Cryo-electron microscopy of cell division in Staphylococcus aureus reveals a mid-zone between nascent cross walls

Authors


E-mail tjb@uoguelph.ca; Tel. (+1) 519 824 4120 ex 53366; Fax (+1) 519 837 1802.

Summary

Cryo-electron microscopy of frozen-hydrated thin sections permits the observation of the real distribution of mass in biological specimens allowing the native structure of bacteria to be seen, including the natural orientation of their surface layers. Here, we use this approach to study the fine ultrastructure of the division site, or septum, of Staphylococcus aureus D2C. Frozen-hydrated sections revealed a differentiated cell wall at the septum, showing two high-density regions sandwiched between three low-density zones. The two zones adjacent to the membrane appeared as an extension of the periplasmic space seen in this organism's cell envelope and showed no distinguishing structures within them. Immediately next to these were higher-density zones that corresponded to nascent cross walls of the septum. Unexpectedly, a rather broad low-density zone was seen separating cross walls in the septum. This mid-zone of low density appeared inflated and without visible structures in isolated cell walls, which showed only the high-density zones of the septum. Here, we suggest that frozen-hydrated thin sections have captured a highly fragile septal region, the mid-zone, which results from the dynamic action of autolysis and actively separates daughter cells during division. The two zones next to the membranes are periplasmic spaces. Immediately next to these are the growing cross walls composed of peptidoglycan, teichoic acid and protein.

Introduction

Peptidoglycan is the main cell wall component in almost all bacteria, constituting a single huge macromolecule that forms a highly cross-linked network and outlines the shape of the bacterium (Beveridge, 1981; Archibald et al., 1993; Scheffers and Pinho, 2005). This bag-shaped molecule (or sacculus) provides cells with a rigid enough exoskeleton to cope with high turgor pressures throughout the cell cycle. Still, peptidoglycan is an elastic meshwork that expands and contracts in response to variations in osmolarity and ionic strength of the medium (Koch et al., 1987; Doyle and Marquis, 1994). The dynamic nature of peptidoglycan is especially recognized during cell division, when the sacculus has to expand in size to eventually allow the formation of two daughter cells. In Gram-positive bacteria, peptidoglycan expansion involves turnover of muropeptides (Doyle et al., 1988), autolysis of the network and incorporation of precursors for cell wall growth (Höltje, 1998; Nanninga, 1998), and attachment of large amounts of teichoic acids and wall proteins (Navarre and Schneewind, 1999; Neuhaus and Baddiley, 2003) and remarkably, all these processes are carried out while maintaining an intact and functional sacculus.

In the model rod-shaped bacteria, Escherichia coli and Bacillus subtilis, cell wall growth progresses with incorporation of precursors along side walls, leading to cell elongation (Cabeen and Jacobs-Wagner, 2005; Scheffers and Pinho, 2005). Once the cell doubles in size, addition of new wall material at the division site allows the formation of the new poles of each daughter cell (old poles are typically inert and show little wall synthetic activity; Archibald et al., 1993). The division site is where Gram-negative and Gram-positive counterparts differ mostly in the mode of wall assembly. Most Gram-negative cells constrict progressively at the mid-cell until fusion of cell envelope structures allow the separation of daughter cells, and constriction is presumably driven by the contractile FtsZ ring structure (Nanninga, 1998; Weiss, 2004). This is not septation. Although the Z ring is also found in Gram-positive bacteria, these organisms have a constant cell diameter and form a characteristic division annulus (or septum) at the division site (Beveridge, 1989; Errington et al., 2003). Here, the membrane invaginates and new wall material is added to the pre-existent side wall, forming a cross wall that grows in a manner resembling a closing iris, until a complete septum is formed (Beveridge and Matias, 2006).

