Electron cryotomography: a new view into microbial ultrastructure

Abstract

Electron cryotomography (ECT) is an emerging technology that allows thin samples such as small bacterial cells to be imaged in 3-D in a nearly native state to “molecular” (∼4 nm) resolution. As such, ECT is beginning to deliver long-awaited insight into the positions and structures of cytoskeletal filaments, cell wall elements, motility machines, chemoreceptor arrays, internal compartments, and other ultrastructures. Here we briefly explain ECT, review its recent contributions to microbiology, and conclude with a discussion of future prospects.

Introduction

After over a century of intense microbiological research, basic metabolism is now largely understood and the complete sequences of nearly 1000 bacterial species are available. Light microscopy has shown us the marvelous diversity of microbial cell shapes and, more recently through fluorescent tags, the approximate locations of numerous proteins and gene loci within those shapes. Higher resolution techniques like X-ray crystallography and NMR spectroscopy have produced atomic models of thousands of important macromolecules. Given this remarkable progress, our persistent ignorance about many of even the most fundamental microbial “cell biological” processes is surprising. We still don't know, for instance, how bacteria generate and maintain their characteristic shapes, establish polarity, organize their genomes, segregate their chromosomes, divide, and in some cases move. The problem in large part has been the lack of a technology that allows us to “see” these processes and their key underlying structures: the resolution gap between light microscopy and atomic models is the realm of electron microscopy (EM), but the “traditional” EM specimen preparation methods of chemical fixation, plastic embedment, sectioning, and staining that have been used in the past failed to preserve the needed details. In just the last few years, the emergence of ECT has opened a new window into microbial ultrastructure [1,2] that promises to revolutionize bacterial cell biology.

Electron cryotomography

Briefly, in ECT suspensions of intact cells are spread into thin films across EM grids, plunge-frozen in liquid ethane, transferred into an electron cryomicroscope, and imaged iteratively while being tilted incrementally. Plunge-freezing prevents ice crystallization, immobilizing the proteins and other cellular structures in their native states and locations. Recording a series of images at different angles allows a full, three-dimensional (3-D) reconstruction (“tomogram”) of the specimen to be calculated. Thus ECT produces 3-D images of intact cells in a nearly native state, without the artifacts of fixation, dehydration, plastic-embedment, sectioning, or staining. Resolutions of a few nanometers are typically achieved, though the resolution and interpretability are critically influenced by specimen thickness and crowdedness. Thus for slender bacteria, the positions and even orientations of individual ribosomes can be detected, but larger cells such as Bacillus subtilis are prohibitively thick. For thicker cells, biofilms, or tissues, the alternative specimen preparation technique “high-pressure freezing” rapidly freezes samples up to 0.5 mm in thickness under pressures of ∼2000 bar to suppress ice crystal formation. While still far from routine, techniques are being developed to “cryosection” such frozen-hydrated material, allowing it to also be imaged in a nearly-native, vitreous state [3].

One of the challenges in ECT is to unambiguously identify structures of interest in tomograms. Unfortunately there are as yet no “GFP-like,” genetically-encodable markers. Instead, some structures have been identified by imaging mutant cells where candidate proteins have been overexpressed, depleted, or deleted, and noting which structures are affected [4,5]. Correlative light and cryoelectron microscopy techniques have also been developed in which fluorescence optical images and cryo-EM images are recorded of the same cell [6-9]. Finally, structures have been identified by their structural “signatures” such as their shapes [10], subunit spacings [11], or bilayer structure [12,13].

Two years ago we reviewed the contributions and potential of ECT to microbiology [1] and noted that while the first paper reporting cryotomograms of prokaryotic cells was published just ten years ago [14], by 2006 there had been thirteen others from 4 different groups. As predicted, the activity in the field in the last two years has rapidly expanded: in just the last two years an additional 20 papers have been published reporting work by 10 different groups. Here we review those papers, including the increasingly promising applications of cryosectioning. The results cover diverse subcellular structures including cytoskeletal filaments, cell envelopes and cell walls, surface appendages, internal compartments, and the nucleoid as well as their functions in the processes of cell division, motility, chemotaxis, pathogenity, and cell-cell interaction.

