Electron tomography—a tool for ultrastructural 3D visualization in cell biology and histology

When inspecting the organization of organelles of a particular cell or the cellular arrangement in several tissues at the ultrastructural level, we usually start from two-dimensional images. We then interpolate a three-dimensional (3D) image from sections at different heights of the same site of an electron microscopic (EM) preparation. The resulting reconstruction, conceived in our imagination, can lead to missing details or general misinterpretations. One of the first successful attempts to overcome these problems concerned the 3D reconstruction from ultrathin serial sections varying between 60 and 80 nm. [1‐3]. However, such studies are time consuming and require a skilled technical laboratory staff. Therefore, scientists using electron microscopy sought alternative methods, such as electron tomography (ET).

In this review, we give a short introduction into ET and a historical overview of its development in relevant important centers. We discuss the principle of ET, including possibilities and limitations. We also refer to ET investigations carried out in our department and mention by name scientists who were helpful in establishing this method in our EM laboratory. We also summarize the development of new advanced methods, such as focused ion beam scanning tomography (FIB-SEM), serial block-face scanning electron microscopy (SBF-SEM), axial bright-field scanning transmission electron microscopy (STEM) tomography, and array tomography (AT).

ET takes place in a transmission electron microscope (TEM), in a TEM equipped with scanning device (STEM), or in special scanning electron microscopes at an acceleration voltage of at least 200 kV. In ET using TEM, a semithin section mounted on a grid is inserted into a high-tilt holder, which can be tilted by a eucentric goniometer over a range of maximally −70° to +70°. Typically, depending on the specimen, a range of −65° to +65° is used. The thickness of the semithin sections is restricted to 300 nm in TEM mode and to 1 µm in STEM mode. ET requires a eucentric goniometer with a complex stage, where the grid holder can be moved in directions of the x‑, y‑, and z‑axes. Computer programs for correcting dislocations and providing automatic adjustment of the eucentric height and focus (autofocus) during tilting are therefore indispensable. During tilting, digital images are acquired via a sensitive CCD camera at preset intervals (usually 1°). The tilting limits are set by the edge of the grid and the fixation ring at the holder, which would cover the section at higher tilting angles. This reduces the 3D information and is referred as “missing wedge”. This disadvantage can be partially compensated by carrying out a second tomogram from the same ROI, usually at an angle of 90° to the first axis (dual axis tomography), reducing the missing wedge to a missing cone and providing higher 3D information. The digital images are stored in stack files, which are put through further processing software packages [4]. The next step is the reconstruction, transforming the projections in a Fourier space, from which they are placed as virtual slices into a virtual 3D volume. This method is referred as weighted back projection (WBP). Corresponding structures through the whole 3D space can be traced and connected by the computer in order to obtain a 3D model. Beside the WBP method, the simultaneous iterative reconstruction technique (SIRT; [5]), and the discrete algebraic reconstruction technique (DART; [6]) have been developed in order to improve the image information. Fig. 1 shows a flowchart of the consecutive steps in ET as used in our laboratory. Compared with 3D reconstructions based on serial sections, ET provides significant advantages in z‑axis resolution of the virtual slices which correlates with the voxel size in nanometers, depending on the image resolution and the EM magnification used for ET acquisition.

Investigations using ET started in the last decade of the past century, as summarized in 1997 [7]. Subsequently, several reviews described the technique [8, 9].

To improve the reliability of EM preparations by avoiding chemical artifacts, freezing methods in liquid nitrogen, such as the high-pressure fixation (HPF), were developed. Frozen cells or tissues can be sectioned using cryo-ultramicrotomy. The cryosections can be transferred via a liquid helium-cooled cryostage into a cryo-electron microscope. This method has been applied, above all, by the group of Wolfgang Baumeister at the Department of Molecular Structural Biology, Max Plank Institute for Biochemistry in Martinsried, Germany [7]. Baumeister’s group characterized proteasomes and their subunits by combining crystallographic studies and ET, including single-particle analysis [10]. This group has presented excellent publications on cell organelles [11‐15]. Structural and functional cellular aspects were visualized by using special apparative innovations for correlative light and cryo-electron microscopy [16]. An important contribution in neurology concerned the presynaptic organization of vesicles and surrounding filaments [17], as well as the presence of PolyQ inclusions in Huntington disease [18]. Compelling structural research has been done in bacteria [19‐21] and viruses [22‐25].

