Chapter 6

The Murein Sacculus

JAMES T. PARK

INTRODUCTION

The discovery of the chemically unique, rigid structural component of bacterial cell walls, termed the murein sacculus, is a relatively recent event in the history of microbiology. In 1949, a review of the subject could only characterize the structural unit as alkali insoluble, and it was speculated that perhaps the material was chitinlike since it contained glucosamine. The elucidation of the structure of the cell wall, its biosynthesis, and the discovery of how many antibiotics, including the β-lactams, interfere with its biosynthesis has developed over the ensuing decades into a fascinating story. The chemical characterization of murein confirms the unity and diversity among cellular forms of life as seen even in rigid, structural materials. In plants, cellulose, a β-1,4-linked polymer of glucose, is a major structural material. In the animal kingdom, most invertebrates rely for their exoskeletons on chitin, a β-1,4-linked polymer of N-acetylglucosamine (GlcNAc). And in the realm of bacteria, there is the murein, yet another β-1,4-linked polymer, a heteropolysaccharide containing novel peptides that are species specific in composition. This structural material, murein or peptidoglycan, is a universal cell wall component. Murein is absolutely essential to preserve the integrity of the cytoplasmic membrane from rupture in medium of low osmolality.

The murein sacculus is a truly unique case of molecular architecture in nature because it changes shape during the cell division cycle while continuously being held together by covalent bonds. It was recognized as early as 1964 that morphogenesis of a bacterial cell during growth and division must involve hydrolysis of covalent bonds in the murein sacculus and insertion of new material (62). Only then could both elongation and septation occur. In subsequent years a great deal has been learned about the structure of murein and the enzymes of Escherichia coli involved in murein synthesis. However, relatively little progress has been made in understanding the biochemistry of morphogenesis, i.e., the specific tasks performed by individual enzymes during cell growth and cell division and how the processes are controlled to maintain an intact sacculus during these processes.

This chapter is concerned with the structure of murein and how the murein sacculus changes during growth. For this reason, only the final stages in the synthesis of murein and what is known about the biochemistry of morphogenesis of E. coli will be considered. A detailed discussion of the enzymology of murein synthesis is presented in chapter 68. A useful book devoted to a fuller discussion of the subject of bacterial cell walls was recently published (21).

TERMINOLOGY

The rigid, shape-determining structure in bacterial cell walls is a complex polymer consisting of two amino sugars and at least four amino acids. The term murein, derived from murus (Latin for "wall"), was introduced by Weidel and Pelzer (62) as a trivial name for these newly recognized cell wall polymers. Since the polymer contains roughly equal amounts of polysaccharides and peptides, it belongs to the general class of polymers called peptidoglycans. Thus, mureins represent a large group of closely related peptidoglycans found exclusively in bacterial cell walls. Strictly defined, the term murein bears the same relation to the term peptidoglycan as, for instance, collagen does to protein. However, because practically all known peptidoglycans are mureins, the two terms have come to be used interchangeably.

Mureins are complex heteropolysaccharides. The repeating unit (and slight modifications thereof) has been termed muropeptide (62). Weidel and Pelzer (62) also introduced the term murein sacculus to indicate that the murein of a cell actually exists as one giant molecule in the form of "a rigid bag of the volume and shape of the cell."

CHEMICAL STRUCTURE OF THE MUREIN

The murein of E. coli is composed of GlcNAc, N-acetylmuramic acid (MurNAc; GlcNAc with d-lactic acid ether substituted at C-3), l-alanine, d-glutamic acid, meso-diaminopimelic acid (DAP), and d-alanine, all in equimolar amounts except for d-alanine, which may be present in slight excess or deficit. These components form the basic muropeptide repeating unit of the murein (Fig. 1). The sugars are linked together by β1→ 4 glycosidic bonds. Attached to the carboxyl group of each muramic acid by an amide linkage is a short peptide, l-alanyl-d-isoglutamyl-l-meso-diaminopimelyl-d-alanine. In the murein of E. coli a small percentage of the peptides lack d-alanine, and an even smaller percentage terminate with an additional d-alanine. This composition is shared by nearly all gram-negative bacteria as well as by a few gram-positive rods (54). In some species the carboxyl groups of glutamic acid or DAP (or both) may be amidated, but in E. coli neither is. A unique feature of the peptide backbone of all mureins is the alternating sequence of optical isomers d-l-d-l-d, beginning with the d-lactyl group of muramic acid and including the l configuration of DAP.