Unlike rod-shaped bacteria, cell wall growth in Gram-positive cocci proceeds mostly at the mid-cell with the development of a septum (Scheffers and Pinho, 2005). In a well-characterized coccus, Staphylococcus aureus, microscopy has shown that enzymes involved in the final stages of peptidoglycan synthesis and assembly (i.e. penicillin-binding proteins; PBPs), the cell division protein FtsZ, and autolysins implicated in daughter cell partitioning localize only at the septum (Paul et al., 1995; Yamada et al., 1996; Giesbrecht et al., 1998; Pinho and Errington, 2003; 2005). Further, staphylococci divide through three consecutive planes at approximate right angles to the preceding one so that typical ‘grape-like’ clusters are formed. Concentric circular rings of recently inserted wall polymers at the surface of each newly formed hemisphere (derived from the cross wall at the septum) contrast with the more smooth surface of older walls (Koyama et al., 1977; Amako and Umeda, 1979; Touhami et al., 2004). These rings have been implicated in peptidoglycan synthesis and hydrolysis at the septum (Amako et al., 1982). As the septal region is the primary site of cell wall growth, its processing must be tightly controlled so as to prevent cell rupture in the presence of high turgor pressure. This is particularly relevant in osmotolerant S. aureus, where turgor can be as high as ∼25 atm (Kunin and Rudy, 1991).

Transmission electron microscopy (TEM) of conventional thin sections has been the primary method to study septum formation in S. aureus, showing the basic processes involved during its morphogenesis (Giesbrecht et al., 1997; 1998; Sugai et al., 1997). At the mid-cell, conventional embeddings, as well as freeze-substitutions, reveal the progressive ingrowth of a cross wall on opposite sides of the cell, each cross wall consisting of a concise mass of amorphous wall material, except for a thin dark line in the middle (often referred to as the ‘mid-line’) separating the newly forming cross walls (Fig. 1). Here, active autolysis resulting in a less cross-linked peptidoglycan has been implicated in the enhanced staining of the mid-line, as well as the presence of PBPs (Paul et al., 1995; Touhami et al., 2004). The exact cellular mechanism that balances both wall synthetic and hydrolytic activity at the site of cell division has been difficult to resolve by TEM of conventional embeddings, which has limited accuracy due to the many structural artefacts produced by the harsh chemicals used in this protocol (Beveridge et al., 2006).

Figure 1.

Thin-section of freeze-substituted S. aureus D2C cells. A dividing cell appears with an amorphous mass of wall material between septal membranes, except for a darkly stained mid-line. Arrows point to the mid-line of the septum, and arrowheads to septal membranes. Bar represents 200 nm.

We have investigated the process of cell division and septum formation in S. aureus by cryoTEM of frozen-hydrated sections. This approach makes use of vitrification of specimens to rapidly immobilize structures in a matrix of amorphous ice, ensuring a state of optimal structural preservation. Specimens are subsequently thin-sectioned and imaged at a temperature that maintains the vitreous state (Dubochet et al., 1988; Al-Amoudi et al., 2004). This method has provided exceptional images of the native structure of bacteria and revealed a periplasmic space in four different Gram-positive organisms, including S. aureus (Matias et al., 2003; Eltsov and Dubochet, 2005; Matias and Beveridge, 2005; 2006; Zuber et al., 2006). Further to our previous study on S. aureus that examined the wall organization at non-septal regions (Matias and Beveridge, 2006), we now provide a better view of the developing wall structure at the division site of this important human pathogen. This is of particular interest, because synthesis and assembly of new wall material is mostly carried out at the septum where a good number of antibiotics that are effective against this organism target different stages of cell wall synthesis (Paul et al., 1995; Pinho et al., 2001; Tomasz, 2006).

Results

General remarks on cell alignment and cutting artefacts

Knife marks and compression, together with crevasses and chatter, are typical cutting artefacts seen in frozen-hydrated sections (Fig. 2A). Crevasses and chatter represent the artefacts that most severely impair correct interpretation of inherent structure, as they are not homogeneous in space, and are more commonly avoided when dextran is used as the cryo-protectant (Zuber et al., 2006). In this study, we decided to use glycerol for cryo-protection of cells so as to grow bacteria in its presence, thus avoiding serious osmotically induced artefacts. Frozen-hydrated sections with few or no crevasses could nevertheless be produced and all sections were devoid of chatter (see Fig. 2A; for more details on cutting artefacts, see Al-Amoudi et al., 2005; Matias et al., 2003; Zuber et al., 2006).

Figure 2.

