Electron Tomography

Advanced article

Wanzhong He, Department of Biological Sciences, National University of Singapore, Singapore Jose´-Jesu´s Ferna´ndez, Centro Nacional de Biotecnologı´a – CSIC, Cantoblanco, Madrid, Spain

Article Contents . Introduction . Principle of ET . Specimen Preparation Techniques in ET . Strategies for Data Acquisition . Interpretation of Tomographic Reconstructions . Biological Applications . Illustrative Biological Examples . Conclusions . Acknowledgements

Online posting date: 15th January 2010

Electron tomography (ET) is an emerging electron microscopy (EM) technique for the three-dimensional (3D) visualization of the cellular architecture and molecular organization in native cells at nanometer scale. ET thus bridges the gap between the low-resolution imaging techniques and the high-resolution structural techniques. In ET a series of images is taken from a single specimen at different projecting orientations. The 3D map computed from these images is subsequently subjected to visualization and interpretation. Thanks to several recent innovations in EM, including improved specimen preparation techniques, novel EM hardware and software, reliable genetic and molecular labelling approaches, ET is rapidly becoming a powerful bioimaging tool for precisely dissecting the macromolecular organizations and cellular events captured from living cells in health and disease. Thus many previously unanswered molecular mechanisms responsible for the specific cellular functions could be clearly elucidated by the ET technology.

Introduction ‘The key to every biological problem must finally be sought in the cell’, wrote the classical cell biologist, EB Wilson, in 1925. Imaging the cells has been an important focus of scientists in life sciences for several centuries. Under the

ELS subject area: Cell Biology How to cite: He, Wanzhong; and Ferna´ndez, Jose´-Jesu´s (January 2010) Electron Tomography. In: Encyclopedia of Life Sciences (ELS). John Wiley & Sons, Ltd: Chichester. DOI: 10.1002/9780470015902.a0021877

traditional light microscope, cells were just considered as a ‘biochemical bog’ without too many subcellular structures. The advent of modern cell biology by using electron microscopy to study the fine cellular structures and organizations in chemically fixed cells has revolutionized the concepts on subcellular structures (McIntosh, 2007). In the ‘Golden Age’ of cellular electron microscopy (EM) (1945–1970), the electron microscope became the dominant tool for the discoveries of the cellular ultrastructures and organelles. However, cellular EM has to give its seat to modern light microscopy (LM) due to the several key improvements developed during 1980–1990, such as the fluorescence microscope and the confocal light microscope, which can more easily locate molecules in living cells by using fluorescent labels. Nevertheless, a lot of important biological questions are still unanswered by LM due to its limited spatial resolution (100–500 nm). Thanks to several recent innovations in EM (e.g. specimen preparation techniques, EM hardware and software), the modern cryoEM techniques allow the visualization of biological specimens from atomic to a few nanometer resolution in three-dimensions (Figure 1; Leis et al., 2009; Hoenger and McIntosh, 2009). Specifically, the emerging cellular electron tomography (ET) has attracted more and more interest of cell biologists in recent years. See also: Confocal Microscopy; Electron Microscopy; Fluorescence Microscopy; Fluorescence Microscopy; History of the Optical Microscope in Cell Biology and Medicine; Light Microscopy ET has emerged as the leading technique for threedimensional (3D) structural analysis of unique complex biological specimens (McIntosh et al., 2005; Lucic et al., 2005; Leis et al., 2009; Hoenger and McIntosh, 2009). The great technological advances in the past decade have made it feasible to directly visualize the molecular architecture of organelles, cells and complex viruses as well as cellular dynamic events captured from in vivo cellular environment at close-to-molecular resolution (Medalia et al., 2002; He et al., 2003, 2008; Grunewald et al., 2003; Beck et al., 2004; Kurner et al., 2005; Nicastro et al., 2006; Al-Amoudi et al., 2007; Liu et al., 2008). The resolution attained by ET is rapidly approaching the molecular level (  2 nm), which

ENCYCLOPEDIA OF LIFE SCIENCES & 2010, John Wiley & Sons, Ltd. www.els.net

1

Electron Tomography

Resolution

3D imaging tools NMR X-ray crystallography Electron crystallography

Cryo-EM

Simple

0.1 nm

Protein

0.19−0.9 nm

In vitro

Averaging

0.45−2 nm

2−6 nm

Electron tomography

Confocal light microscopy

High Identical objects

Single particle

X-ray tomography

Best target

Unique object

20−50 nm

Nonaveraging

Protein complex virus Complex cell tissue

100−500 nm

Cell tissue organism

µm-mm

Human body

Low

Complex

In vivo

Medical CT

Figure 1

Comparisons of 3D bioimaging tools.