Fig. 1Chemical structure of the principal muropeptide monomer present in the murein of E. coli and other gram-negative bacteria (C6 of Weidel and Pelzer [62]), showing the favored ringlike structure of the peptide proposed by Barnickel et al. (3). NAcGlc, N-acetylglucosamine; NAcMur, N-acetylmuramic acid; Ac, acetyl.

Source:

The arrangement of muropeptides in the murein of E. coli is shown in Fig. 2. The sugars form linear chains of alternating units of GlcNAc and MurNAc linked β-1→ 4. The MurNAc at the end of each strand is present as a nonreducing 1,6-anhydro sugar (32). The average glycan strand, based on the 1,6-anhydromuramic acid (1,6-anhMurNAc) content of sacculi, is about 30 muropeptides in length (10, 22, 24). Initially, the strands are probably much longer than 30 and become shorter with time since 70% of the glycans have an average length of 9 disaccharide units (28).

Fig. 2Arrangement of muropeptides and glycan strands in the murein sacculus. M, MurNAc;, G, GlcNAc; M*, 1,6-anhMurNAc; , peptide.

Source:

A molecule of lipoprotein is attached to about every tenth muropeptide (8). The covalent link is from the l-carboxyl group of DAP to the epsilon amino group of the carboxy-terminal lysine of the lipoprotein.

Adjacent strands are cross-linked to each other through the peptide side chains. Most of these cross-links are between the carboxyl group of the d-alanine in position 4 of one peptide and the free amino group of DAP of a muropeptide in the adjacent strand. However, it has been shown that up to 20% of the cross-links do not involve d-alanine but, rather, are direct links between the DAP residues of neighboring chains (22). About half of the muropeptides of the sacculus are involved in cross-links, and another 5% are involved in links that hold three muropeptides together (23). These estimates are based on the proportions of muropeptide monomers, dimers, and trimers present in the digests of murein sacculi treated with egg white lysozyme or another β-N-acetylmuramidase produced by the fungus Chalaropsis. These enzymes hydrolyze the β-1→ 4 bonds between C-1 of MurNAc and C-4 of GlcNAc, thereby reducing the glycan chains to disaccharide units. After enzymatic hydrolysis, muropeptides that are cross-linked are present as muropeptide dimers or trimers.

In early studies in which the components of lysozyme digests were characterized, eight components could be separated by paper chromatography. They were designated C1 through C8. The two principal components were C6, the muropeptide monomer consisting of a disaccharide tetrapeptide, and C3, a dimer composed of two C6 muropeptides cross-linked to each other. Also present were lesser amounts of C5, a disaccharide tripeptide, and C4, a dimer identical in composition to C3 but differing in that the two disaccharides of the dimer were glycosidically linked. The cyclic dimer is now known to be an artifact of digestion with egg white lysozyme, since the glycan chains of sacculi are completely hydrolyzed to disaccharides by Chalaropsis β-N-acetylmuramidase. More recently, Glauner and Schwarz (22) have fractionated Chalaropsis digests of E. coli sacculi into 60 components by using high-pressure liquid chromatography. They have characterized 33 of them, which includes nearly all of the components that represent at least 0.1% each of the total muropeptides of the digest. This great number of components is largely accounted for by the fact that monomers, dimers, and trimers are present not only in unmodified form but are also at the ends of chains. Hence these components contain a disaccharide with 1,6-anhMurNAc. Lipoprotein is also found attached to monomers, dimers, and trimers. Adding to the complexity of the Chalaropsis digests is the fact that the peptides of any given component may contain two, one, or no d-alanine residues. Based on these results, it appears that chain termination and lipoprotein attachment occur more or less randomly, irrespective of the location of cross-links. An exception is that lipoprotein is found attached to DAP-DAP linked muropeptide dimers but not to d-Ala–DAP linked dimers. Five of the characterized components contained glycine instead of d-alanine in either position 4 or position 5 of the peptide, indicating a lack of absolute specificity in the synthesis of the precursor pentapeptide.

