Ultrarapid freezing of bacteria (i.e. vitrification) results in optimal preservation of native structure. In this study, cryo-transmission electron microscopy of frozen-hydrated sections was used to gain insight into the organization of the Bacillus subtilis 168 cell envelope. A bipartite structure was seen above the plasma membrane consisting of a low-density 22 nm region above which a higher-density 33 nm region or outer wall zone (OWZ) resided. The interface between these two regions appeared to possess the most mass. In intact and in teichoic acid-extracted wall fragments, only a single region was seen but the mass distribution varied from being dense on the inside to less dense on the outside (i.e. similar to the OWZ). In plasmolysed cells, the inner wall zone (IWZ)'s thickness expanded in size but the OWZ's thickness remained constant. As the IWZ expanded it became filled with plasma membrane vesicles indicating that the IWZ had little substance and was empty of the wall's polymeric network of peptidoglycan and teichoic acid. Together these results strongly suggest that the inner zone actually represents a periplasmic space confined between the plasma membrane and the wall matrix and that the OWZ is the peptidoglycan-teichoic acid polymeric network of the wall.
Transmission electron microscopy (TEM) of conventional thin sections has long been a primary tool to examine the ultrastructure of bacterial boundary layers. Based on the response of bacteria to the Gram reaction, thin sections have shown that Gram-negatives possess a cell wall consisting of an outer membrane and a peptidoglycan layer. This layer is found in a defined space between the plasma and outer membrane, called the periplasmic space, and is filled with periplasm (Hobot et al., 1984; Beveridge and Graham, 1991). Together the outer membrane, the peptidoglycan layer and the periplasm constitute the cell wall in Gram-negative bacteria (Murray, 1963; Beveridge, 1981; 1999; Beveridge and Graham, 1991; Matias et al., 2003).
For conventional embeddings, it is well recognized that the harsh treatment that bacteria are subjected to during fixing, dehydration and embedding can both denature and extract essential cell envelope constituents, thereby inducing structural artefacts (Beveridge et al., 2005). A more recent cryo-technique, freeze-substitution, provides better preservation and a more natural view of bacteria in thin section (Beveridge, 1999; 2000). Here, bacteria are vitrified by rapid freezing, and are chemically fixed, stained and dehydrated at −80°C without thawing (Graham and Beveridge, 1990). They are then embedded in plastic and thin sectioned. Thin sections of freeze-substituted Gram-positive walls appear to show more complexity than seen by more traditional means (Umeda et al., 1987; Paul et al., 1993; Graham and Beveridge, 1994; Beveridge, 2000). This is most apparent in Bacillus subtilis, which has been used as a model Gram-positive bacterium (Fig. 1), where a three-zoned wall is seen (Graham and Beveridge, 1994). Here, the wall region immediately above the plasma membrane is an electron dark zone followed by a more electron translucent zone. The outermost (and third) zone is a fibrous layer. Each of these zones is compatible with the concept of cell wall turnover (Koch, 1983; Koch and Doyle, 1985; Beveridge and Graham, 1991; Archibald et al., 1993; Beveridge, 2000), even when the new ‘scaffold’ (Dmitriev et al., 2003) or older ‘horizontal layer’ (Vollmer and Höltje, 2004) models for peptidoglycan arrangement are considered. Like conventional thin sections, freeze-substitutions reveal no apparent periplasmic space between the plasma membrane and the cell wall (Fig. 1). Even though modern transmission electron microscopes currently resolve some biomaterials at better than 0.5–1.0 nm (often by selected area electron diffraction), major questions still remain about the organization of Gram-positive cell walls and this is mainly due to our inability to preserve and view their fabric in its natural state.
In this report, we use frozen-hydrated thin sections to provide a more natural view of the B. subtilis cell envelope in order to differentiate the wall from the plasma membrane, to more accurately determine the location of the wall's polymeric network, and to detect the possible existence of a periplasmic space. We chose this method because by rapidly freezing and vitrifying the cells, molecular motion and hydrolytic enzymes are rapidly stopped thereby preserving ultrastructure in the best possible way (Dubochet et al., 1988; Harris, 1997). If the vitrified cells are thawed, the vast majority of cells regains viability and once more become active metabolic cellular units. The difficulty of this cryo-technique, though, is that, unlike thin sections from conventional and freeze-substitution preparations that use water to float and stretch sections during sectioning, the production of frozen-hydrated sections is done on a dry knife. Accordingly, compression of cells and the development of discontinuities within the ice during sectioning impact clarity. Yet, these difficulties can be surmounted and the results are gratifying as they ensure a re-evaluation of Gram-positive envelope structure. Unlike traditional TEM methods, the frozen-hydrated sections used in our study are not stained with heavy metal contrasting agents and imaging relies on differential mass within the cells.