Frozen-hydrated S. aureus.
A. At medium magnification, cells appear with ribosomes as ∼20-nm-dense particles dispersed throughout the cytoplasm and the membrane is bound by a bipartite wall at non-septal regions (short black arrow). Cross walls at the septum appear with a more intricate organization (white arrow). Note that the cell at the centre of the image has a developing septum, while the cell at the bottom left shows a completed septum. The same cell looks more ellipsoidal than circular because of compression along the cutting direction. Long black arrows point to knife marks and arrowheads to crevasses.
B and C. High-magnification images with corresponding digital densitometry scans.
B. At non-septal regions, the cell envelope consists of a plasma membrane (short arrow) followed by a periplasmic space (white arrowhead) and an outer wall (black arrowhead). The periplasmic space possesses lower contrast and density than the outer wall, which displays an almost constant density throughout its thickness (contrast is directly proportional to density in these images; see Matias and Beveridge, 2006).
C. At the septum, two HDZs (short arrows) are seen sandwiched between three low-density zones (arrowheads); each HDZ displays a nearly constant level of density along its thickness. Long arrows point out cutting directions. Bars represent 250 (A) and 50 nm (B and C).

Staphylococcus aureus possesses five distinct zones at its septum

At higher magnification, S. aureus showed a cytoplasm packed with ribosomes and without visible aggregation of DNA (Fig. 2A). At non-septal regions, the cell envelope consisted of a plasma membrane surrounded by a periplasmic space and cell wall matrix (or outer wall; Fig. 2B). This differed substantially at the septum where five zones of alternating densities were observed between the two boundary membranes (it should be noted that contrast is directly proportional to density in frozen-hydrated sections; Fig. 2C; Matias and Beveridge, 2006). Like most Gram-positive bacteria, S. aureus is seen here developing a true septum, i.e. the diameter of cells remain constant (cells do not constrict) and cross walls are derived from the progressive addition of wall material to the outer wall, forming a characteristic ‘T-like’ structure in thin section, which is only a thin slice of the septal annulus. The organization of the two zones adjacent to the membrane at the septum compared well with the wall structure of the cell envelope (cf. Fig. 2B and C). In both cases, a low-density zone was seen between the membrane and a higher-density zone. The low-density zone next to the membrane appeared as an extension of the periplasmic space. The high-density zones (HDZs) appeared to be similar to the outer wall, showing a nearly constant level of density throughout their thickness (Fig. 2B and C).

A septal mid-zone of low density was also seen (Fig. 2C) that was positioned where the mid-line existed in differently processed cells (Fig. 1). This newly found mid-zone possessed low density, similar to that observed in the zones adjacent to the septal membranes. This mid-zone must be the highly stained mid-line which we now see expanded into its natural native state. Although this mid-line has been attributed to a ‘splitting system’ with its inherent ‘murosomes’, presumably involved in cell separation and consisting of concentrically arranged rings of wall substance (Giesbrecht et al., 1998), our cryoTEM data showed the middle zone of the cross wall to have no discernible structures within it (Fig. 2C).

Fine details of the S. aureus septum

High-magnification images of selected regions of the growing septum in native cells provided more detail of the five zones of the septum and of how cell division could proceed. Tight apposition between the membrane and HDZ of the septum was only observed at the tip of the ingrowing septum (Fig. 3A). The close junction between membrane and septal tip suggested an intimate relationship between them as new wall polymers are presumably laid down. This seems reasonable because the site of the growing tip should be one of the most active regions of polymer transfer and network assembly (e.g. via membrane-embedded PBPs). At regions at the very tip, each low-density zone appeared thinner than the adjacent (periplasmic) space, showing a closer contact here between HDZs and the septal membrane (Table 1; Matias and Beveridge, 2006). The HDZ in these adjacent regions was 10.7 nm thick, possessing only a fraction (56%) of the outer wall thickness (Table 1). These were presumably thinner because they have yet to receive their full complement of wall polymers and they should continue to expand until they achieve true outer wall thickness. Images showing the progressive inward growth of HDZs suggest that it is the membrane at the tip of the septum that drives most of the cell wall synthesis as the septum grows. Meanwhile, at the opposite side of the HDZ at the tip, the lower density zone is beginning to form and gets thicker upstream from the tip; an enzyme complex could be breaking down peptidoglycan fibres here, thus separating the initial single HDZ at the tip into two separate and more mature HDZs (Fig. 3A and B). This progressive thickening was expected because developing HDZs should eventually become bona fide cell walls once division is complete and daughter cells have separated. In support of this, HDZs of cells showing completed septa appeared even thicker than developing (or uncompleted) septa, more closely approaching the outer wall thickness (i.e. 70% of outer wall thickness; Table 1).