MOLECULAR ARCHITECTURE OF THE MUREIN

Based on the DAP content and size, the sacculus of E. coli is believed to consist mainly of a single monomolecular sheet of murein (7, 62). Analysis by neutron and X-ray small-angle scattering techniques indicates that as much as 80% of the sacculus exists as a monolayer and that the remainder is triple layered (39). Other data are consistent with at least some of the sacculus having more than one layer (40, 48). The orientation of the sugars and peptide side chains in strands of peptidoglycan remains a matter of conjecture. This is due primarily to the fact that the material does not have a crystalline structure, so X-ray diffraction patterns lack detail. Nonetheless, Debye-Scherrer rings indicating periodicities of about 1 nm and 0.44 nm have been reported (20). These patterns are believed to reflect the length of the disaccharide unit in the strand, 1.03 nm, and the distance between strands. However, this polymer is remarkably flexible. Even a multilayered, highly cross-linked murein such as that found in staphylococci can undergo considerable swelling and shrinking in response to the osmotic strength of the medium in which the bacteria are suspended (38). The scatter corresponding to 0.44-nm repeat distances is almost the same as the interchain distance found in chitin, where the chains are tightly hydrogen bonded to each other. However, in murein this is probably the minimum distance between chains when sacculi are in the dry state. In fact, a more extended structure for the muropeptide cross-links seems likely in an aqueous environment. Based on the estimates of the actual muropeptide content of sacculi by Braun et al. (7), the average distance between chains would be 1.25 nm. Initially, most models for the structure of murein assumed a chitinlike conformation in which the sugar rings present a flat surface with all the peptide side chains pointing in one direction and the glycan chains closely stacked and stabilized by hydrogen bonds (for a review, see reference 50). It has been recognized for some time that certain properties of murein are not readily compatible with this type of model, namely, the elasticity of the polymer and its susceptibility to hydrolysis by egg white lysozyme and other β-muramidases.

MOLECULAR ARCHITECTURE OF A GLYCAN STRAND OF THE MUREIN

A model proposed by Barnickel et al. (4) seems to overcome these objections and, in addition, provides a plausible explanation for the facts that in E. coli only half of the muropeptides participate in cross-links and that in gram-positive bacteria sacculi consist of multilayered sheets of murein cross-linked to each other. In this model the sugars do not lie in a flat plane but are tilted with respect to each other to accommodate the lactic acid moiety of muramic acid. Simply put, muropeptides are twisted in such a way that each strand of murein is actually a helix with each peptide side chain being exposed approximately 90° around the helix relative to its neighbors. With this configuration, only every fourth muropeptide would be positioned to form a cross-link with a strand on its left. Similarly, a cross-link would occur with a strand on the right only with every fourth muropeptide whose peptide is pointing to the right (Fig. 2). Three of every four muropeptides are on opposite sides of a given strand relative to their immediate neighbors and hence may not readily be linked to the same neighboring strand. Considering the strands on each side, this permits 50% of the muropeptides to form cross-links, which is roughly the number observed in E. coli.