General remarks on the freezing and vitrification of bacteria
Although high-pressure freezing results in a 10-fold increase in the vitrification depth (Sartori et al., 1993), a cryo-protectant was still required to ensure that deleterious cytoplasmic nucleation of ice crystals was inhibited. In our experience, and that of others (Dubochet et al., 1983; 1988), the use of cryo-protectants is a necessary requirement for efficient vitrification of large aggregates of bacteria for cryo-sectioning. Usually dextran, sucrose or glycerol are used as cryo-protectants (Dubochet et al., 1983; Matias et al., 2003). In this present study, we chose glycerol over dextran or sucrose because it has a lower molecular weight, can be used at a lower concentration (10% versus 15–20% w/w) and is satisfactory for most Gram-positive bacteria (V.R.F. Matias and T.J. Beveridge, unpublished). The reduction in glycerol concentration for cryo-protection and vitrification can be attributed to the lower freezing point of the glycerol (compared to dextran or sucrose) at these concentrations. Furthermore, cells tend to pack tighter when pelleted by centrifugation in glycerol, resulting in a smaller interstitial volume outside cells. This is important as cells and their structures tend to more readily vitrify than the interstitial fluid because of their relative higher dry weight content.
In order to minimize osmotic effects associated with the use of glycerol, bacteria were grown with this low molecular weight cryo-protectant in the growth medium (This was instead of shocking them by brief immersion in the cryo-protectant immediately before high-pressure freezing). Compared to growth without cryo-protectant, the only noticeable difference with glycerol was a slight decrease in the growth rate (Td = 29 min for glycerol versus Td = 20 min without glycerol). During division, cells did not separate as rapidly during growth in glycerol (resulting in short chains of cells) but had developed a typical single cell phenotype by stationary phase. This same alteration in growth pattern was also seen with growth in (20% w/w) dextran or (15% w/w) sucrose. Admittedly, glycerol can be a carbon source for some Gram-positive bacteria, including B. subtilis, but under our growth and cryo-protectant regimen, this did not alter vitrification. It is possible, though, that glycerol could have had a subtle effect on structural detail such as the size of regions within the cell envelope.
Frozen-hydrated thin sections and associated cutting artefacts
At low magnification, frozen-hydrated sections of B. subtilis showed cutting artefacts that are typically associated with the technique (Fig. 2A and Dubochet et al., 1988). In contrast to resin-embedded samples, where water is used to float and stretch sections during sectioning, frozen-hydrated sections cannot be collected on a fluid surface as there are no liquids with high enough surface tension that can be used at such low temperatures (i.e. −140 to −180°C). Accordingly, stresses created during cryo-sectioning remain on the sections as they collect on the dry knife edge. Cutting artefacts include knife marks, crevasses and compressions associated with the cutting direction. Contaminating ice crystals can also be found on the surface of frozen sections because the sections act as a cold trap for water vapour in the air.
Thin copper tubes were used in our high-pressure freezing system for rapid energy conduction during vitrification. The diameter of the tubes was so narrow that capillary action on these rod-shaped bacteria aligned them longitudinally to the tube axis. This also appeared to be a close-packing phenomenon as the cells were highly concentrated. Consequently, only cross-sections of the cylindrical region of cells were obtained (Figs 2A and 3). Some cross-sections look more oblong than circular because of compression during sectioning, which reduces the section length in the cutting direction with a corresponding increase in section thickness (Fig. 2B).