Cytoskeleton

The discovery of the bacterial cytoskeleton is a fascinating tale, considering that not long ago cytoskeletal proteins were thought to exist exclusively in eukaryotic cells. It is now known that there are homologs of all three main classes of eukaryotic cytoskeletal proteins (tubulin, actin, and intermediate filaments) in bacteria (FtsZ, MreB, and CreS, respectively). Indeed it is now apparent that the bacterial cytoskeleton is not only complex, but probably ubiquitous, and is involved in numerous cellular functions [15]. Each filament's subcellular localization has been investigated mainly by immuno-EM, immunofluorescence and GFP-based light microscopy. For example, FtsZ was first localized to the mid-cell division site by immuno-EM [16] and then visualized as a ring structure by fluorescence light microscopy [17]. Traditional EM (involving chemical fixation, dehydration, embedding and staining) failed to reveal actual filaments, however, probably because the polymers were dynamic and depolymerized somehow during specimen preparation. Applying whole-cell ECT on dividing C. crescentus cells, we visualized short, arc-like filaments close to the cytoplasmic membrane at the constriction site (Fig. 1A and Supplementary Movie) [5]. The filaments' position, orientation, time of appearance, and correlation in numbers and length to expression levels and stabilities in several mutants all showed they were in fact composed of FtsZ. The 3-D images of FtsZ filaments in wild-type cells strongly supported a force-generating role of FtsZ filaments during cell division first proposed by Erickson [18].

Figure 1. Cytoskeletal filaments.

(A) 8-nm tomographic slice of a dividing Caulobacter crescentus cell, showing the filaments in cross-section (small dark dots near the center of the circles). Since the reconstruction can be visualized in 3-D, these filaments can be segmented (Inset). Filaments (red), outer membrane (yellow), and inner membrane (blue). Scale bar 100 nm. Adapted from [5] with permission from Nature Publishing Group. (B) Vitreous cryo-section of an E. coli cell carrying a high-copy ParMRC-bearing plasmid. The nominal thickness of the section is 50 nm. A bundle of filaments are shown in the circle. Top inset: enlarged view. Bottom inset: illustration showing the cellular plane (grey) seen in the image. Scale bar 100 nm. Adapted from [11] with permission from the American Association for the Advancement of Science.

While numerous other filament bundles have now been visualized in cryotomograms of bacteria (those reviewed in [1], [19-21]), so far only a few are identified. Recently Salje et al. cryosectioned E. coli cells over-expressing ParM, the actin-like protein that segregates R1 plasmids, and saw filament bundles (Fig. 1B) [11]. ECT of vitreous sections containing the filament bundles revealed a diffraction pattern similar to that derived from in vitro purified and assembled ParM protofilaments, thus identifying the filament bundles as ParM. In wild-type (low-copy number) strains, the authors observed bundles of three to five intracellular ParM filaments localized within the periphery of the nucleoid, strongly supporting the model that only one filament is needed to separate each plasmid pair.

Cell Envelope and Peptidoglycan

Bacterial cell envelopes shield the cytoplasm and genetic material from the environment. The mycobacterial cell envelope is of paramount medical interest because it constitutes a permeability barrier for antibiotics and is essential for virulence. Hoffmann et al. applied both cryosectioning and whole-cell ECT to investigate the cell envelope structure of Mycobacterium smegmatis, Mycobacterium bovis BCG, and Corynebacterium glutamicum [12]. With close-to-focus images and tomograms showing paired individual leaflets, they proved that the mycobacterial outer layer is indeed a bilayer structure, which they labeled the “mycobacterial outer membrane” (Fig. 2A). By further imaging a mycolic acid-deficient C. glutamicum mutant, the authors observed the absence of the outer membrane and concluded that mycolic acids are constituents of the outer membrane. In a similar study, Zuber et al. cryosectioned the same species and mutants and obtained similar results, although the authors proposed a different model of the mycobacterial outer membrane [13]. These studies highlighted how important it can be to preserve specimens in their native (in this case frozen-hydrated) state.

Figure 2. Cell envelope.