allows the study of the 3D organization of the structural components at a detail sufficient for the identification of macromolecular complexes, the analysis of their spatial distribution and their interactions in the native cellular context (Nickell et al., 2006; Brandt et al., 2009). ET offers a unique potential to bridge the gap between the low-resolution imaging techniques (Medical CT (computerized tomography), light microscopy and X-ray tomography; Le Gros et al., 2005) and the high-resolution structural techniques (X-ray crystallography, nuclear magnetic resonance, electron crystallography and single particle cryo-EM) (Figure 1). Those traditional highresolution structural techniques require the averaging of thousands to millions of identical copies of objects isolated from cells to achieve high resolution. However, the important spatial-temporal information of these objects in the living cells is totally lost by such in vitro studies. The proteins and their complexes need to be present in specific locations in response to the internal or external stimuli at certain circumstances in native cellular environments to perform their functions properly. Furthermore, the isolated proteins or complexes might also be unstable and subjected to conformational changes after the isolation. However, the protein expression levels might be either extremely low or high in cells, which would thus make them very difficult if not impossible to be isolated for structural study or to be traced individually by the low-resolution LM (100–500 nm). LM can only visualize fluorescently labelled proteins; most of the cellular structures and proteins without labels are invisible. In contrast to LM, ET can 2

visualize everything (e.g. vesicles, skeletons and proteins) in the cells at a few nanometer resolutions. For these cases, ET will play a unique role in dissecting and mapping out the molecular architectures and dynamic events from a series of ‘snapshots’ of the living cells at nanometer scale. In conclusion, ET will be a powerful tool to address the fundamental questions in cell biology and other biomedical sciences and, in fact, it has already provided exciting insights into a wide range of biological processes. See also: Macromolecular Structure Determination by X-ray Crystallography; Macromolecular Structure Determination: Comparison of Crystallography and NMR; Single Particle EM; Two-dimensional Electron Crystallography

Principle of ET ET is a computerized reconstruction method for generating the 3D map of the specimen from a series of 2D EM projection images recorded by a charge-coupled device (CCD) camera at different projecting orientations (Figure 2). The projection images for ET can be at any random orientations. One of the simplest strategies for collecting projection images is recording a series of projection images by simply rotating the specimen around a tilt axis at different angles, typically over a tilt range +/2608 or 708 and at small increments (1–28) (Figure 2a). Such projection images are called a single-axis tilt series. Sometimes, for better angular coverage, another tilt series is taken with the specimen rotated by 908 (called dual-axis tilt series). For

ENCYCLOPEDIA OF LIFE SCIENCES & 2010, John Wiley & Sons, Ltd. www.els.net

Electron Tomography

Fourier space

Real space e−

e−

e−

e−

FT at 0° at

−4 5°

Pr oj

5 −4 at

ec

n

tio

tio

n

ec °

Projection at 0°

Projection at 0° (b)

(a) Projection

1 Projection

°FT

at 4

oj Pr

Angular range [−50°,+50°]

45

5°

FT

at

3 Projections

5 Projections

15 Projections

IFT

25 Projections (c) Reconstruction

(d) Back-projection

Figure 2 Projection acquisition and tomographic reconstruction. (a) Projection. (Left) Tilt series are acquired by tilting the specimen around a tilt axis (here perpendicular to the sheet) at small increments over a limited tilt range. (Centre) Projection process of an object made up of four dots with different radii at 0o tilt. (Right) Acquisition of several projections of a tilt series from the object. (b) Assembling the Fourier transform (FT) of the three projections at 2458, 08 and 458 ((a), right) in the Fourier space, which is progressively filled as more projections are acquired and their FTs assembled. Afterwards, an inverse FT yields the reconstructed object. The ‘missing wedge’ is marked with grey triangles. (c) The reconstructed object under ideal conditions (sampling and covering). (d) Reconstruction of the object from the tilt series with back-projection. From left to right: the reconstruction with tilt series over the range [2508,+508] with 1, 3, 5, 15 and 25 projections. The influence of the limited tilt range and the angular sampling are apparent.

illustration purposes, Figure 2a shows a sketch of the acquisition of the projection images for a ‘specimen’ made up of four dots with different radii. Acquisition of the tilt-series of a biological specimen on a traditional manually operated EM is almost an impossible task. This is because the electron-dose sensitive biological specimens would be quickly destroyed under the high electron dose introduced by the manually searching, tracking, focusing and recording of images. The development of the automated data acquisition procedures on EM was crucial for acquiring a tilt series under low-dose imaging conditions (Koster et al., 1992; Dierksen et al., 1992, 1993). Even with the automated acquisition system, the projections in tilt series are not well aligned due to the imperfections of the mechanical and the electron optical systems. A more accurate alignment of the images is needed afterwards by computational procedures. The computational alignment is intended to mutually set the images to a common coordinate system by correction for the shifts and possible rotations and magnification changes. The standard procedure for the alignment starts at the preparation stage, where colloidal gold particles are

included in the biological sample to be used as electrondense fiducial markers. Once the tilt series is acquired, a coarse alignment is first performed by cross-correlation, a mathematic alignment approach. Then the coordinates of those markers are determined manually or semi-automatically throughout the images of the tilt series for further fine alignment. Using the coordinates of those markers, the images are then mutually aligned by means of a least square procedure aiming to minimize the alignment error as a function of the shifts, rotations and other parameters included in the alignment model (Frank, 2006). The reconstruction problem in ET is then to obtain the 3D structure of the specimen from the set of aligned images in the tilt series. The mathematical principles of tomographic reconstruction are based on the central section theorem, which states that the Fourier transform (FT) of a 2D projection of a 3D object is a central section of the 3D FT of the object (Frank, 2006). Based on this theorem, a series of real space projections corresponds to a series of central slices in Fourier space. Therefore, an inverse FT (IFT) of these aligned slices will generate the 3D reconstruction of the specimen (Figure 2b – c). One problem of this