Figure 3.

High-magnification images of the S. aureus septum.
A. HDZs appear as high contrast double strips extending from the outer wall to the tip of the invaginating membrane. A concise thin mass of HDZ joins the two strips of HDZ together at the tip (arrowhead). Note that a short piece of outer wall bridges the two strips of cross wall, thereby connecting them at the cell periphery (short arrow). This ‘bridge’ is considerably thicker than the outer wall beyond this region.
B. The HDZs are thinner near the septal tip, where they show close association with the membrane. This is not as apparent with the septal tip at the left because it is not aligned properly with the electron beam. Long arrows point out cutting directions. Bars represent 50 nm.

Table 1.  Measurements on structures and compartments of the septum of S. aureus.a
Thickness of structure/compartment (nm)Cells showing uncompleted septumCells showing completed septumWall fragments showing uncompleted septumWall fragments showing completed septum
  • a. 

    Average ± standard deviation of 10 measurements, unless stated otherwise; measurements were corrected for compression, as described in Experimental procedures.

  • b. 

    The uneven spacing of the middle low-density zone made measurement insignificant.

  • c. 

    This refers to the region of the outer wall situated between HDZs.

  • d. 

    Average ± standard deviation of eight measurements.

  • e. 

    This refers to regions of the outer wall further away from the outer wall bridge.

  • NA, not applicable.

Peripheral low-density zone12.4 ± 3.613.4 ± 2.1NANA
High-density zone10.7 ± 1.613.4 ± 2.514.0 ± 3.516.7 ± 2.7
Middle low-density zone12.8 ± 2.813.0 ± 3.5NAbNAb
Outer wall ‘bridge’c39.5 ± 6.7d29.9 ± 14.6d37.9 ± 7.130.4 ± 8.9
Outer walle18.8 ± 3.119.1 ± 3.528.1 ± 4.527.7 ± 4.0

Cell wall fragments

To help in the interpretation of the five zones seen at the septum, frozen-hydrated sections of cell wall fragments were produced so as to visualize septa that were both free of cytoplasm and not under the confines of a plasma membrane. Here, dextran was used as a cryo-protectant to better ensure that sections were crevasse-free (Fig. 4). Dextran produced slightly thinner outer wall HDZs than glycerol, but this difference was within the error of measurement (cf. Table 1 and Matias and Beveridge, 2006).

Figure 4.

Cell wall fragments.
A. Wall fragments show only the HDZs after mechanical breakage and SDS boiling. There is a large separation between HDZs at the septal regions (black arrows). Note that rupture of the sacculus (white arrow) takes place further away from the septum.
B. Fragments displaying completed septa also show large separations between HDZs (arrow).
C–E. High-magnification images of septal regions close to their junction with the outer wall; arrowheads point to the inner end(s) of HDZs, short arrows to the outer wall bridge, and long arrows to cutting directions. Outer walls appear rather thin because they are all almost perpendicular to the cutting direction, where reduction of a measured structure due to compression is maximal (see Experimental procedures for more details).
C. Fragments with incomplete septum appear connected to a thicker outer wall on their outer end (arrow), while the thickness of HDZs decrease rapidly as they approach their inner ends. At the inner end, the HDZs are connected to one another by a thin mass of HDZ in about half of these fragments (arrowhead).
D. In the other half of fragments displaying incomplete septa, the inner ends of the HDZs are not connected (arrowheads), presumably sheared during cell breakage; the outer wall between HDZs also appears thicker than in other regions.
E. In fragments showing complete septa, an empty ‘inflated’ mid-zone is seen between HDZs; the HDZ of the bridge at the outer end appears thicker (arrow). No structure is seen within the matrix of the HDZs in C–E. Bars represent 200 (A and B) and 50 nm (C–E).