Structure of frozen-hydrated B. subtilis
The cytoplasm of B. subtilis was filled with large well-preserved ribosomes and thin DNA fibres that were dispersed throughout the cytosol (Fig. 3A) (Conventional protocols using chemical fixation show small compacted ribosomes with the DNA condensed into a fibrous central mass in the cytoplasm; Beveridge, 1989a). In frozen-hydrated sections, the cell wall was particularly well preserved in the non-deformed regions of the cells (Figs 2B and 3A) and, strikingly, walls appeared to be bipartite, with a 22 nm inner zone (IWZ) showing less contrast than a 33 nm outer zone (OWZ) (Fig. 3B and Table 1). As contrast is directly proportional to density in frozen-hydrated samples (Dubochet et al., 1983), our results showed that the B. subtilis wall possesses two regions of different distinct densities. This view differs from the tripartite format seen in freeze-substituted cells, which show a heavily stained inner zone followed by a translucent zone and a more heavily stained fibrous zone at the outer surface of the wall (Fig. 1; Graham and Beveridge, 1990; 1994).
Table 1. Measurements on structures and compartments of B. subtilis.a
Cell wall fragments
Teichoic acid-extracted cell wall fragments
. Average ± standard deviation of 12 measurements.
. Taken across the cell from outer face to outer face of the wall.
. The uneven spacing of the inner wall zone made measurement impossible.
To help make correlation easier between freeze-substitution and frozen-hydrated sections of intact cells, cells were mechanically broken with a French press and SDS boiled so that cell wall fragments could be isolated, frozen and cryo-sectioned (Fig. 4A; for images of freeze-substituted cell wall fragments, see Graham and Beveridge, 1994). Surprisingly, cross-sections of wall fragments did not reveal them to be bipartite but instead to be monopartite (Fig. 4B). The cell walls consisted of a 45 nm thick single-zoned dense matrix, presumably the OWZ, which retained the original curvature of the cell (Fig. 4B and Table 1). Indeed, cell shape was retained by the wall fragments (Fig. 4A) implying a certain rigidity and prepatterned contour. Compared to the cell wall seen on intact cells, the thickness of the OWZ had expanded (from 33 nm to 45 nm). As hot SDS treatment solubilizes wall-associated proteins and contaminating debris from wall fragments (e.g. membrane, ribosomes, DNA, etc.; Hancock and Poxton, 1988) and leaves the cell wall intact (i.e. the peptidoglycan-teichoic acid matrix; Archibald et al., 1993), this is strong evidence that the OWZ is in fact the cell wall. The cross-section diameter of the cell wall fragments remained unchanged when measured from outer face to outer wall face (Table 1) but the wall matrix expanded inwards, which could be a response to the absence of turgor pressure once the cells were broken. The absence of the IWZ in isolated wall fragments implies that the inner region of the wall seen on cells is composed mostly of soluble less dense components that were washed away during the isolation of walls. Mass distribution is not always readily apparent in frozen-hydrated sections but densitometry of these fragments suggested that there was a mass decrease from inner to outer face (Fig. 6B). This will be further discussed in a latter section.
Removal of teichoic acid
Since our cells were grown in relatively rich medium containing phosphate, the walls were in the teichoic acid state (Neuhaus and Baddiley, 2003). This was confirmed by phosphorus analysis. As alkali removes this polymer from the peptidoglycan network, we treated our cell wall fragments with NaOH removing 92.8% of teichoic acids (also based on phosphorus analysis) resulting in fragments composed predominantly of peptidoglycan. This removal of teichoic acid had a profound influence on the wall as seen in frozen-hydrated sections because the cellular shape and rigidity of the fragments were lost (Fig. 5A). Higher magnifications revealed that the walls were thinner (∼34 nm thick; Table 1), more bendable and less dense than untreated walls (cf. Figs 4B and 5B). As, in our 168 strain, teichoic acid accounts for approximately 50% of the dry weight of wall fragments (Koch and Doyle, 1985; Archibald et al., 1993), a loss of contrast and a reduced thickness were expected once the teichoic polymers were removed. In an indirect way, this removal of teichoic acid and the maintenance of a monopartite structure help confirm that the OWZ is the actual cell wall of the cell.