ARRANGEMENT OF GLYCAN STRANDS IN THE MUREIN SACCULUS

It is implicit from Fig. 2 that glycan strands run parallel to each other. Although this arrangement is consistent with the formation of an orderly cross-linked monolayer, there is no direct evidence for it. However, there is indirect evidence indicating not only that the glycan strands are parallel to each other, but also that they are perpendicular to the axis of the cell and hence follow the circumference of the cell’s cylinder. This was shown by partial digestion of sacculi with a penicillin-insensitive endopeptidase which hydrolyzes the cross-links (60) to reveal the individual glycan strands to be predominantly oriented perpendicular to the rod and by demonstration that sonication ruptures sacculi perpendicular to their axis (59). Since about 70% of the strands average only 9 nm in length and the remainder average 45 nm in length (28), at least 80 strands with an overall average length of 30 nm, aligned end to end, are needed to span a circumference of 2,500 nm once. Since adjacent strands are cross-linked to each other, a continuous rigid layer is formed. It is not known whether the short strands are exactly perpendicular to the axis of the cell and hence appear to form rings around the circumference, or whether slight pitch exists so that, in effect, they form a helix.

The arrangement of murein strands at the poles of the cell is unknown. It seems probable that as the septa grow inwards to form the new poles of daughter cells, strands are anchored to the existing sacculus by transpeptidation. This would form a template for continued ingrowth by laying down a series of ever-smaller concentric rings of murein. Since septa consist of two new poles being formed simultaneously, the question arises as to whether the two developing poles are covalently linked to each other at any time during their formation. The few published electron micrographs of septa do not provide a definite answer to this question, though fragmented sacculi held together by their septa have been seen. At some point in the division process, the new cell poles are clearly separated by an unexplained space of uniform width that is observed in cells of the envA mutant, a chain-forming strain (9). Burdett and Murray (9) also demonstrated that the outer membrane does not invaginate simultaneously with the inner membrane and the sacculus. If the space separating the new, adjacent poles were occupied by a few molecules of a murein-associated enzyme, this might account for the remarkably fixed distance observed between the two poles formed in the septum.

Electron micrographs of another chain-forming, division-defective mutant of E. coli provide an entirely different picture. The septa of the envC mutant, E. coli PM61, appear fused in some regions and irregularly separated in other parts of the same septum (Fig. 3) (49). This suggests a sequence for septum formation in which the two poles are initially fused and then separated by a specific enzyme (amidase or endopeptidase) to allow insertion of new outer membrane. In normal E. coli cells these processes occur in rapid succession; thus, suggestions for such a sequence may only be seen in chain-forming strains.

Fig. 3E. coli PM61 (envC), showing apparent fusion of the murein of the two cell poles in the septum.

Source:

RELATIONSHIP OF THE MUREIN SACCULUS TO THE OUTER MEMBRANE AND THE CYTOPLASMIC MEMBRANE

The murein sacculus is located in the periplasm, the space between the outer membrane and the cytoplasmic membrane. When examined by electron microscopy of thin sections fixed and stained by conventional methods, the sacculus of E. coli is usually seen as a dense layer 2 to 3 nm thick, associated with the outer membrane and separated from the cytoplasmic membrane. However, micrographs of specimens prepared by less disruptive methods show that the murein layer may be from 7 nm to 15 nm thick and that the space between the two membranes is of constant thickness (1, 19, 29, 40) (Fig. 4). This could indicate that murein is a double layer or, alternatively, it may simply reflect the native state of hydration of the polymer. The outer membrane is presumed to be firmly anchored to the sacculus by the covalently linked lipoprotein molecules that are embedded in the outer membrane. Additional tight association of the outer membrane and the sacculus may be provided by the outer membrane porins, especially the ompC and ompF gene products, which are known to remain associated with the murein sacculus even after envelopes are heated at 60°C with sodium dodecyl sulfate (53). Altogether, the lipoproteins and porins provide over 400,000 possible contacts between the murein sacculus and the outer membrane.

Fig. 4(a) Thin section of E. coli, showing the uniform distance between the outer and inner membranes and the thick, space-filling layer containing the murein in the periplasmic space. Reprinted by permission from reference 29. (b) Thin section of E. coli, again showing, by a different method, a relatively thick murein layer.

Source: Reprinted by permission from reference 40.