(A) Vitreous cryo-section of a Mycobacterium bovis BCG cell. The section has a nominal thickness of 35 nm. Scale bar 100 nm. Adapted from [12] with permission. © 2008 by The National Academy of Sciences of the USA. (B) 10-nm tomographic slice through an E. coli sacculus. The individual glycan strand densities lie in the plane of the sacculus, roughly perpendicular to the saccular long axis. Scale bar 500 nm. Adapted from [22] with permission. © 2008 by The National Academy of Sciences of the USA.

The mechanical strength of bacterial cell walls arises from a molecular “bag-like” exoskeleton of peptide-crosslinked glycan strands called the sacculus. While the chemical composition and subunit structure of peptidoglycan has been known for decades, the overall architecture of the sacculus was unclear. Two fundamentally different models of 3-D peptidoglycan organization had been suggested, namely the “Layered” and “Scaffold” models. Gan et al. applied ECT to intact sacculi purified from two Gram-negative bacteria, E. coli and C. crescentus [22]. In cryotomograms of both preparations, sacculi were seen to contain a single peptidoglycan layer in which the individual glycan strands were oriented in the plane of the sacculus perpendicular to the long axis of the cell and spaced approximately 5-8 nm apart (Fig. 2B). This observation ruled out the “Scaffold” model and instead established a “Disordered, Circumferential, Layered” model.

Motility and Surface Appendages

Bacterial cells exhibit wonderfully diverse modes of motility including swimming, swarming, twitching, and gliding. The most studied mode of motility, swimming, is driven by one or more flagella which are rotated by a membrane-embedded motor. In spirochetes, the flagella are located in the periplasmic space between the outer and cytoplasmic membranes. Three spirochetes have now been imaged with ECT. In a recently cultured termite gut spirochete, Triponema primitia, Murphy et al. revealed novel structures including bowls, arches, fibrils, and two layers of peptidoglycan sandwiching the flagella between them (Fig. 3A) [23]. These results, combined with the earlier ECT structure of the flagellar motor from the same species [24], are consistent with the “rolling cylinder” model of spirochete motility originally proposed by Berg [25]. ECT of the pathogenic spirochete Treponema denticola revealed periplasmic flagella, cytoplasmic filaments and plate-like structures, and a patella-like periplasmic structure at one polar tip [20]. The cytoplasmic filaments, earlier suggested as CfpA (a unique protein in spirochetes) [26], were visible in both wild-type and aflagellate strains. Finally, Charon et al. reported the structure of the Lyme disease spirochete Borrelia burgdorferi [27]. In contrast to the stacked bundle observed previously by traditional EM [28], by ECT it was seen that the periplasmic flagella adopt a flat-ribbon configuration.

Figure 3. Surface appendages.

(A) 10-nm tomographic slice through a Treponema primitia cell, showing novel surface appendage structures. Scale bar 200 nm. Adapted from [23] with permission from Blackwell Publishing Ltd. (B) 3-nm tomographic slice through the middle of a Flavobacterium johnsoniae cell, showing the outer membrane (OM), cytoplasmic membrane (CM), cell surface filaments (F) and added latex spheres (S). Inset: enlarged view of the boxed area showing the peptidoglycan layer (P) and a patch (A) underneath the outer membrane. Scale bar 300 nm. Adapted from [29] with permission from the American Society of Microbiology.

The gliding bacterium Flavobacterium johnsoniae was shown to possess tufts of ∼5-nm-wide cell surface filaments emanating from the inner surface of the outer membrane (Fig. 3B) [29]. These filaments were absent in a non-motile gldF mutant cell but were restored in the same mutant complemented with plasmid-encoded GldF, a component of a putative ATP-binding cassette transporter. The cell surface filaments are unlikely to be composed of any Gld proteins, however, since the Gld proteins are known to localize inside the outer membrane.