It is interesting that the isolated walls showed a retention of cellular shape, which was however, lost after the extraction of teichoic acids. This is consistent with the irregularly shaped cells that resulted from the controlled depletion of an important enzyme for the synthesis of teichoic acids (Bhavsar et al., 2001). It also points to a substantial role for teichoic acids for the rigidity of cell walls. It seems that interactions between both the phosphoryl and protonated d-alanyl groups of teichoic acids with the carboxyl groups of the peptidoglycan (Neuhaus and Baddiley, 2003) could restrict the conformation of peptidoglycan fibres, aiding rigidity and retention of shape.
Polymeric differentiation of the cell wall into zones of mass distribution
Closer examination of the OWZ at higher magnification revealed an asymmetric distribution of mass throughout the wall thickness (Fig. 6A). The OWZ appeared with progressively less contrast from the inner face to the outer face, which is similar to the differentiation seen on walls of freeze-substituted cells (between the intermediate and outermost regions; Fig. 1). At this point, it must be emphasized that the contrast seen in freeze-substitution preparations is generated by heavy metal stains, thereby making the detection of different wall zones easier. The outermost region of the wall is especially better seen where high hydrolysis rates generate many reactive sites for metal binding and higher contrast. Conversely, frozen-hydrated sections show less contrast on the outer face of the OWZ because of its lower local density because autolysins, involved in cell wall turnover, have solubilized the polymeric network in this region (Fig. 6A). Similar wall differentiations were also observed in native and teichoic acids-extracted wall fragments where the wall was denser on the inner face compared to the outer face as shown by density tracings of the corresponding images (Fig. 6B,C). The tracings of all preparations (Fig. 6A–C) showed a progressive decrease of wall mass from inner face to outer face, which is what would be expected with wall turnover progressing from the inside layers of polymeric network to the outside.
Interestingly, the removal of teichoic acids affected the wall uniformly from inside to outside, so that no change in monopartite infrastructure was seen, suggesting that these anionic polymers were not segregated to any specialized region within the wall, thereby confirming previous experimentation (Neuhaus and Baddiley, 2003). However, extraction of teichoic acids reduced the wall thickness by 10 nm (Table 1) indicating that either teichoic polymers extend about 10 nm above the surface of the peptidoglycan network or that their interaction with peptidoglycan expands the entire wall fabric by this amount of extension.
In order to further investigate how the IWZ integrates into the cell envelope structure of B. subtilis, intact cells were plasmolysed so as to artificially increase the separation between the wall fabric and the plasma membrane. In this case cells were grown without a cryo-protectant and then subjected to aqueous solutions of increasing osmolarity (10% glycerol, 20% sucrose, 20% glycerol and 20% glycerol-5% NaCl solutions). The 20% glycerol-5% NaCl solution caused the best plasmolysis of cells and, because glycerol was used, vitrification could immediately be done. These plasmolysed cells possessed compacted cytoplasms and gaps were seen between the plasma membrane and the cell wall (cf. Fig. 3A with Fig. 7A–C); the protoplasts were shrinking (as the water was being drawn out of them) and the space between wall and membrane was increasing. The density of the space resembled the IWZ except that it was thicker. High magnifications of this space showed it to be of low density (i.e. similar to the ice surrounding the cells) so that there could be little actual substance to it (Fig. 7D). Above this space a much denser layer, the OWZ, could be seen. The OWZ now approached the thickness of the cell wall fragments (Table 1), which would be expected as plasmolysed cells do not exert turgor pressure (or surface tension; Koch, 1983) on the walls (i.e. the walls would not be as stretched and compacted as in normal cells).