The wall fragments were identified by the observation of their HDZs and septa could readily be identified (Fig. 4A and B). The septal ‘double band’ pattern seen in wall fragments appeared to extend inwards from the outer wall and showed larger separations between them than seen in native cells (cf. Figs 2A and 4B). Even at high magnification, no structure was seen associated with either side of septal cross walls (Fig. 4C–E). As there was no indication of a low-density zone associated with the isolated walls, the low-density regions seen adjacent to the membranes at the septum in intact cells must be composed of soluble constituents, probably similar to those in the periplasmic space (Matias and Beveridge, 2006). These soluble constituents were removed from the fragments during isolation or boiling in SDS.

More strikingly, the mid-zone of cross walls in wall fragments also appeared devoid of detectable substance (Fig. 4C–E). In intact cells, turgor pressure ensured a constant thickness in this mid-zone but in wall fragments the absence of turgor allowed expansion. The observation that cross walls of wall fragments have enlarged mid-zones suggested that these zones lacked components cross-linked to both adjacent cross walls to hold them together (Fig. 4D and E). Thus the mid-zone of the septum resembled the periplasmic space in that it is expandable and without substantial quantities of cross-linked components. Conversely, the dark double bands (i.e. HDZs) of the septum seem to represent the peptidoglycan–teichoic acid–wall protein network of nascent walls, because they possessed similar contrast compared with the cell wall matrix at the cell envelope and also resisted boiling SDS (cf. Figs 2C, 4D and E).

As observed previously in S. aureus and B. subtilis (Matias and Beveridge, 2005; 2006), comparison of the data on wall fragments with native cells showed that the outer wall expanded in thickness upon isolation of the fragments (Table 1). This was expected, because the fabric of the wall matrix contracts its surface area in response to the absence of turgor pressure once the cells are broken, becoming thicker and denser. Interestingly, the same thickening was observed for the septal HDZs (Table 1), helping confirm that these HDZs are in fact nascent cross walls made up of the same cross-linked polymeric constituents as outer walls. Furthermore, HDZs of fragments showing completed septum were thicker than those in fragments displaying incomplete septa, consistent with the progressive addition of mass to cross walls as ingrowth advanced until eventually reaching the maximal thickness of the outer wall (Table 1).

Closer examination of the leading edge of septal HDZs in fragments showed their ends as thin extensions of HDZ (Fig. 4C), which confirmed that seen in intact cells (Fig. 3A and B). Sometimes, both shorter and longer septa were completely separated at their ends (Fig. 4D) but we believe this separation was due to the shearing of the ends during cell breakage. As these ends are so thin and possibly not as highly cross linked as adjacent matrix, they would be highly susceptible to shear. In concert, these observations corroborate those seen in intact cells suggesting that ingrowing HDZs are synthesized as a single growth point, which rapidly separates into two planes by enzyme action.

Septum fusion

In a thin section running right through the middle of a septating cell, the exact moment of cross wall fusion was captured (Fig. 5A). Two globular-like structures, one large and a much smaller one, were observed in the middle of the cell between the two membrane planes (Fig. 5B), apparently connecting the opposite HDZs. Although this cell looks notably compressed, it is unlikely that compression induced the formation of such structures, because compression is homogenous in space and has no effect on the width of structures perpendicularly oriented to the cutting direction (Dubochet et al., 1988; Matias and Beveridge, 2005; Zuber et al., 2006). Cross wall growth in S. aureus proceeds in a manner similar to a closing iris until cross wall fusion completes the septum (Amako and Umeda, 1979; Giesbrecht et al., 1998), but the exact moment of fusion has never been observed before in its hydrated state. These last moments of amalgamation, as seen in Fig. 5, are crucial for sealing the gap between the two cells and it seems complicated with high input from the membrane. These final steps of amalgamative polymerization are crucial for the survival of each daughter cell because the typical polymer network cannot be maintained during the final annealing steps. A misstep here can lyse cells – polymer alignment and bonding must be subtly different from the norm at the centre of a septating cell and, for this reason, a septation scar can be seen by atomic force microscopy at the centre of each new cross wall (Touhami et al., 2004).

Figure 5.

Cell wall fusion. (A) Lower-magnification image of a section running through the middle of a septating cell. The box designates the region where the two tips of the septum have joined and is shown in higher magnification in B. Here, a globular structure connecting cross walls from opposite sides of the cell is seen. Bars represent 200 (A) and 50 nm (B).