ENCYCLOPEDIA OF LIFE SCIENCES & 2010, John Wiley & Sons, Ltd. www.els.net

3

Electron Tomography

Specimen Preparation Techniques in ET

approach is related to the interpolation in Fourier space. The standard method for tomographic reconstruction is weighted back-projection (WBP), which essentially is equivalent to the Fourier approach just described but working in real space (Frank, 2006). WBP distributes the specimen mass present in images evenly over computed back-projection rays. The sum of the back-projection rays from the images in the series represents the density distribution of the original object. The term ‘weighted’ in WBP derives from the fact that a weighting of projection images is necessary to properly represent the high-frequency information in the reconstruction. For illustration purposes, Figure 2d shows a sketch of the reconstructions from a set of projections with back-projection. The limited tilt range in ET results in a region empty of information in the Fourier space, so-called ‘missing wedge’ (or called ‘missing pyramid’ in the dual-axis tilt-series case), marked with grey triangles in Figure 2b. The resolution of the reconstruction is thus anisotropic. In real space, it produces artefacts as blurring of the spatial features in the beam direction, making some features appear as elongated in that direction and others not resolved at all (Figure 2d). A +/270 tilt range involves that 22% of the information is missing. The use of dual-axis series significantly reduces the missing information (down to 7% in the case of +/270 tilt range). Figure 2d shows the effect of the missing wedge in the reconstruction, which has the dots elongated in the beam direction. The angular sampling, that is, the interval between successive tilt images, is another point affecting the resolution of the reconstruction. The more the projection images, the better the angular sampling, and as a consequence the better the resolution. Figure 2d clearly shows the improvement of the reconstruction when the number of projections increases.

Bio-samples

Preparing biological specimens with high quality has been one of the major technical challenges in ET (McIntosh et al., 2005; Leis et al., 2009; Hoenger and McIntosh, 2009). ET is mainly applied for studying molecular architectures and cellular events in living cells and tissues, which makes the specimen preparation even more technically demanding than the other traditional cryo-EM techniques – more focused on studying macromolecular assemblies isolated from cells. To prepare biological specimens suitable for ET study, the following factors have to be taken into account: (1) the cellular structures and molecular distributions in living cells are very dynamic; (2) the aqueous biological specimen is unstable under the high vacuum and electron beam in EM; (3) the specimen thickness suitable for EM imaging is usually limited to 200–500 nm; (4) the image contrast of the unstained specimen (containing light elements such as carbon, nitrogen, oxygen and nitrogen) is extremely low and (5) identification of the proteins in cells with EM is difficult. Therefore, the specimen preparation methods must be designed to deal with these challenges: (i) the specimen needs to be immobilized by rapid cryo-fixation techniques; (ii) the aqueous biological specimen must be either dehydrated or frozen-hydrated; (iii) the bulk specimen has to be thin sectioned; (iv) better image contrast should be provided (e.g. thin enough or heavy metal staining) and (v) they should be compatible with other protein recognition or labelling techniques. In this section, we will discuss challenges and strategies for preparing specimens with high quality in ET (Figure 3). See also: Transmission Electron Microscopy: Preparation of Specimens

Fixation techniques

Water in samples Frozen-hydrated

zing

Small objects: <500 nm isolated proteins & organelles, viruses, bacterium, thin regions of cells

ee e fr

(<500 nm) Directly fixed on grid

Neg

ative

stain

ing

ree re f

u

ress

hp

Hig

(Bulk sample) (Fixing) Chem

ical f

Figure 3

4

Cryo-ET

g

Plun

zing

Large objects: >0.5−200 µm bacterium, cells, tissues, embryos, whole organisms

Data acquisition

ixatio

n

Dehydrated

RT-ET

Frozen-hydrated

Cryo-ET

Freeze-substitution

RT-ET

(30−300 nm) (Thin-sectioning) Dehydrated

Specimen preparation techniques for electron tomography. ENCYCLOPEDIA OF LIFE SCIENCES & 2010, John Wiley & Sons, Ltd. www.els.net

RT-ET

Electron Tomography

Traditionally, the dehydrated EM specimens have been commonly prepared either by negative staining of small objects or by positive staining of thin-sections of conventional chemically fixed bulk samples; however, those samples are subjected to serious fixation and staining artefacts, which limit the final resolution (Tokuyasu, 1973; Steinbrecht and Mu¨ller, 1987; McIntosh et al., 2005; Frank, 2006). The image contrast of unstained specimens is very low under the electron microscope. Staining the specimen with heavy metal salts, such as uranyl acetate, can produce high contrast and reduce serious collapsing (Frank, 2006). In the conventional chemical fixation, the fixatives might take a few seconds to 120 min to penetrate into the cells or tissues to crosslink the molecules in the specimen under room temperature, which makes it unsuitable for capturing the ultrastructures and fast cellular dynamic events. Owing to these limitations, the

(a)

200 nm

(b)

500 nm

200 nm

(d)

(c)