Many years ago it was shown that the cytoplasmic membrane of growing cells is attached to the outer membrane at a few hundred sites termed adhesion or fusion sites (5). Actual fusion of the two membranes at these sites has never been seen by electron microscopy and may in fact not occur. It therefore seems plausible that many of these "fusion sites" represent places where new strands of murein are being covalently attached to the murein sacculus while their growing ends are still attached to bactoprenolpyrophosphate, the carrier lipid present in the cytoplasmic membrane that is involved in murein and lipopolysaccharide synthesis (43, 46). As is shown below, about 200 such connections between the sacculus and the cytoplasmic membrane may exist in growing cells. Similarly, other "fusion sites" may represent places of synthesis of lipopolysaccharides or outer membrane proteins.

PROCESSES INVOLVING CHANGES IN THE MUREIN SACCULUS

Elongation

Each generation, the cell doubles the length of its murein sacculus by an elongation process that is enzymatically distinct from septation. An early reason for believing that elongation and septation involve separate pathways for synthesis and incorporation of murein is that septation can be blocked by a variety of means without affecting elongation (55). It seems self-evident that the septum must be made entirely of newly synthesized murein, but elongation could occur either by intercalation of new murein or by outgrowth of new wall from a fixed site. This problem is addressed in the next paragraph. There is evidence suggesting that the two processes proceed separately and alternately. Autoradiographic measurements of the distribution of [³H]DAP in the sacculi of pulse-labeled cells have demonstrated that the rate of elongation slows during the period of septum formation (64). This suggests that some needed component of the elongation machinery is not fully active during septation. Based on certain assumptions about the condition of the cells during the cell division cycle, the observed variation in the rate of murein synthesis has been predicted and described mathematically (13, 14) (see chapter 104). Conversely, when elongation takes place, murein synthesis by the septation pathway is shut down, as shown by the following. Penicillin-binding protein 3 (PBP 3) is the enzyme primarily responsible for murein synthesis during septation (6). When furazlocillin, which specifically binds and inactivates PBP 3, is added to dividing cells, murein synthesis is inhibited by up to 40%, but it causes no inhibition when added to filamenting cells (6).

Heteropolysaccharide synthesis generally involves the assembly of a lipid-linked repeating unit of the polymer on the inner side of the cytoplasmic membrane, followed by polymerization and finally release of the polysaccharide on the other side of the membrane. The end result is the random extrusion of material to form a slime or capsule, or, with some degree of organization or self-assembly, to produce a rigid, structural unit such as those containing cellulose or chitin. Murein, in contrast, though initially synthesized by the rules of heteropolysaccharide synthesis, must be assembled according to a specific plan that allows the covalently closed sacculus to enlarge (i.e., elongate) and then divide into two separate sacculi without loss of integrity. Experiments designed to determine the fate of newly formed strands have revealed some aspects of the plan used by E. coli to achieve elongation (11, 12).