Chemotaxis

Bacteria sense nutrient gradients through an array of proteins that form a complex at the cell pole. As for the structures described above, despite rich information about individual components and their functions, the overall architecture of the complex was unclear. Two contradicting models regarding its organization had been suggested based on crystallographic and other data [30]. Following work with negatively stained samples, through ECT of wild-type E. coli and mutant strains over-expressing the serine chemoreceptor Tsr, Zhang et al. visualized striations at the poles and identified them as chemoreceptor arrays [31]. Briegel et al. confirmed that similar structures in C. crescentus cells were in fact chemoreceptor arrays by correlating fluorescent light microscopic images (where the receptors had been labeled with a fluorescent tag) with electron cryotomograms of the same, lightly fixed cells (Fig. 4A) [9]. Analysis of the wild-type Caulobacter array structures revealed that the receptors were arranged in a 12-nm hexagonal lattice with “trimers of receptor dimers” at each vertex. A concurrent study by Khursigara et al. confirmed the lattice spacing and the trimers-of-dimers model in the same species, but emphasized that the arrays were not perfectly hexagonal [21]. In an exciting combination of tomographic, crystallographic, and “single-particle-like” techniques, Khursigara et al. went on to show that the HAMP domains in E. coli mutants over-expressing Tsr undergo conformational changes due to ligand binding and methylation (Fig. 4B and C) [32].

Figure 4. Intracellular structures.

(A) 15-nm tomographic slice of a C. crescentus cell showing a polar chemoreceptor array (arrow). Inset: fluorescence light microscopy image of the same cell showing the co-localization of McpA-mCherry with the polar array seen by ECT. Scale bar 50 nm. Adapted from [9] with permission from Blackwell Publishing Ltd. (B, C) ‘Compact’ (B) and ‘Expanded’ (C) conformations of the E. coli chemoreceptor Tsr visualized by ECT and sub-volume averaging. Adapted from [32] with permission. © 2008 by The National Academy of Sciences of the USA. (D) 4-nm tomographic slice of isolated Halothiobacillus neapolitanus carboxysomes. Scale bar 100 nm. Adapted from [34] with permission from Elsevier Ltd. (E) Tomographic slice of a Prochlorococcus MIT9313 cell showing the intracytoplasmic membrane. Scale bar 100 nm. Adapted from [36] with permission from the American Society of Microbiology. (F) 20-nm tomographic slice of a Rhodopseudomonas viridis cell exhibiting photosynthetic membranes. Scale bar 200 nm. Adapted from [37] with permission from Elsevier Ltd. (G) 8-nm slice of a Bdellovibrio bacteriovorus cell exhibiting a distinct nucleoid region. Scale bar 150 nm. Inset scale bar 25 nm. Adapted from [19] with permission from the American Society of Microbiology.

Carboxysomes

The carboxysome, a proteinaceous shell that encapsulates bacterial ribulose 1,5-bisphosphate carboxylase/oxygenase (RuBisCO), is the most well characterized member of a family of polyhedral bodies termed bacterial microcompartments that encapsulate key enzymes [33]. In cyanobacteria and many chemoautotrophic bacteria, the encapsulated RuBisCO catalyzes the first step of carbon fixation in the Calvin-Benson-Bassham cycle. The structure of carboxysomes has been proposed to increase the local concentrations of substrate and enzyme and sequester the reaction from useless side reactions. Schmid et al. first reported the structure of carboxysomes by ECT from Halothiobacillus neapolitanus (Fig. 4D), showing that they were regular icosahedra but with different sizes. These authors averaged sub-classes and suggested that the different sizes might be a result of different packing arrangements of the shell proteins [34]. Iancu et al. imaged carboxysomes from Synechococcus strain WH8102, showing that they too were regular icosahedra, but suggested instead that the different sizes arose from different T-numbers [35]. Iancu et al. went on to show through simulation that the concentric shells of RuBisCO seen in both studies could be explained by simple close packing.

Photosynthetic membranes

Two ECT studies have characterized bacterial photosynthetic membranes [36,37]. Ting et al. applied ECT to both frozen-hydrated cells and sections of two closely related, ecologically important cyanobacteria (strains of Prochlorococcus) [36]. Compared to the MIT9313 strain, the MED4 strain (which has one of the smallest genomes (1.66 Mbp) of any known photosynthetic organisms) had a smaller cell volume, a smaller carboxysome, less intracytoplasmic lamellae, and a minimal cell wall architecture (Fig. 4E).