Frequently, membrane vesicles extruded into the artificial gap (Fig. 7C) and these were filled with cytoplasmic material (Fig. 7E). These resembled the ‘mesosome bodies’ found in Bacillus megaterium after plasmolysis in sucrose solutions stronger than 1 M (Weibull, 1965). Contraction of the protoplast during plasmolysis resulted in a considerable reduction in the total area of the membrane, and hence a fraction of it was excised as membrane vesicles. It is also possible that distinct regions of the membrane were strongly bound to the cell wall via penicillin-binding proteins (PBPs; Blumberg and Strominger, 1972), lipoteichoic acids (Neuhaus and Baddiley, 2003) and lipoproteins (Sutcliffe and Russel, 1995; Sutcliffe and Harrington, 2002), thereby pulling away these areas as the protoplast retracted from the cell wall and helping to develop the vesicles. As the protoplasts continued to shrink during plasmolysis, these vesicles were forced out of the cell into this space until they were confined by the OWZ (Fig. 7E). If this space (i.e. the IWZ) consisted of significant substance and was an essential part of the cell wall matrix, the vesicles would not have room to develop and be present. Whatever is in the IWZ, it is quite compactable, is of low density, and can be deformed by the action of the vesicles. These data again suggest that the IWZ is less substantial (as it has little density) than the OWZ, and that the actual wall is the OWZ. It is probable that the IWZ is a periplasmic space, which is confined between the plasma membrane and the cell wall and is filled with a variable and low concentration of periplasmic components. Even though this region has low mass, it must still contain important components such as PBPs, lipoteichoic acids, enzymes and secreted proteins. Indeed, the concentration of substance in this periplasmic region, although low, must be great enough to resist compression resulting from turgor pressure exerted by the protoplast. Additionally, lipoteichoic acids, which are embedded into the plasma membrane and extend into the cell wall, could aid in connecting the membrane, periplasmic space and cell wall amalgam together.
Existence of a periplasmic space
In this study, we present the structure of B. subtilis 168 and its cell envelope by cryo-TEM of frozen-hydrated sections. The plasma membrane is surrounded by a low-density 22 nm thick zone, which is enclosed by a higher-density 33 nm thick zone. Our results strongly suggest that the IWZ is a periplasmic space while the OWZ is the actual cell wall consisting of a peptidoglycan-teichoic acid polymeric matrix and associated proteins. By convention, using the terminology employed for the Gram-negative envelope (Beveridge and Graham, 1991; Beveridge, 1999; Matias et al., 2003), the periplasmic space and its constituent periplasm should be considered an essential but less substantial part of the cell wall.
The conclusion that the IWZ represents an extra-protoplasmic compartment between the plasma membrane and the peptidoglycan-teichoic acid wall matrix comes from a number of observations. First, only one substantive wall zone was observed above the plasma membrane with intact cells, and this same zone was seen in wall fragments (i.e. the OWZ) and to a lesser extent in teichoic acid-extracted fragments. Second, the OWZ and both types of wall fragments possessed a differentiated wall structure, showing higher density at the wall inner face compared to the outer face. The IWZ was not differentiated. Third, membrane vesicles were able to extrude into the IWZ. Fourth, after extrusion the membrane vesicles were deformed and limited by the OWZ indicating the solid nature of this outer layer, whereas the IWZ was non-deforming. And, fifth, the IWZ had very low density (little substance) and the OWZ had high density (elevated substance) as determined by their innate electron scattering power.
The conclusions drawn from our hydrated-frozen section data could be interpreted as diametrically opposed to previous freeze-substitution experiments (Amako et al., 1982; Umeda et al., 1987; Graham and Beveridge, 1990; 1994; Beveridge and Graham, 1991; Graham et al., 1991; Paul et al., 1993; Beveridge, 1995; 2000), but this is not so. Freeze-substitution, although a cryo-technique that accurately preserves cells, provides entirely different information than that provided here by frozen-hydrated sections. In freeze-substitution, cells are vitrified and then chemically substituted at −80°C (i.e. chemically fixed, dehydrated and embedded in plastic; Beveridge et al., 2005). Before the bacteria can be imaged in thin section, they must be stained by heavy metal salts (such as uranium and lead) to increase the scattering power of the biomaterial beyond that of the embedding plastic. Electron microscopic images of freeze-substituted bacteria, then, depend on the binding of heavy metal stains to reactive sites in the biomaterial (primarily electronegative groups; Beveridge, 1989b) and, accordingly, reveal where these reactive sites are (Koval and Beveridge, 1999). This is different from frozen-hydrated sections, which reveal where regions of high differential mass occur (i.e. as opposed to the density of vitrified water).
Correlation of previous freeze-substitutions of B. subtilis (Graham and Beveridge, 1994; also see Fig. 1) with our current frozen-hydrated section results (e.g. Fig. 3B) suggests a periplasmic space exists that is filled with low-density material (frozen-hydrated sections) and that this material (i.e. periplasm) is highly reactive with heavy metal stains so as to produce high contrast in the region (freeze-substitution). This correlation is diagrammed in Fig. 8.