Growing cells were pulse-labeled with [¹⁴C]DAP or [³H]DAP. Dimers representing the cross-links between two adjacent strands of murein were isolated from the sacculi following digestion with a muramidase that cleaves the β-1,4 bond between carbon 1 of muramic acid and carbon 4 of glucosamine. The distribution of label between the two halves of the dimers was then determined. The two muropeptides of the dimer are termed donor and acceptor, where the donor represents the newly formed strand of murein, since it contains the d-Ala-d-Ala bond needed to drive the cross-linking reaction. Also, the half-life of pentapeptides containing d-Ala-d-Ala in murein is less than 1 min (17) (see chapter 68 for details of biosynthesis pathway). The original data based on experiments of this type indicated that two strands might be inserted together (12), provided that no septal murein was being synthesized—an assumption that proved to be incorrect (44). Data obtained with nondividing cells indicated that initially about 95% of the [¹⁴C]DAP was in the donor position (15, 44). This represents strong evidence that new strands are inserted one at a time. The percentage of label present in the donor position remains constant for 8 min (12) and then gradually increases in the acceptor position. During the first 8 min of growth, in cells with a generation time of 48 min, about 400,000 muropeptides are inserted into the sacculus of each cell. This value is equivalent in length to about 180 times the circumference of the cell, so the relatively constant ratio of donor to acceptor was surprising since it indicates that few of the new strands were cross-linked to each other. One way to explain this observation is that new strands are inserted simultaneously at about 200 sites, each separated from other insertions by preexisting murein (see model, Fig. 5). An average cell would have over 1,100 hoops of glycan strands spaced about 1.25 nm apart, excluding the poles of the cell. Thus, 200 new strands inserted randomly would on average be 5 or 6 strands apart. The rise in acceptor radioactivity after the first 8 min, however, suggests that for a given enzyme complex laying down a new strand the odds are quite high that it will make contact with the strand it inserted 8 min earlier. This indicates to me that a new strand is continuous, but that subsequently breaks are caused by the action of a transglycosylase, and that strand insertion is more likely to follow a given groove of two strands rather than jump to a different groove. Thus synthesis and insertion are processive. The essential features of the elongation process are that (i) each of about 200 separate enzyme complexes independently synthesizes and inserts single strands of murein, and (ii) the enzymes involved, which are integral membrane proteins, move unidirectionally around the circumference of the cell about once every 8 min, inserting new strands. After 8 min, new strands begin to be cross-linked to the radioactive strands inserted 8 min earlier, and hence relatively more radioactivity appears in acceptors (Fig. 5).

Fig. 5Structure of the murein at a growth site for elongation of the murein sacculus as it may exist during a pulse-labeling experiment. The stippled, helical strand represents a new radioactive strand of murein inserted between preexisting strands initially (A), and later when, as a result of continued insertion along a helical path, a radioactive strand contacts the radioactive strand formed 8 min earlier (B). The arrows between strands indicate the direction of cross-links in dimers from donor to acceptor strands.

Source:

Given that the covalently closed murein sacculus must be maintained intact to protect the cell from osmotic rupture, the question arises: how can the cell elongate without ever experiencing breaks in its sacculus? One way to do this, proposed by Koch (37), is the so-called "make-before-break" mechanism. To do this the cell would add a new strand underneath the monolayer of murein and cross-link it to each of two adjacent strands which are under stress. Only then would an enzyme cleave between the two adjacent strands in the sacculus and introduce the new strand into the stress plane. The difficulty with this approach is that it requires a very special enzyme that cleaves cross-links that are under stress only and at the same time recognizes only those stressed cross-links that are underlain by a new strand. Since there are several enzymes in the vicinity of the sacculus that not only can cleave unstressed murein, but that are likely to cleave bonds under stress even more readily, I find the make-before-break idea highly unlikely. A variation of this idea has been put forward by Holtje (30), in which three strands formed simultaneously underneath the existing sacculus are cross-linked to each other and to two strands of the sacculus that are separated from each other by one strand. The "in-between" strand is then removed by the combined action of endopeptidases and transglycosylases and the three new strands become members of the stressed layer. In addition to the special requirements of any make-before-break model, the three-strand insertion model is untenable because it is well established that only one strand is inserted at a time during elongation of the sacculus (15, 16, 44). However, as will be noted later, producing multiple strands adjacent to each other simultaneously and then removing one or more, as suggested by Holtje, is an appealing method to achieve septation and cell separation.

Concern about preserving all cross-links at all times is probably overdone. It seems likely to me that the cell can manage with less than a normal number of cross-links in its sacculus. There are about 3.5 × 10⁶ muropeptides in a sacculus (63) and only about 700,000 cross-links so that considerably less than half of the muropeptides are involved in cross-links. Certainly the cell manages very well without its sacculus being completely cross-linked. Thus it is difficult to argue that make-before-break is necessary. A simpler way to elongate the murein sacculus would be to insert new strands after a very few cross-links are cut, and the new strand is inserted and cross-linked to neighboring strands to immediately take up the stress. It would require that an endopeptidase, such as PBP 4 (21), associate and work in conjunction with the two-headed enzyme, presumably PBP 2 (47), that makes the new strands and cross-links them to old strands during elongation.