200 nm

(e)

low-resolution specimens prepared with such conventional methods are not commonly used in ET. See also: Immunoelectron Microscopy Currently, modern ET mainly works with highresolution specimens prepared with the rapid freezing techniques, such as plunge freezing (Dubochet et al., 1988) or high-pressure freezing (HPF) (Moor, 1987), which can preserve atomic to molecular resolution (Figure 3 and Figure 4). Rapid freezing techniques can capture cellular dynamic events and ultrastructures by physically immobilizing the molecules in the living cells within milliseconds without forming ice crystals, which are presumably the most accurate approaches for preparing high-quality biological specimens for ET. Plunge freezing is very effective for rapidly freezing small objects (5500 nm in the thinnest region) that can be directly deposited or grown on the hydrophilic carbon-coated EM grid by quickly submerging

100 nm

(f)

500 nm

(g)

100 nm

Figure 4 The ultrastructures of cells preserved by different specimen preparation techniques. (a) and (b) Well-preserved cellular organization and ultrastructures observed in ileal epithelial cells in suckling rats prepared with HPF/FSF – microtubules (pointed by red arrows) attached to tubular vesicles; ordered structures on the inner surface of large vesicles (pointed by blue arrow); membranes of vesicles and mitochondria are well-preserved; (unpublished results) (c) Molecular organizations of desmosomes and cell skeletons observed in mouse skin prepared with HPF/FSF (reproduced from He et al. (2003), with permission from the American Association for the Advancement of Science). (d) Vesicle with ordered structures observed in ileum of suckling rat prepared with HPF/FSF (unpublished result). (e) Tracing endocytosed 1.4 nm Nanogold particles by HPF/FSF-gold-enlarging techniques (He et al., 2008) (unpublished result). (f) Coated vesicles in thin region of fibroblast prepared by plunge freezing (reproduced from Koning et al. (2008) with permission from Elsevier). (g) High-resolution ultrastructures of desmosomes in human skin prepared by frozen-hydrated sectioning (reproduced with permission from Al-Amoudi et al. (2005b)). ENCYCLOPEDIA OF LIFE SCIENCES & 2010, John Wiley & Sons, Ltd. www.els.net

5

Electron Tomography

it into liquid ethane or liquid propane and extracting the heat at least 21048C s21 (Dubochet et al., 1988). However, for large objects, like cells and tissues (  10–300 mm), plunge freezing is no longer effective because the vitreous water formed around the outer layer of the specimen acts as a thermal barrier that drastically decreases the speed for extracting the heat generating in the inner portion of the specimen. Therefore, inevitably ice crystals would be formed in the specimen, hence destroying the ultrastructures. HPF can fully vitrify the bulk specimen (up to 200–300 mm) by reducing the ice crystallization speed drastically by briefly using  2050 bar hydrostatic pressure and followed with rapid freezing (Moor, 1987; McDonald, 2007). See also: Electron Cryomicroscopy The vitrified bulk specimen (0.5–300 mm) has to be either directly sectioned to produce 30–300 nm frozen-hydrated sections by cryo-sectioning (Al-Amoudi et al., 2003, 2007), or further processed with freeze-substitution fixation (FSF) and resin infiltration, embedding, polymerization for cutting 50–300 nm plastic thin sections (He et al., 2003, 2008; McIntosh et al., 2005), to make it suitable for ET study. Ideally, the thickness of the specimen should be smaller than the inelastic mean-free-path, 200 nm (350 nm) for 120 kV (300 kV) electrons, to avoid strong multiple scattering and inelastic scattering, which would strikingly degrade the image quality (Grimm et al., 1998). To date, cryo-ET of frozen-hydrated sections from bulk specimens is still technically demanding and difficult, because of the knife-induced mechanical distortion, such as crevasses, section compression and instability (Al-Amoudi et al., 2005a, 2007); It is worth exploring other alternative approaches, such as focused ion beam technology (Marko et al., 2007), to avoid the mechanical distortions. The majority of cryo-ET applications are limited to frozenhydrated natural thin objects, such as viruses, tiny bacteria, thin regions of cells, isolated proteins and organelles, which can be easily vitrified on the EM grid by the simple plunge freezing technique (Koning and Koster, 2009). Although frozen-hydrated samples can preserve close to atomic resolution faithfully (Figure 4f and g), however, the direct interpretable resolution achieved by cryo-ET is still limited to 3–6 nm due to the limitations of the electron dose sensitivity of unstained biological specimens and the low image contrast. FSF is an alternative approach for processing bulk frozen-hydrated specimens for achieving  5 nm resolution, by slowly dissolving and replacing the frozen cellular water by cold organic solvent (e.g. acetone, methanol) containing fixative (e.g. OsO4, uranyl acetate, glutaraldehyde, tanic acid) at approximately 2908C; at such low temperature other large molecules (e.g. proteins) and organelles are kept in immobilized state, whereas those small molecules (fixatives) can penetrate slowly into the cells and cross-link the proteins during the gradually temperature-raising processes. The HPF/FSF specimen has much less artefacts than those prepared by conventional chemical fixation (Steinbrecht and Mu¨ller, 1987); the  5 nm ultrastructures and cellular organizations are 6