Comparative genomic analyses found differences in genes between the two strains including those involved in peptidoglycan synthesis. Konotry et al. applied whole-cell ECT to the anaerobic purple photosynthetic bacterium Rhodopseudomonas viridis (Fig. 4F), documenting the tunnel-like structures connecting the photosynthetic membranes to the inner membrane and the highly packed arrangement of the flattened sacs [37].

Nucleoid and cell-cell interaction

In traditional EM images, the nucleoid, which is supposed to contain the compacted genome, was often seen as a lighter area within the cytoplasm. In higher detail and reliability, cryotomograms of thin bacterial cells often show a central ribosome-free area that has an obviously finer texture than the rest of the cytoplasm. A particularly clear example is seen in the highly bent Bdellovibrio bacteriovorus (Fig. 4G) [19]. ECT of frozen-hydrated sections of Gemmata obscuriglobus (a member of the phylum Planctomycetes) revealed a network of double-membrane compartments that probably enclose packed chromatin, an organization speculated to increase the radiation tolerance of this species [38]. ECT also revealed the structure of the contact site between two symbiotic bacteria, Ignicoccus hospitalis and Nanoarchaeum equitans [39], and the 3-D ultrastructure of an uncultivated, ultra-small archaeon in acid mine drainage (AMD) biofilm samples [40].

Future prospects

In addition to these dominantly cryotomographic studies, others have begun to include ECT reconstruction as just one of a set of approaches to characterize cells and processes [41-44]. Thus in combination with other new imaging and experimental methods, ECT is precipitating a new era in bacterial cell biology. Improvements in both the quality and number of cryotomograms being produced should be expected. Beginning with the sample, further improvements in methods for cryosectioning will make cryotomography of serial sections possible for larger cells and even biofilms. The development of genetically-encodable, electron-dense tags to unambiguously identify specific macromolecules in cryotomograms is sorely needed [45-47], as are further improvements in the technologies for correlating light and electron cryomicroscopy [6-9].

Instrumentally, the development of direct electron detectors, phase plates, and aberration correctors could dramatically improve image quality [48,49]. Automation of image acquisition, tilt-series alignment, and reconstruction [50,51], coupled with the development of databases for the management and recovery of the large datasets produced by ECT [52](J. Ding et al., in preparation) will facilitate the imaging of very large numbers of cells, as is sometimes necessary. Computational advances such as contrast-transfer-function correction and improved methods for denoising, template-matching, sub-volume averaging, and segmentation will all enhance the interpretability of the tomograms [53] [54]. Finally, as more and more institutions acquire the expensive microscopes needed and more and more practitioners are trained, the numbers of studies applying this technology will also obviously increase.

Supplementary Material

Supplementary Movie

The movie is a three-dimensional tour through the ECT reconstructions of four cells: two NA1000 cells and two cells over-expressing FtsZG109S. The NA1000 cells are introduced with a fLM image of YFP-labeled FtsZ in NA1000 cells (labeled “Caulobacter crescentus PvanA::ftsZ-eyfp (MT196)” in the movie), showing that FtsZ normally localizes as a ring at midcell. Next a cryo-EM image of a dividing NA1000 cell is shown, zooming in on the division site, which will be reconstructed. The original tilt-series of images is shown (labeled “Tilt Series”), followed by the full 3-D reconstruction (labeled “Tomogram”) slice-by-slice. The segmented inner membrane (blue), outer membrane (gold), and FtsZ filaments (red) are then shown from a range of views while the raw reconstruction is flashed on and off to give the viewer a sense for the interpretability of the data. Next a second NA1000 cell is shown (labeled “Another cell more deeply constricted”) in a similar but abbreviated fashion without the tilt series or full reconstruction. The two cells over-expressing FtsZG109S are introduced with another fLM image using FtsZ-YFP (labeled “FtsZ G109S”), showing that FtsZ accumulates in the extended division sites. An additional fly-around of one particular filament and its connection to the cell wall is included (labeled “Connections to peptidoglycan”). Finally the second cell over-expressing FtsZG109S is shown (labeled “Another mutant cell”), wherein an additional cytoplasmic cytoskeletal filament bundle (orange) is seen. Three particularly well-resolved, long filaments that could almost be traced over the top and bottom of the cell were organized as interlaced spirals. Adapted from [5] with permission from Nature Publishing Group.