Daughter cell separation

Division in Gram-positive bacteria consists of two sequential events, septation and cell separation. The traditional view of cell separation was simple – after the septum was complete, hydrolysis down its middle (i.e. mid-line) by autolysins proceeded from cell periphery to centre until the two daughter cells were separated (Beveridge, 1981). In S. aureus, division could be more complicated because pockets of hydrolysis (murosomes) have been seen in the septum (Giesbrecht et al., 1998). Now, with frozen-hydrated sections an entirely different view is seen: separation of the two septal planes occurs very early during septum ingrowth forming a mid-zone of low density, non-adhesive, soluble matter. Indeed, the two daughter cells would immediately start cell separation as ingrowth occurred (i.e. a constrictive division) except that the septum is held together by a HDZ at its initiation point at the cell periphery (Figs 3A and 4A–E). Analysis of this outer wall ‘bridge’ in frozen-hydrated cells and wall fragments showed it to be much thicker than the outer wall at the early stages of septation (Figs 3A, 4C and D, and 6A; Table 1). Such a thick outer wall at the septum likely protects cells against the potential suicidal action of autolysins, which are acting immediately below it as the septum grows. Once the septum was completed, the outer wall bridge grew thinner (Table 1). The large standard deviation seen during this later stage of septation seems to reflect the progressive stages in the thinning of the outer wall bridge before daughter cells split. Such reduction in the thickness of the outer wall bridge is seen in Fig. 6 until, eventually, the bridge is completely hydrolysed and the cells separate. Interestingly, as this bridge thins, the septal HDZs thicken, presumably to ensure that this new wall matrix can withstand full turgor pressure once cell separation occurs.

Figure 6.

Final stages of cell division seen at the cell outer surface. A–C show the progressive thickening of HDZs (white arrowheads) and reduction of the thickness of the peripheral bridges between cross walls (black arrowheads). Long arrows point out cutting directions. Bar represents 50 nm.

Discussion

CryoTEM of frozen-hydrated sections allowed frequent observation of septum formation in S. aureus and revealed five distinct zones of different densities within septa: two HDZs appearing ‘sandwiched’ between three low-density zones. Our frozen-hydrated data on the S. aureus septa indicate that the two HDZs are the peptidoglycan–teichoic acid–wall protein matrix of nascent cross walls, the two low-density regions adjacent to the membranes are periplasmic spaces that continue from the outer envelope region, and the septal mid-zone corresponds to a soluble zone confined between nascent cross walls (Fig. 7A). This interpretation is based on the following evidence: (i) the two zones adjacent to the membrane continue as an extension of the periplasmic space and is confined between a membrane and a cell wall matrix as seen at the cell envelope; (ii) like the periplasmic space, the zones adjacent to the membrane and the mid-zone of the cross wall have low density and no distinguishable structures within them; (iii) the HDZs of the septum were the only septal structures seen in cell wall fragments; (iv) contrast of the different septal zones was similar to the two zones of the cell envelope; and (v) large separations were observed between the two dark wall zones of the septum of wall fragments, which were devoid of internal structures.

Figure 7.

Cell wall growth in S. aureus.
A. As a result of the mode of cell wall growth, a periplasmic space (PS) is found not only at the cell surface between the plasma membrane (PM) and cell wall (or outer wall; OW), but also at the septum between the membrane and nascent cross walls (CW); additionally, a soluble mid-zone (MZ) develops in-between cross walls at the septum; figure is drawn out of scale to make visualization of enveloping structures easier.
B. Proposed model for the growth cell walls. Simultaneous insertion of new wall material to an ingrowing septum on the outer surface of the membrane by cell wall synthases (CWS) and breakdown of peptidoglycan fibres by cell wall hydrolases (CWH) on the opposite side of the septal tip allows the growth of cross walls from the outer wall (lighter grey areas indicate newly incorporated wall material; 1 and 2). A thicker outer wall between cross walls (arrowhead) would secure the cross wall and ensure the cell integrity at the division site against the potential suicidal action of autolysins. Upon septum completion at the centre of the cell (3), cell wall synthases at the septal tip would disperse along the septal membrane to add mass to cross walls so that their thicknesses reach that of the outer wall; concurrently, proteolytic processing of autolysins found at the outer face of the outer wall bridge would activate lytic enzymes to hydrolyse the thick peptidoglycan at this region of the outer wall, eventually allowing physical separation of daughter cells.