quite reliable in the most specimens prepared with HPF/ FSF (Figure 4a – e), although they may contain other inevitable artefacts (such as staining artefacts, structure alterations and possibly some aggregations), which presumably affect the high resolution (53–5 nm). HPF/FSF prepared thin sections are easy to cut and very stable under electron beam and offer high image contrast, which makes the screening and imaging less challenging. HPF/FSF can also be easily cooperated with other high-resolution EM labelling techniques (Figure 4d; He et al., 2008; Staehelin and Kang, 2008). So far, ET of HPF/FSF prepared plastic section under room temperature is still the preferred approach for studying the cellular ultrastructures and the snapshots of dynamic cellular events of the bulk biological specimens at  5 nm resolution or better (McIntosh et al., 2005; Hoenger and McIntosh, 2009). Identification of the proteins in cells is extremely important; however, it is impossible to directly identify the small molecules by their shapes at 3–6 nm in noisy tomograms, even if it is possible to identify some large complexes (Leis et al., 2009). Therefore, it is highly desirable for the development of other high-resolution tracing (He et al., 2008), immunolabelling (Staehelin and Kang, 2008), clonable labels (Mercogliano and DeRosier, 2007; Diestra et al., 2009) and other chemical labelling techniques for identifying the proteins in the tomograms.

Strategies for Data Acquisition Modern cryo-EM systems configured for ET usually include several subsystems – illumination system (electron gun), computer-controlled specimen stage, vacuum system, optical system, imaging recording system (energy filter, CCD, film, etc.) and automated control system. The computer-controlled cryo-specimen stage, field emission gun (FEG), CCD and automated control system are essential for low-dose tilt-series data acquisition, which can reduce the dose consumption at 40–100 fold range compared to manual data collection. Several new technologies, such as spherical aberration corrector (Cs-corrector), direct electron detector (DDD), phase plate (Danev et al., 2009), improved lens and novel cryo-stages, will provide further improvements of the ET data quality in the near future (Hoenger and McIntosh, 2009). There are two schemes for ET data acquisition: keeping the specimen under room temperature for tilt-series data acquisition (RT-ET) and maintaining the specimen under frozen-hydrated condition (below 21658C) for tilt-series data collection (cryo-ET). Those heavy-metal salts stained and dehydrated specimens (Figure 3) are quite stable, offering high image contrast and insensitivity to the electron-dose damage. Thus, the specimen screening and imaging is much easier. Therefore, the tilt-series data acquisition is less technical challenging compared with that of the electron-dose sensitive frozen-hydrated specimen. The plastic thin sections (30–500 nm) usually need to be first stabilized by exposure to some electron dose

ENCYCLOPEDIA OF LIFE SCIENCES & 2010, John Wiley & Sons, Ltd. www.els.net

Electron Tomography

(  104 e nm22) before the tilt-series data collection (  2  105 e nm22). Usually, a tilt-series covers a +/2708 tilt range with constant increment of 18 (total: 141 projections); for dual-axis tilt-series, the specimen is rotated 908 for another tilt-series data collection. For the electron beam sensitive frozen-hydrated specimen, the specimen must be loaded into the cryo-holder, which maintains the temperature under 21658C to keep the vitreous state, for low-dose imaging under cryo-EM. Radiation damage can be drastically reduced by trapping the ionizing products (e.g. H2) and removing the heat generated by inelastic scattered electrons under cryo-EM condition. The data acquisition strategies and schemes have been proposed and tested (Grimm et al., 1998; Koster et al., 1992). To achieve 2–4 nm resolution, the specimen must be very thin (200– 350 nm, for 120–300 kV cryo-EM); the total electron-dose should be limited to 2500–6000 e nm22 for recording 100– 300 projections by a cosine scheme for allocating the exposure time for each projection, and ensuring the 2– 5 e pixel21 on the CCD to make the signal-to-noise ratio above the detectable limit (Grimm et al., 1998). The specimen thickness relative to the electron beam at high tilt angle turns out to be much thicker than in the untilted case (e.g. 2–3 times thicker at 60–708), which is far above the inelastic scattering mean-free-path; thus, the multipleand inelastic scattered electrons will drastically degrade the image quality. Using energy filter to remove those inelastic electrons can improve the image quality for thick

specimens or high tilt cases. However, the frozen-hydrated specimen should not be thicker than 500 nm, otherwise the attained resolution is much worse and there is no advantage over other methods. The image contrast of the biological specimen is very low and thus a higher underfocus value (2–6 mm) usually has to be set to enhance the image contrast, though at the expense of messing up the high resolution. The development of new phase plates (Danev et al., 2009) or the Cs-corrector (Haider et al., 2008) might be able to overcome this problem.

Interpretation of Tomographic Reconstructions Although reconstructed tomograms contain a wealth of information, their interpretation is complicated due to a number of factors: the artefacts due to the missing wedge, the low signal-to-noise ratio and the inherent biological complexity. Significant efforts are spent to facilitate the interpretation by several stages of post-processing of the tomograms: noise reduction, segmentation and identification of macromolecular complexes (Figure 5). Tomograms are significantly corrupted by noise, which precludes their visualization and analysis. Standard linear filtering techniques based on local averages, Gaussian kernels or low pass Fourier filtering succeeds in reducing

V

M

(a)

(b)

(c)

Figure 5 Interpretation of tomograms. (a) Vaccinia virus (top) denoised with anisotropic nonlinear diffusion (bottom; Cyrklaff et al., 2005). (b) Cellular context of tubular vesicles (right) modelled by manual segmentation (left; reproduced from He et al. (2008) with permission from Nature Publishing Group). (c) Atlas of the 80S ribosome (bottom) built from a section of S. cerevisiae (top) after detection with template matching, alignment and averaging (Pierson et al., 2008). ENCYCLOPEDIA OF LIFE SCIENCES & 2010, John Wiley & Sons, Ltd. www.els.net