The working group of Abraham J. Koster (section Electron Microscopy, department Molecular Cell Biology at the University of Leiden, Netherlands) published articles providing significant technical improvements of ET [26‐29]. Several papers contributed to a better understanding of the structure and function of the endoplasmic reticulum, the Golgi apparatus, and related organelles [30‐33], as well as of the endothelium-specific Weibel–Palade bodies [34, 35].

Early investigations leading to 3D visualization of the Golgi apparatus were also undertaken by the research group of Kathryn E. Howell at the University of Colorado School of Medicine, Department of Cell and Developmental Biology [36, 37].

Our primary and central interest was in the 3D visualization of the dynamic structure of the Golgi apparatus in cultured human hepatoblastoma cells under the influence of several environmental factors. ET was implemented in our laboratory in 2004, under helpful instructions by Wim Voorhout (FEI), Bruno M. Humbel, and Jean Luc Murk, Wolfgang Baumeister, Jürgen M. Plitzko, and Matthias Eibauer from the Max Plank Institute of Biochemistry in Martinsried.

In our ET research group of the Department of Cell Biology and Ultrastructure Research (Head: Margit Pavelka), Adi Ellinger was one of the most enthusiastic members together with Margit Pavelka, Josef Neumüller, Claudia Meisslitzer-Ruppitsch, Carmen Ranftler, and Monika Vetterlein, and the technicians Peter Auinger, Ivanna Fedorenko, Ulrich Kaindl, Beatrix Mallinger, Thomas Nardelli, and Elfriede Scherzer.

Throughout many years, the complex and dynamic Golgi apparatus architectures were at the center of our interest. To avoid artifacts due to chemical fixation, we used high-pressure freezing followed by freeze substitution and resin embedding (Fig. 2). We visualized the endocytic cellular routes using horseradish peroxidase (HRP)-labeled plant lectin wheatgerm agglutinin (WGA) combined with diaminobenzidine (DAB) cytochemistry [38, 39].

We were interested in correlative microscopy, since light microscopy allows the visualization of structures and metabolic processes using fluorescent markers or tracers while EM can only demonstrate this in fixed, high-pressure cryo-fixed, or frozen specimens but with the advantage of high resolution. Our challenging approach was to combine correlative microscopy with ET: Cell organelles, labeled with fluorescent compounds, were subjected to photooxidation of DAB. In the presence of DAB, cells were photobleached by epifluorescence. Aldehyde-fixed cells on plastic petri dishes were incubated in a DAB solution and irradiated with light at 488 nm, until no fluorescence could be detected. Light microscopically, a dark precipitate was seen at the area of irradiation. This area was marked on the petri dish and further processed for EM. The DAB precipitates at the previously fluorescent sites could then be visualized also at the ultrastructural level [40‐43].

The highly dynamic nature of the Golgi apparatus could be studied with the aid of the non-metabolizable glucose analogue 2‑deoxy-D-glucose (2DG). Treatment of cultured cells with 2DG causes dis- and reorganization of the Golgi apparatus stacks and correlated with changes in the cellular ATP levels. The Golgi apparatus is maintained in the form of Golgi bodies, from which the stacks can be rebuilt after 2DG removal [44, 45].