Initial observations on Gram-negative cell envelopes by freeze-substitution suggested that these periplasmic spaces where so filled with periplasmic materials that a gel was formed, the ‘periplasmic gel’ (Hobot et al., 1984). Yet, frozen-hydrated sections suggested that the Gram-negative periplasm was a low-density material (Matias et al., 2003). Like the periplasm of B. subtilis seen in our present study, the periplasms of Escherichia coli and Pseudomonas aeruginosa also appear to be filled with low-density but intensely reactive (staining) substance (Matias et al., 2003). This new information adds to our growing understanding of periplasmic spaces in prokaryotes (Graham et al., 1991; Beveridge, 1995; Merchante et al., 1995).
The existence of a periplasmic space in Gram-positive bacteria could be advantageous; it would provide a space to manoeuvre enzymes in between the plasma membrane and cell wall, away from the highly negatively charged wall polymers and sufficiently apart from the wall to avoid steric crowding. In S. pneumoniae and S. aureus, crystal structure determination of PBPs has shown that they could extend 13–9.7 nm above the membrane (Pares et al., 1996; Lim and Strynadka, 2002). This approximates the 7.5 nm length of PBP5 in E. coli (Nicholas et al., 2003; C. Davies, pers. comm.), which is close to the distance of 9.3 nm between the plasma membrane and the peptidoglycan layer seen in frozen-hydrated sections of this bacterium (Matias et al., 2003). It is probable that PBPs require a certain amount of free space within the periplasm to catalyse the development of new wall fabric.
Interpretation of the mass distribution within the peptidoglycan-teichoic acid network (i.e. OWZ)
One additional feature that is not seen in the frozen-hydrated sections is the ‘fringe’ that is seen at the top of the wall in freeze-substitutions (Figs 1 and 8), which has been attributed to cell wall turnover (Graham and Beveridge, 1994) because this is a region that is actively being hydrolysed into soluble polymers (Mobley et al., 1984; Koch and Doyle, 1985; Archibald et al., 1993). It is thought that the hydrolysis of bonds in this region of the wall results in the disassembly and agglomeration of wall components into a fibrillar fringe during dehydration, which is readily seen because of the staining of many of these exposed reactive sites with heavy metals during freeze-substitution. This fringe is not seen in frozen-hydrated sections because biological structures are immobilized in their fully hydrated state during vitrification, preventing the highly hydrolysed outermost wall from agglomerating into a fibrillar material. Instead, frozen-hydrated sections showed progressively less contrast (lower density) through the wall thickness from inside to outside on cells and isolated wall fragments (Figs 3B, 4B and 5B). This is also in agreement with the concept of cell wall turnover and corroborates previous freeze-substitution results on the differentiation of the B. subtilis cell wall (Fig. 8).
The traditional view of peptidoglycan organization in bacterial walls is that the glycan strands are laid down as interconnected polymeric sheets (Pink et al., 2000; Vollmer and Höltje, 2004). Gram-positive walls would have multiple sheets sitting on top of and covalently linked to one another (Koch and Doyle, 1985) in a manner that should provide innate molecular infrastructure throughout the wall's thickness (Pooley, 1976; Graham and Beveridge, 1994; Touhami et al., 2004). More recently, a ‘scaffold’ model has been proposed for peptidoglycan organization (Dmitriev et al., 2003). These two competing models (sheet versus scaffold) have been extraordinarily difficult to detect at the molecular level by any high resolving means and our current study does not determine this issue any better than previous studies. It does, however, clarify molecular domains or zones by density and charge within the Gram-positive wall and these domains must indirectly reflect polymeric molecular organization.
Our classical textbook view of Gram-positive walls as being thick, amorphous and undifferentiated, given by conventional thin sections, does not reflect the dynamic properties and presumably differentiated state required for this stress-bearing structure. A Gram-positive wall must resist high turgor pressure of about 2626 kPa (Thwaites and Surana, 1991) while presenting high turnover rates during cellular growth (Archibald et al., 1993). Intuitively, insertion of new peptidoglycan strands occurs at the wall's inner face, the age of the wall increases with the distance from the membrane, and old wall material is shed from the outer wall face at approximate rates of 50% per generation (Pooley, 1976; Mobley et al., 1984; Merad et al., 1989; Archibald et al., 1993). As better methods for preservation and resolution are acquired, our appreciation of the intricacies of these walls increases. At this time, cryo-TEM appears to be a new modern approach that holds great promise as it has shown a differentiation of wall mass that supports the current concept of wall turnover and it has confirmed the existence of a periplasmic space.