7

Electron Tomography

the noise, but at the expense of blurring edges and features. This has led to the development of more sophisticated noise reduction techniques that preserve the structural features of interest. Anisotropic nonlinear diffusion is by far the standard noise reduction algorithm in the field (Frangakis and Hegerl, 2001; Fernandez and Li, 2003). It ensures preservation and enhancement of features of interest while cleaning the noise. Segmentation intends to decompose the tomogram into its structural components by identifying the sets of voxels that constitute them. Though tedious, manual segmentation is the simplest and the most common approach, which consists in that the user assigns the structural features using visualization tools (He et al., 2008). Several automatic or semi-automatic approaches based on sophisticated mathematical functions have been proposed in the field (reviewed in McIntosh, 2007). However, none of these methods has stood out as a general applicable method yet. Most of them are sensitive to noise, though their success is greatly improved if combined with the noise reduction methods, and in any case the results should always be supervised. Identification and interpretation of the individual molecules within crowded cellular environments in the tomograms at current direct achievable resolution (4–6 nm) is still extremely challenging but very important. There are several strategies aiming for molecular interpretation of tomograms: (1) direct identification of macromolecular configurations between known molecules (He et al., 2003; Al-Amoudi et al., 2007); (2) electron dense marker labelling (Mercogliano and DeRosier, 2007; He et al., 2008; Staehelin and Kang, 2008; Diestra et al., 2009); (3) templatematching techniques (Ortiz et al., 2006), followed by (4) subvolume averaging and classification (Beck et al., 2004; Nicastro et al., 2006; Bartesaghi et al., 2008; Liu et al., 2008; Winkler et al., 2009). Mapping out spatial-temporal distributions of the molecules in cellular environments at a few nanometer resolutions will make ET a very powerful bioimaging tool for cell biologists in the future.

Biological Applications As discussed earlier, ET offers many advantages over other bioimaging techniques for studying the flexible, unique ultrastructures and the spatiotemporal distributions of molecules in native cells or tissues at nanometer scale. Combined with genetic techniques and cell biology tools, ET can be used for answering a variety of biological questions, such as cell motility, development events, infectious and genetic diseases, etc. Currently, the biological applications of ET are mainly limited by the specimen preparation techniques and electron dose sensitivity of the biomaterials. Cryo-ET of rapidly frozen natural thin objects (5500 nm), that is viruses, tiny bacteria, thin regions of cells, isolated proteins and organelles, is less technically challenging and is quite easy to do. So far, about 80 such cryo-ET applications have been 8

performed successfully in many labs (Koning and Koster, 2009). Cryo-ET of thin sections of frozen-hydrated bulk specimens is technically demanding and suffers from several inevitable artefacts as discussed earlier; therefore, so far such applications are quite limited (Koning and Koster, 2009; Al-Amoudi et al., 2005a, 2007; Leis et al., 2009; Hoenger and McIntosh, 2009). The direct interpretable structures in the noisy tomograms reconstructed from the low-dose projection images are very limited; usually only the membranes, cell skeletons, large complexes and virus structures could be directly identified. Subvolume averaging and classification (Beck et al., 2004; Nicastro et al., 2006; Bartesaghi et al., 2008; Liu et al., 2008; Winkler et al., 2009) is required for interpreting some large rigid complexes at higher resolution. RT-ET of thin sections of HPF/ FSF prepared bulk samples is more feasible, and it has been widely used for studying snapshots of cellular ultrastructures and molecular events (e.g. cell skeletons, vesicles, molecular distributions, etc.), at approximately 4– 5 nm resolution (He et al., 2003, 2008; McIntosh et al., 2005; Staehelin and Kang, 2008; Hoenger and McIntosh, 2009). For the applications in cell biology, reliable electron dense marker labelling (Mercogliano and DeRosier, 2007; He et al., 2008; Staehelin and Kang, 2008; Diestra et al., 2009) will be very important for interpreting the cellular events and molecular arrangements in cells, since it is impossible or unreliable to identify most of the small proteins or complexes by their shapes at 2–6 nm resolution. A few illustrative biological examples are showed in Figure 5.

Illustrative Biological Examples Vaccinia virus is one of the largest (  360 nm) and most complex viruses whose 3D structure at 5 nm resolution was elucidated for the first time in 2005 by ET (Cyrklaff et al., 2005). Vaccinia virus is well known because it was successfully used for vaccination against the smallpox disease. The tomograms were computed with WBP from tilt series with images in the tilt range [2608, 608] at an interval of 28. After noise reduction and segmentation, interpretation was possible. The virus is a brick-shaped structure with dimensions of 360  270  250 nm, composed by an outer membrane and an inner core where the genetic material is located. A membrane that has a palisade of spikes surrounds the core. Figure 5a shows a slice of a tomogram containing a Vaccinia virion. In newborn rats, neonatal Fc receptor (FcRn) transfers IgG from milk to blood by apical-to-basolateral transcytosis across intestinal epithelial cells. ET was applied to make such epithelial transcytosis directly visible by using new labelling and detection methods to map out individual endocytosed nanogold-labelled IgG Fc within transport vesicles at 4 nm resolution (He et al., 2008). The specimen was prepared with HPF/FSF followed by FSF-based gold enlargement procedures, then embedded in resin and polymerized. A 180-nm plastic section was cut with ultramicrotome for imaging under a 120-kV EM at room