Our data on septum formation in S. aureus indicate that cross walls are initially synthesized at the tip of the ingrowing septum and that this tip rapidly separates into two separate planes. The tip is in immediate and intimate contact with the plasma membrane, whereas the opposite end of the septum possesses a bridge of wall material that cements the septum together at its periphery. All of these structures of the septum must be under the influence of turgor pressure. Clearly, this pressure can only affect regions where a pressure differential exists and the traditional view has been that only the outer peripheral wall has such a differential. We agree. Yet, the low density of the mid-zone presents us with an enigma because it is newly discovered and (we suspect) composed of soluble constituents. It, like periplasms, could be compressible and therefore subjected to turgor pressure.

At tip of the septum, close proximity between the growing point and a region of autolytic activity could be necessary to direct the synchronized synthesis of each septal plane (at the same rate) to ensure equal symmetric growth of each. Here, it is tempting to envision the existence of a multienzyme complex composed of highly active PBPs, autolysins and other division proteins, but the visual complexity of these growth points is so complex (Fig. 5B) that differential densities are so far meaningless. Yet, the septum is a structural amalgam of newly assembled and rearranging polymers and we attempt to reconcile this together with our new structural data derived from frozen-hydrated sections in Fig. 7B. Here, we show where we believe the most active sites of assembly (PBPs) and disassembly (autolysins) reside. Interestingly, we have not been able to distinguish localized concentrations of so-called division proteins (e.g. polymerized tubulin analogues) from the general density of the cytoplasm surrounding division sites. This is strange because the high density of such assemblies should be readily apparent in our frozen-hydrated sections.

In the past decade, there have been many studies reporting on the septal localization of proteins involved in cell division, which assemble a cytoplasmic apparatus via the polymerization of the tubulin-like FtsZ protein, resulting in a large cytoplasmic protein complex named the FtsZ ring (Errington et al., 2003). As mentioned before, we have not been able to distinguish anything resembling a cytoplasmic tubulin-like structure. Yet, Zuber et al. (2006) have recently observed contrasted dots on the inside face of the membrane at the septal tip in other Gram-positive cocci; these could be the first images of polymerized FtsZ proteins using cryoTEM but more unambiguous identification is necessary. Septal rings are routinely observed with the use of FtsZ–green fluorescent protein fusions and fluorescence microscopy, which give rather low-resolution images (Nanninga, 1998). Nevertheless, such techniques have shown that cell wall synthesis occurs specifically at the septum in S. aureus, and that FtsZ recruits PBP2 to this site (Pinho and Errington, 2003).

At the peripheral surface of the septum, a thick bridge was observed between cross walls and should be necessary to resist turgor pressure at this potentially vulnerable region, where autolysins likely assist in cross wall growth and eventually trigger cell partitioning. This bridge is also necessary to hold the septum together as an intact structure during its ingrowth.

The growth of cross walls in S. aureus as two separate planes should greatly facilitate cell partitioning. After all, because the septum has been split and is only held together at the peripheral bridges, these same bridges need only be hydrolysed and the daughter cells would separate. There are only two peptidoglycan hydrolases involved in cell separation: Atl, a bifunctional protein that through proteolytic processing generates two extracellular peptidoglycan hydrolases, and Sle1, an N-acetylmuramyl-L-alanine amidase (Kajimura et al., 2005). Atl positions itself as a ring around the cell that corresponds to the zone of division (Yamada et al., 1996). Interestingly, double mutants lacking both Atl and Sle1 form irregular clusters with significantly impaired growth. This contrasts with other Gram-positive bacteria that always divide through the same medial plane, most notably B. subtilis and Streptococcus pneunomiae, in which the inactivation of genes for peptidoglycan hydrolases associated with cell separation causes filamentation of cells but without major reduction in the growth rate (Margot et al., 1999; Ohnishi et al., 1999; De Las Rivas et al., 2002). In fact, a recent cryoTEM study of frozen-hydrated sections suggests there are differences in the mode of cell separation between staphylococci and other Gram-positives, such as Streptococcus gordonii and Enterococcus gallinarum (Zuber et al., 2006).