Bacterial strain and growth conditions
Bacillus subtilis 168 was grown at 37°C to mid-exponential growth phase (optical density at 470 nm, 0.5–0.7) in trypticase soy broth (TSB). Once good growth was obtained, the cells were transferred to TSB containing 10% (w/w) glycerol, and harvested cells were washed three times in 50 mM Hepes (N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid, pH 7.0) or 10% (w/w) glycerol in 10 mM Hepes. For growth, whenever glycerol was added, it was used as a cryo-protectant. A pellet of cells grown in TSB with glycerol and washed in buffered glycerol was directly used for freezing (i.e. vitrification).
Isolation of cell wall fragments
Bacteria grown in TSB were resuspended in 50 mM Hepes (pH 7.0), 50 µg ml−1 DNase, 50 µg ml−1 RNase and French pressed twice at 18 000 psi. The suspension was spun (3000 g) to remove any intact cells. The remaining supernatant, containing the cell wall fragments, was boiled in 4% (w/v) SDS for 2 h (Sprott et al., 1994). Fragments were washed five times in deionized water and three times in 12.5% glycerol 10 mM Hepes using a Beckman ultracentrifuge (150 000 g for 30 min at 4°C). This higher glycerol concentration was necessary to better protect the cell walls during freezing. For the extraction of teichoic acids, water-washed cell wall fragments were treated with 0.5 M NaOH for 90 min at room temperature (Hancock and Poxton, 1988). TA-extracted cell walls were washed five times in deionized water and three times in 16% glycerol in 10 mM Hepes. Again, this higher glycerol concentration better protected the extracted walls during freezing.
Plasmolysis of B. subtilis was proceeded after growth in TSB by washing cells twice using either 10% (w/w) glycerol, or 20% (w/w) glycerol, or 20% (w/v) sucrose or 20% glycerol−5% NaCl, all prepared in 10 mM Hepes (pH 7.0). In all cases, the various additions were necessary for either cryoprotection or osmotic stabilization of the plasmolysed cells. Pellets obtained were used for freezing and sectioning.
Freezing and sectioning of bacteria
Pellets of samples for freezing were further centrifuged five times (6000 g for cells and 16 000 g for cell wall fragments, for 1 min each time) and the supernatant was removed by touching it with a filter paper (Whatman). This allowed the vitrification of samples at a lower concentration of the cryo-protectant by reducing the amount of interstitial fluid available (see text). Each pellet was drawn into a disposable plastic syringe, which was attached to a thin copper tube for eventual freezing. The bacteria or walls were injected into the tube from the syringe and immediately vitrified using a Leica EM PACT high-pressure freezer. The copper tube ensured rapid transfer of latent heat from the samples aiding vitrification. Frozen samples were sectioned in a Leica cryo-ultramicrotome to a nominal thickness of 50 nm using a 45° diamond knife (Diatome), and mounted on carbon-coated 1500-mesh copper grids (Al-Amoudi et al., 2002; Matias et al., 2003).
Cryo-transmission electron microscopy and chemical analyses
Grids containing the frozen hydrated thin sections were mounted into a Gatan cryo-holder for direct observation at −170°C in a LEO 912AB energy-filtered cryo-TEM operating at 100 kV. Images were taken using a 1024 × 1024 pixel slow-scan CCD camera (Proscan). Images were stored and analysed using analySIS (SIS, Munster, Germany) software. Length measurements were done on the least deformed regions of cells (regions of high curvature; Fig. 2B).
Phosphorus analysis of cell walls was performed using a Varian Vista-pro radial ICP with a Cetac ultrasonic nebulizer.
This work was supported by a Natural Science and Engineering Research Council of Canada (NSERC) Discovery grant to T.J.B. V.R.F.M. is recipient of a PhD scholarship from CNPq/Brazil. Microscopy was performed in the NSERC Guelph Regional STEM Facility, which is partially funded by an NSERC Major Facility Access grant to T.J.B.