Our ET experience in cell biology encouraged other research groups to join and collaborate with us. One project initiated by Johannes Huber, Division of Gynecological Endocrinology and Reproductive Medicine, Medical University of Vienna, dealt with the propagation of endothelial progenitor cells (EPCs) from human cord blood. As a distinct maturation criterion, the formation of endothelium-specific Weibel–Palade bodies (WPBs) was demonstrated. These cigar-shaped rod-like bodies form microtubular structures with a diameter of 13 ± 2 nm. Using ET, we were able to visualize the supply of newly formed WPBs from TGN as well as bifurcated WPBs that could only be demonstrated by ET ([46]; Fig. 3). In cooperation with Herbert Stangl and Clemens Röhrl, Institute of Medical Chemistry, Center for Physiology and Pathophysiology, Medical University of Vienna the uptake and intracellular transport of high-density lipoprotein were followed in EPCs by using fluorescence microscopy and ET [47].

A more recent collaboration involved the research group of José Martínez-Menárguez from the Department of Cell Biology and Histology, Biomedical Research Institute of Murcia (IMIB-Arrixaca-UMU), University of Murcia, Spain. We studied the complex 3D architecture of the Golgi apparatus of principal cells of the rat epididymis [48]. ET revealed an abundant Golgi network and a system of branching interconnected cords.

A permanent collaboration was set up with the Blood Donation Center of the Austrian Red Cross for Vienna, Lower Austria and Burgenland (Wolfgang R. Mayr, Christof Jungbauer and Renate Renz) and the Department of Blood Group Serology and Transfusion Medicine of the Medical University of Graz (Thomas Wagner). The interaction of platelets (PLTs) with bacteria has been shown in PLT concentrates after spiking with Staphylococcus aureus. The uptake of these bacteria into the open canalicular system (OCS) was visualized by using ET [49, 50]. Bacterial contamination leads to activation of PLTs and to the formation of platelet microparticles (PMPs; Fig. 4) responsible for causing unexpected and fatal transfusion reactions [51].

A very challenging project has been carried out together with Robert Borny from the Division of Cardiovascular and Interventional Radiology, Department of Biomedical Imaging of the Medical University of Vienna: ET 3D models were created in order to visualize nucleophilic cross-linked dextran-coated iron nanoparticles [52].

A further collaboration with the research group of Gerd Leitinger and Dagmar Kolb from the Institute of Cell Biology, Histology, and Embryology, Center for Molecular Medicine (ZMM), and the Core Facility “Ultrastructure Analysis”, Center for Medical Research (ZMF), Medical University of Graz focused on investigations of the presynaptic densities in the central nerve system of locusts. Using ET, the architecture of the L‑neuron output synapses revealed that vesicles adjacent to the active zone of the presynaptic density exhibit a highly ordered arrangement, provided by electron-dense filamentous strands. At the site of filamentous extensions of the presynaptic densities, the so called Bruchpilot protein, important for flying and running, was identified by preembedding immunogold labeling [53].

A challenging collaboration started in 2008 with the research group of Kristijan Jezernik Faculty of Medicine, Institute of Cell Biology, University of Ljubljana, Slovenia concerning the nature and function of fusiform vesicles (FVs) in the rat urinary bladder. These organelles represent parts of urothelial plaques, located as microdomains at the apical plasma membrane of the superficial epithelial cells. They are involved in the change of shape of the urothelium during extension or contraction of the urinary bladder. Using ET, we were able to show that these organelles exhibit the shape of flattened discs organized in the form of stacks. It was demonstrated that urothelial plaques contain uroplakins, which are enriched in the post-Golgi compartment [54, 55].

Further ET projects were realized in collaboration with Siegfried Reipert and colleagues from the Core Facility of Cell Imaging and Ultrastructure Research, Faculty of Life Sciences, University of Vienna on the visualization of tubulo-helical membrane arrays (THUMA) in PtK2 rat kangaroo epithelial cell lines [56] and on intracellular mesocrystalloid structures in the ovisac of the sea branchiopod Artemia franciscana [57].