Experimental procedures

Bacterial strains and growth conditions

Staphylococcus aureus Newman D2C was grown at 37°C to a mid-exponential growth phase (optical density at 470 nm, 0.5–0.8) in either trypticase soy broth (TSB) or TSB containing 10% (w/w) glycerol (used as a cryo-protectant). Harvested cells (centrifuged at 6000 g for 5 min) were washed three times in 50 mM HEPES (N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid, pH 7.0) or in 10 mM HEPES containing 10% (w/w) glycerol. The pellet of cells grown in TSB with glycerol and washed in buffered glycerol was directly used for freezing.

Isolation of cell wall fragments

Cell wall fragments were isolated as described previously (Matias and Beveridge, 2006). Briefly, S. aureus grown in TSB was resuspended in 50 mM HEPES (pH 7.0) containing 50 mg ml−1 DNase and 50 mg ml−1 RNase, and 0.5 ml aliquots of the pellet was added to 1.5 ml of 0.1 mm zirconia/silica beads in 2 ml tubes. Bacteria were mechanically broken by two cycles of 1 min agitation at 4600 r.p.m. in a Bead Beater (Biospec), with cooling of the tubes on ice between agitations. The supernatant was centrifuged to remove intact cells and residual beads (3000 g). The remaining supernatant, containing the cell wall fragments, was boiled in 4% (wt/vol) SDS for 2 h (Sprott et al., 1994). Wall fragments were washed five times in deionized water and resuspended with an equal volume of 40% dextran (42 kDa) in 20 mM HEPES (pH 7) for freezing.

Freezing and sectioning of bacteria

Pellets of bacteria or wall fragments were injected into thin copper tubes and immediately vitrified using a Leica EM PACT high-pressure freezer. 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 1000-mesh copper grids (Matias et al., 2003; Matias and Beveridge, 2005).

Freeze-substitution

For freeze-substitution of cells, copper tubes with frozen bacteria were cut, under liquid nitrogen, into roughly 2-mm-long pieces. These pieces were transferred to small vials containing 0.5 ml of freeze-substitution medium (2% osmium tetroxide, 2% uranyl acetate in anhydrous ethanol), and vials were transferred to a Leica AFS. Freeze-substitution was carried out as described earlier (Matias and Beveridge, 2006).

CryoTEM

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 cryoTEM operating at 120 kV. Energy filtering improves image contrast by eliminating inelastically scattered electrons, which causes a blurring effect on micrographs. Zero-loss energy filtered images were taken using either a 1024 × 1024 or a 2048 × 2048 pixel slow-scan CCD camera (Proscan). Freeze-substituted cells were observed at room temperature, under the same operating conditions. Images were stored and analysed using analySIS (SIS, Munster, Germany) software.

Length measurements

Length measurements were corrected for compression in frozen-hydrated sections as described by Zuber et al. (2005). Compression reduces the section length in the cutting direction with a corresponding increase in section thickness. This is why frozen-hydrated sections of spherical specimens, like S. aureus cells, appear elliptical and not circular. During cryo-sectioning, the long axis of the ellipse (perpendicular to the cutting direction) remains unchanged, whereas the axis parallel to the cutting direction becomes shortened. Compression (c) was calculated from the following equation: c = 1 − (d1/d2), where d1 and d2 are, respectively, the short and long axes of the ellipse measured on many cells and wall fragments. Average compression was 0.32 ± 0.05 (n = 20) for cells, and 0.357 ± 0.086 (n = 20) for wall fragments. Size measurements of envelope structures were corrected for compression according to the equation

image

where d′ and d are, respectively, the measured and corrected dimensions, c is the compression and α is the angle between the measured dimension and the cutting direction. The latter equation shows that compression is maximal when the measured structure is in the same direction of cutting (α = 0°) and that compression does not affect measurements of structures perpendicularly oriented to the cutting direction (α = 90°; i.e. in the least deformed regions of cells; Matias and Beveridge, 2005; 2006).

Acknowledgements

This work was supported by a Natural Science and Engineering Research Council of Canada (NSERC) Discovery grant to T.J.B. Microscopy was performed in the NSERC Guelph Regional Integrated Imaging Facility (GRIIF), which is partially funded by an NSERC Major Facility Access grant to T.J.B. During part of this study, VRFM was supported by funds from AFMnet-NCE to T.J.B.

Ancillary