by Hall, Hartwieg and Nguyen
Electron tomography has been in development for 20 years, and as
the availability of this equipment increases, it is becoming one of the leading EM techniques for C. elegans. Electron tomography
became feasible with the advent of computer-controlled stages on the newer model
TEMs, more reliable CCD cameras and the development of powerful analytical
software for the creation and annotation of 3D tomograms (Franck, 2006). Just as the nematode seems to be ideally suited for the high-pressure freezing technique due to its small size, the nematode is also well suited for electron tomography due to its simplicity and the wealth of anatomic information already available.
Electron tomography techniques were pioneered for intact C. elegans by John White, Thomas Muller-Reichert, Richard McIntosh, Kent McDonald, and their colleagues. Early objects of interest have included the mitotic spindle in the early embryo (O’Toole et al., 2003; O’Toole and Muller-Reichert, 2008; Pelletier et al., 2006), the ultrastructure of the mitochondrion (Kanazawa et al., 2008), and the shape and organization of ribosomes in the rough endoplasmic reticulum (Leapman et al., 2004). The Hall lab has recently been introduced to this technology through the auspices of the New York Structural Biology Center, with technical help from KD Derr and William Rice. Many subcellular structures within C. elegans are immediately of interest in both wild type and mutant backgrounds, such as intercellular junctions, synapses, sensory endings, and apical structures at the luminal border of the intestine (Stigloher et al., 2011; Topalidou et al., 2012).
Animals preserved by HPF followed by freeze substitution provide the ideal specimens for electron tomography. It is also possible to collect frozen thin sections to be viewed on a cryo-stage within the TEM, without thawing or plastic embedment. This is useful for 3D studies of molecular complexes. Fine details of isolated molecular complexes can also be viewed by electron tomography after spreading them on thin films (Ben-Harush et al., 2009). Alternately, intact fast frozen tissue can be embedded in plastic for viewing in semithin sections (80–500 nm thick). In the latter case, the limits on viewing may depend on the strength of the electron beam needed to penetrate through the section at higher tilt angles.
TEMs with a field emission gun (FEG) offer higher power electrons for use with thicker specimens (from 200 kV up to three million kV). The examples shown below were produced on an FEI Tecnai20 electron microscope equipped using a 200 kV FEG, viewing sections at thicknesses from 80 to 250 nm (ETFIG 1).
The SerialEM program was used to control the microscope stage’s X,Y position, the degree of tilt, and the exact focus, collecting and organizing about 100 sequential EM images while the stage is gradually tilted from –70° to + 70° (Mastronarde, 2005). Those sequential images were then processed using the Protomo software package to create the initial tomogram (Winkler and Taylor, 2006). For some applications, the semithin sections may be coated with gold beads on one side as alignment markers in calibrating the exact relations between all objects within the tilt series for tomogram production (Franck, 2006) (ETFIG 2). In other cases, information based upon structural features inherent inside the tissue itself (e.g., ‘‘markerless alignment’’) has proven sufficient to provide ‘‘weighted back projections’’ in 3D space (ETFIG 3).
ETFIG 2 ETFIG 3
Using Fourier transforms to compare objects from hundreds of different angled views through the same semithick section, it becomes possible to separately resolve microscopic details even when some objects lie ‘‘behind’’ one another within the EM section. One begins by collecting many images around one or two axes of tilt for a single image space, then computing a tomographic reconstruction for each tilt axis, and combining those single axis tomograms into a dual axis tomogram having even higher resolution (ETFIG 2 and ETFIG 3) (Franck, 2006). The ultimate limits on resolution can be difficult to measure exactly, and may be subject to both specimen quality and total scope magnification. However, this technology allows one to achieve much better views of objects in the 3–30 nm size range (smaller than the typical thin section thickness) that were previously obscure in single sections. Furthermore, by combining tomograms across serial sections, one can also build accurate 3D models of larger volumes that are large enough to reconstruct whole organelles or perhaps
even whole cells in exacting detail. The precision of these models exceeds any previous 3D model based upon standard thin section reconstructions.
A dual tilt tomogram from one semithick section consists of a data model that can be viewed with the appropriate software from multiple angles, allowing identification and annotation of structures within the three-dimensional volume. The McIntosh laboratory at Boulder has produced free software called IMOD that is very useful for several key steps in data analysis (Kremer et al., 1996). Several commercial annotation packages are also available, including Amira and Maya (see Computer based tools). After annotation, the 3dmod module of IMOD provides only a limited roster of visualization tools to display the model. For that reason, we have been moving the annotated 3dmod models into Amira (see WormAtlas Movie Gallery). Other available visualization packages include Blender and Cinema 4D. The 3dmod module allows one to draw objects in multiple colors, to model objects as lines, spheres, or to trace open or closed contours having more complex shapes. Better tools are still needed for modeling more diffuse features, such as networks of fine filaments (basal lamina, terminal web, actin networks, etc), or to allow edges of objects to be defined accurately by ‘‘thresholding.’’ Properties of the electron tomogram data (noisiness, poor contrast) make these diffuse features difficult to address at present.
Special TEM specimen holders are necessary in order to collect images up to very high tilt angles (+/–70°) for tomography. Going to extremely high tilts reduces the ‘‘missing wedge’’ of data space for Fourier analysis and improves final resolution.
The IMOD program can be difficult to learn, but this program is widely used, and new features are still being added and improved. The program is supplied as freeware from the University of Colorado in Boulder. The McIntosh lab offers a course about once per year to introduce it to new users and an online user group is availble to answer questions.
