanced FIB-SEM technology hereby presented delivers continuous stable and reliable operation for months to years, and is optimized for fine resolution whole cell imaging.

Open Data

The resulting wealth of data arising from the task performed here, is unique and comprehensive, and covers multiple cells, with all their fine ultrastructural details at sub-organelle level. The authors made available 10 datasets containing 7 common wild-type cultured cell samples, and 3 different tissue samples. The authors made all the data available at https://openorganelle.janelia.org, which allows the wide scientific community to browse, download and explore the data. Within this database, the metadata and pre-selected views of interest are provided, while data can be further explored using an in-built visualization tool. Moreover, the repository contains segmentations into various categories of interest allowing quantification of numbers, sizes, shapes, contacts, locations, proximity, and multiple other parameters regarding organelles and proteins within the whole cell.

Key findings

1. HeLa cell: HeLa cells are human cervical cancer cells widely used across multiple research fields. Having such cells imaged fully and to great detail, provides a strong reference point for the multiple conditions and studies for which these cells are models. The authors present here a whole HeLa cell including all its organelles, such as the mitochondrial network, the centrosome, the Golgi apparatus, and the nuclear envelope. The dataset already shows the advantages of 3D isotropic imaging, which provides details that no single 2D section would allow visualizing.
2. Cytotoxic T cell attacking an ovarian cancer cell: The authors give this as an example of cell-cell interactions, focusing on the immunological synapse. They discuss that although the immunological synapse has been a main focus of interest, two-dimensional studies lack sufficient detail of this important interaction. Imaging of this interaction with FIB-SEM allowed a complete map of the complex membrane topology at the interface between the T cell and its target. Zooming into images allowed visualizing key features including membrane interdigitation, flat membrane apposition, and filopodia of the target cell trapped between two cells. Altogether, FIB-SEM allowed label-free localization of all organelles in the T cell, including those necessary for target killing and cytokine secretion.
3. Mouse pancreatic islets. Pancreatic islets are microorgans consisting mainly of beta, alpha, delta polypeptide cells and endothelial cells. Among them, beta cells secrete insulin stored in secretory granules to maintain blood glucose homeostasis. Beta cells have a polyhedral shape, and whole cell imaging at high resolution is known to be challenging. Large-scale high-resolution FIB-SEM enables the analysis of ultrastructural differences between beta cells within an islet, as well as sub-cellular features, such as ribosomes or the cytoskeleton. In their work, the authors were able to image the cytoskeleton, as well as interactions between beta cells including primary cilia, their connections to neighbouring cells, and the intermingling of microvilli. They also report close contacts of the ER and insulin secretory granules in 3D, and provide a comprehensive 3D representation of microtubule networks and how they interact with other organelles. Moreover, they observed insulin secretory granules enriched near the plasma in association with microtubules, and independent of extracellular glucose levels.
4. Drosophila brains. The authors investigated the central complex, and whether it contains characteristic synaptic motifs (including polyadic synapses, rosette synapses, or any new synaptic motif). They report that FIB-SEM allows reconstruction not only of large neuronal objects, but also smaller intracellular components such as synaptic vesicles, dense-core vesicles and microtubules.

What I like about this preprint

This is altogether a fascinating step forward both in the technology that enables this type of imaging, and in the fact that the resulting data was made publicly available to the scientific community and it can be further interacted with and used according to user-specific needs. I think reference databases in microscopy are often missing, and one such as this, with such impressive ultrastructural detail will be extremely valuable, and could set an important precedent for cell biology and imaging.

Open questions

1. How accessible and/or easily implementable are the modifications you report for FIB-SEM? Do you envisage that many labs world-wide will be able to perform this imaging, and contribute to the fascinating database you have begun?
2. You mention that the data you have acquired and which you discuss in this work is made available in https://openorganelle.janelia.org. Can groups that are able to perform eventually, this type of imaging and use the datasets you provide here as reference, upload their findings and comparisons to eg. pathological conditions of other sorts? What you have provided here is a fascinating basis for the scientific community.
3. The time required to perform this detailed imaging is still very large. Many labs have the limitation that if only one microscope is available, it cannot be blocked for months at a time. Do you envisage that isotropic 3D EM will eventually be possible, albeit in shorter amounts of time?
4. You tested your developments in various types of cells, and you discuss some of the obstacles presented by some of them. Do you envisage other limitations that readers and potential users should be aware of when exploring multiple types of samples?

References

1. Xu et al, Isotropic 3D electron microscopy reference library of whole cells and tissues, bioRxiv, 2020.

 

doi: https://doi.org/10.1242/prelights.26566

Read preprint (No Ratings Yet)