ENCYCLOPEDIA OF LIFE SCIENCES & 2010, John Wiley & Sons, Ltd. www.els.net

Electron Tomography

temperature. The tomogram was reconstructed with WBP from a dual-axis tilt series of projection images, with tilt range [2708, 708] at an interval of 18. The individual nanogold-labelled IgG Fc can be identified in the projection images, and traced to different types of vesicles in a 3D model generated from the tomogram (Figure 5b). Visualization of the macromolecular organization of thick eukaryotic organisms such as yeast Saccharomyces cerevisiae is possible thanks to cryo-sectioning techniques. These techniques allowed the extraction of thin sections (50 nm) of S. cerevisiae to perform ET (Pierson et al., 2008). Tomograms were reconstructed with WBP from tilt series in the angular range [2708, 708] at steps of 1.58. Ribosomes in the cell environment were automatically detected by means of template matching techniques, using as a template a previous ribosome density map obtained by single particle EM. The ribosomes were aligned and an average was computed. Finally, the ribosomes were put back into the 3D space in their location with the determined orientations, thus building an atlas of the 80S ribosome of S. cerevisiae (Figure 5c).

Conclusions The advances in electron tomography in recent years have opened a window for the direct visualization of the molecular organization of cells at nanometer scale in 3D. ET is rapidly emerging as a powerful tool filling the resolution gap between the high-resolution structural techniques (X-ray crystallography, NMR (nuclear magnetic resonance) and single particle cryo-EM) and the lowresolution bioimaging tools (X-ray tomography and confocal light microscopy). In conjunction with the advances in specimen preparation techniques, EM hardware, novel labelling techniques and imaging processing techniques, ET will play a crucial role in dissecting the cellular structures and functions at a new horizontal level.

Acknowledgements The authors thank the financial support of the grants SBIC-SSCC R-154-000-377-305, AcRF R-154-000-402112, MCI-TIN2008-01117 and JA-P06-TIC-01428.

References Al-Amoudi A, Diez DC, Betts MJ and Frangakis AS (2007) The molecular architecture of cadherins in native epidermal desmosomes. Nature 450: 832–837. Al-Amoudi A, Dubochet J, Gnaegi H, Lu¨thi W and Studer D (2003) An oscillating cryo-knife reduces cutting-induced deformation of vitreous ultrathin sections. Journal of Microscopy 212: 26–33. Al-Amoudi A, Dubochet J and Norle´n L (2005b) Nanostructure of the epidermal extracellular space as observed by cryo-

electron microscopy of vitreous sections of human skin. Journal of Investigative Dermatology 124: 764–777. Al-Amoudi A, Studer D and Dubochet J (2005a) Cutting artefacts and cutting process in vitreous sections for cryo-electron microscopy. Journal of Structural Biology 150: 109–121. Bartesaghi A, Sprechmann P, Liu J et al. (2008) Classification and 3D averaging with missing wedge correction in biological electron tomography. Journal of Structural Biology 162: 436– 450. Beck M, Forster F, Ecke M et al. (2004) Nuclear pore complex structure and dynamics revealed by cryoelectron tomography. Science 306: 1387–1390. Brandt F, Etchells SA, Ortiz JO et al. (2009) The native 3D organization of bacterial polysomes. Cell 136: 261–271. Cyrklaff M, Risco C, Fernandez JJ et al. (2005) Cryo-electron tomography of vaccinia virus. Proceedings of the National Academy of Sciences of the USA 102: 2772–2777. Danev R, Glaeser RM and Nagayama K (2009) Practical factors affecting the performance of a thin-film phase plate for transmission electron microscopy. Ultramicroscopy 109: 312–325. Dierksen K, Typke D, Hegerl R and Baumeister W (1993) Towards automatic electron tomography II. Implementation of autofocus and low-dose procedures. Ultramicroscopy 49: 109–120. Dierksen K, Typke D, Hegerl R, Koster AJ and Baumeister W (1992) Towards automatic electron tomography. Ultramicroscopy 40: 71–87. Diestra E, Fontana J, Guichard P, Marco S and Risco C (2009) Visualization of proteins in intact cells with a clonable tag for electron microscopy. Journal of Structural Biology 165: 157–168. Dubochet J, Adrian M, Chang JJ et al. (1988) Cryo-electron microscopy of vitrified specimens. Quarterly Reviews of Biophysics 21: 129–228. Fernandez JJ and Li S (2003) An improved algorithm for anisotropic nonlinear diffusion for denoising cryo-tomograms. Journal of Structural Biology 144: 152–161. Frangakis AS and Hegerl R (2001) Noise reduction in electron tomographic reconstructions using nonlinear anisotropic diffusion. Journal of Structural Biology 135: 239–250. Frank J (ed.) (2006) Electron Tomography: Methods for ThreeDimensional Visualization of Structures in the Cell. Berlin: Springer. Grimm R, Singh H, Rachel R et al. (1998) Electron tomography of ice-embedded prokaryotic cells. Biophysics Journal 74: 1031– 1042. Grunewald K, Desai P, Winkler DC et al. (2003) Threedimensional structure of herpes simplex virus from cryoelectron tomography. Science 302: 1396–1398. Haider M, Mu¨ller H, Uhlemann S et al. (2008) Prerequisites for a Cc/Cs-corrected ultrahigh-resolution TEM. Ultramicroscopy 108: 167–178. He W, Cowin P and Stokes DL (2003) Untangling desmosomal knots with electron tomography. Science 302: 109–113. He W, Ladinsky MS, Huey-Tubman KE et al. (2008) FcRnmediated antibody transport across epithelial cells revealed by electron tomography. Nature 455: 542–546. Hoenger A and McIntosh JR (2009) Probing the macromolecular organization of cells by electron tomography. Current Opinion in Cell Biology 21: 89–96.