One of the exciting features of electron tomography is that one may begin to visualize objects that were not apparent in any standard TEM view, while looking ‘‘live’’ on the microscope. This means that finding the features of interest is not always straight forward, and it will be worthwhile to try several places in the search for the ‘‘right’’ locale. But one must also be very careful not to spend any time viewing poorly frozen tissues, since any freeze damage will only make bigger problems in interpretation as one moves from standard TEM images to tomograms.
ETFIG 1: Philips Technai 20 transmission electron microscope is used for electron tomography. This microscope (available from FEI, Portland, OR) has an excellent tilting goniometer to control views of the sample at many angles, and a bottom-mounted digital camera for data collection. The microscope can be programmed to collect images for many hours automatically, achieving the correct focus, magnification, tilt, and image framing on its own, and by comparison to previously collected images.
Click pictures for new window with figure and legend, click again for high resolution image
ETFIG 2: Electron tomogram of a rectal muscle. A dual axis electron tomogram shows diffuse elements of the basal lamina and within the cuticle, including layers of filaments spanning the lamina, and long thin filaments anchoring the depressor muscle cell to the rectal cuticle. Ventral is to the left. This is not an electron micrograph. The tomogram has been resliced at a favorable angle. Such fine features are never seen well in standard thin sections. HPF/FS sample. Image capture by Leslie Gunther, FEI Technai20 TEM. Tomogram calculated using gold particles as reference marks. This tomogram can also be viewed as a set of movies.
ETFIG 3: Electron tomogram of a touch dendrite. A. Raw image of ALM dendrite within the dual axis electron tomogram, face on, prior to any annotation. This is a mathematical model built after Fourier analysis, not a micrograph. Note that the microtubules are clustered near a large vesicle, which is likely cargo to be moved by microtubule-based motors inside the dendrite. HPF/FS sample. Image capture by KD Derr (NYSBC), Technai20 TEM. Tomogram calculated using weighted back projection from internal features by Bill Rice (NYSBC). Scale bar, 100 nm. B. Hand-annotated elements of the ALM dendrite and its neighborhood. Note that the microtubule bundle bends to accommodate passage of the vesicle. Purple, microtubules; red, ribosomes; yellow, large vesicle; blue, small vesicle; green, ALM plasma membrane. IMOD annotation by Kristin Politi (see Topalidou et al., 2012). This tomogram can also be viewed as a movie.
Ben-Harush, K., Wiesel, N., Frenkiel-Krispin, D., Moeller, D., Soreq, E., Aebi, U., Herrmann, H., Medalia, O., Collins, F.S. and Goldman, A.E. 2009. The supramolecular organization of the C. elegans nuclear lamin filament. J. Mol. Biol. 386: 1392–1402. Abstract
Franck, J. 2006. Electron Tomography: Methods for the Three-Dimensional Visualization of Structure in the Cell, 2nd Edition. Springer, Berlin, p. 455. Abstract
Hall, D.H., Hartweig, E. and Nguyen, K.C.Q. 2012. Modern electron microscopy methods for C. elegans. Methods Cell Biol. 107: 93-149. Abstract
Kanazawa, T., Zappaterra, M.D., Hasegawa, A., Wright, A.P., Newman-Smith, E.D., Buttle, K.F., McDonald, K., Mannella, C.A. and van der Bliek, A.M. 2008. The C. elegans Opa1 homologue EAT3 is essential for resistance to free radicals. PLoS Genet. 4: e1000022. Article
Kremer, J.R., Mastronarde, D.N. and McIntosh, J.R. 1996. Computer Visualization of Three-Dimensional Image Data Using IMOD. J. Struct. Biol. 116: 71–76. Abstract
Leapman, R.D., Kocsis, E., Zhang, G., Talbot, T.L. and Laquerriere, P. 2004. Three-dimensional distributions of elements in biological samples by energy-filtered electron tomography. Ultramicroscopy 100: 115–125. Abstract
Mastronarde, D.N. 2005. Automated electron microscope tomography using robust prediction of specimen movements. J. Struct. Biol. 152: 36–51. Abstract
O’Toole, E.T., McDonald, K.L., Mantler, J., McIntosh, J.R., Hyman, A.A. and Muller-Reichert, T. 2003. Morphologically distinct microtubule ends in the mitotic centrosome of C. elegans. J. Cell Biol. 163: 451–456. Article
O’Toole, E.T. and Muller-Reichert, T. 2009. Electron tomography of microtubule end-morphologies in
C. elegans embryos. Meth. Mol. Biol. 545: 135–144. Article
Pelletier, L., Ozlu, N., Hannak, E., Cowan, C., Habermann, B., Ruer, M., Muller-Reichert, T. and Hyman,
A.A. 2006. The C. elegans centrosomal protein SPD-2 is required for both pericentriolar material recruitment and centriole duplication. Curr. Biol. 14: 863–873. Article
Stigloher C., Zhan, H., Zhen, M., Richmond, J. and Bessereau, J.L. 2011. The presynaptic dense projection of the Caenorhabiditis elegans cholinergic neuromuscular junction localizes synaptic vesicles at the active zone through SYD-2/Liprin and UNC-10/RIM-dependent interactions. J. Neurosci. 31:4388-96. Article
Topalidou, I., Keller, C., Kalebic, N., Nguyen, K.C., Somhegyi, H., Politi, K.A., Heppenstall, P., Hall, D.H. and Chalfie, M. 2012. Genetically separable functions of the MEC-17 tubulin acetyltransferase affect microtubule organization.
Curr. Biol. 22:1057-65. Article
Winkler, H. and Taylor, K.A. 2006. Accurate marker-free alignment with simultaneous geometry determination and reconstruction of tilt series in electron tomography. Ultramicroscopy 106: 240–254. Abstract