ENCYCLOPEDIA OF LIFE SCIENCES & 2010, John Wiley & Sons, Ltd. www.els.net

9

Electron Tomography

Koning RI and Koster AJ (2009) Cryo-electron tomography in biology and medicine. Annals of Anatomy 191: 427–445. Koning RI, Zovko S, Ba´rcena M et al. (2008) Cryo electron tomography of vitrified fibroblasts: microtubule plus ends in situ. Journal of Structural Biology 161: 459–468. Koster AJ, Chen H, Sedat JW and Agard DA (1992) Automated microscopy for electron tomography. Ultramicroscopy 46: 207– 227. Kurner J, Frangakis AS and Baumeister W (2005) Cryo-electron tomography reveals the cytoskeletal structure of Spiroplasma melliferum. Science 307: 436–438. Le Gros MA, McDermott G and Larabell CA (2005) X-ray tomography of whole cells. Current Opinion in Structural Biology 15: 593–600. Leis A, Rockel B, Andrees L and Baumeister W (2009) Visualizing cells at the nanoscale. Trends in Biochemical Sciences 34: 60–70. Liu J, Bartesaghi A, Borgnia MJ, Sapiro G and Subramaniam S (2008) Molecular architecture of native HIV-1 gp120 trimers. Nature 455: 109–113. Lucic V, Forster F and Baumeister W (2005) Structural studies by electron tomography: from cells to molecules. Annual Review of Biochemistry 74: 833–865. Marko M, Hsieh C, Schalek R, Frank J and Mannella C (2007) Focused-ion-beam thinning of frozen-hydrated biological specimens for cryo-electron microscopy. Nature Methods 4: 215–217. McDonald K (2007) Cryopreparation methods for electron microscopy of selected model systems. In: McIntosh JR (ed.) Cellular Electron Microscopy in ‘Methods in Cell Biology’ (Series eds: Wilson L and Matsudaria P), vol. 79, pp. 23–56. San Diego: Academic Press. McIntosh JR (ed.) (2007) Cellular Electron Microscopy, Volume 79 of Methods in Cell Biology. San Diego: Academic press. McIntosh JR, Nicastro D and Mastronarde D (2005) New views of cells in 3D: an introduction to electron tomography. Trends in Cell Biology 15: 43–51.

10

Medalia O, Weber I, Frangakis AS et al. (2002) Macromolecular architecture in eukaryotic cells visualized by cryoelectron tomography. Science 298: 1209–1213. Mercogliano CP and DeRosier DJ (2007) Concatenated metallothionein as a clonable gold label for electron microscopy. Journal of Structural Biology 160: 70–82. Moor H (1987) Theory and practice of high pressure freezing. In: Steinbrecht RA and Zierold K (eds) Cryotechniques in Biological Electron Microscopy, pp. 175–191. Heidelberg: Springer. Nicastro D, Schwartz C, Pierson J et al. (2006) The molecular architecture of axonemes revealed by cryoelectron tomography. Science 313: 944–948. Nickell S, Kofler C, Leis AP and Baumeister W (2006) A visual approach to proteomics. Nature Reviews Molecular Cell Biology 7: 225–230. Ortiz JO, Forster F, Kurner J, Linaroudis AA and Baumeister W (2006) Mapping 70S ribosomes in intact cells by cryoelectron tomography and pattern recognition. Journal of Structural Biology 156: 334–341. Pierson J, Fernandez JJ, Bos E et al. (2008) Visualization of the 80s ribosome in situ using cryo-electron tomography of vitreous sections. Proceedings of the European Microscopy Congress 2008, Vol. 3: Life Science, pp. 89–90, Berlin, Springer. Staehelin LA and Kang BH (2008) Nanoscale architecture of endoplasmic reticulum export sites and of Golgi membranes as determined by electron tomography. Plant Physiology 147: 1454–1468. Steinbrecht RA and Mu¨ller M (1987) Freeze-substitution and freeze-drying. In: Steinbrecht RA and Zierold K (eds) Cryotechniques in Biological Electron Microscopy, pp. 149–172. Berlin: Springer. Tokuyasu KT (1973) A technique for ultracryotomy of cell suspensions and tissues. Journal of Cell Biology 57: 551–565. Winkler H, Zhu P, Liu J et al. (2009) Tomographic subvolume alignment and subvolume classification applied to myosin V and SIV envelope spikes. Journal of Structural Biology 165: 64–77.

ENCYCLOPEDIA OF LIFE SCIENCES & 2010, John Wiley & Sons, Ltd. www.els.net