In summary, elongation of the murein sacculus is a multisite process, which involves the insertion of strands at about 200 separate locations simultaneously. The enzymes responsible move continuously in one direction around the circumference. This is perhaps the first example of directed movement of membrane-bound biosynthetic enzymes.

Septation: Formation of Polar Caps

Septation is a very complex process that involves both temporal and topological controls. Two entirely new polar caps of murein must be formed, and the murein sacculi of the two daughter cells must be cleaved from each other by one or more murein hydrolases without loss of integrity to either. As summarized in an excellent review by Donachie et al. (18), as many as 15 genes may be required for septum formation, two for cell separation, and possibly two more for initiation of septation (see also chapter 101). Seven genes are most clearly implicated in septation, because mutants therein are temperature-sensitive filament formers (fts). These are: ftsI, ftsQ, ftsA, ftsZ, ftsW, ftsN, and ftsL (mraR). What is known of the role of these genes in cell division is discussed in chapter 101. One of these genes codes for a PBP (53): ftsI, also known as sep or pbpB (2), is the structural gene for PBP 3. The amino acid sequence derived from the DNA sequence of this gene suggests that it is an ectoprotein anchored in the cytoplasmic membrane by its amino terminus (41). Ishino and Matsuhashi (33) have shown that it has both transpeptidase and transglycosylase activity. This is the only PBP clearly essential for survival, in the sense that all ftsI mutants are unable to form colonies under restrictive conditions. The mutants are unable to divide under these conditions and hence form long filaments, indicating that elongation continues but septation is blocked. Certain β-lactam antibiotics such as piperacillin and furazlocillin, which selectively bind to PBP 3 at the MIC, appear to specifically inhibit synthesis of the murein laid down during septum formation (6).

As mentioned above, a process by which multiple strands are laid down in ever smaller concentric circles, employing a multienzyme complex with several PBP 3 molecules (a "divisome"; 42) operating at the leading edge of the constriction, would lead to two new multilayered cell poles fused to each other. These can then be separated from each other by enzymatic hydrolysis with an endopeptidase or amidase. In keeping with this notion, Romeis and Holtje have demonstrated that PBP 3 and PBP 7 are selectively adsorbed by a protein-affinity column containing lytic transglycosylase (51). PBP 7 has been shown to be a d,d-endopeptidase which cleaves the cross-links in intact sacculi (52). Thus it can be envisioned that a divisome that includes PBP 3, PBP 7, and lytic transglycosylase might produce multiple strands to form the new poles and then separate the daughter cells by removing one or more of the strands with PBP 7 and the lytic transglycosylase. In support of the idea that strands of murein are hydrolyzed during cell separation, it has been reported that murein turnover is greater in normally growing cells than in filamenting cells, indicating that considerable losses occur during cell separation (44). This speculative view only pertains to pole formation and cell separation: the assembly and activation of the divisome presumably involves several other fts gene products, such as FtsZ, FtsA, FtsQ, FtsL, etc. (see chapter 101). The determination of the biochemical function of these gene products in cell division is one of the more difficult but fascinating chapters that remains to be written about E. coli. At present, it is a little like knowing the members of the orchestra, but not what instrument each plays or how they cooperate to make music.

Maintenance of Rod Shape and the Role of PBP 2

Another mystery of the murein sacculus story is how the cell retains the rod shape of its sacculus during growth and division. It has been known since 1975 that PBP 2 is required for continued elongation of the cylindrical wall (57). In fact, PBP 2 appears to be required for most of the net synthesis of murein during elongation (45). Loss of PBP 2 function by mutation causes E. coli to grow as spherical cells (56). Inactivation by mecillinam, which binds selectively to PBP 2, causes cells to gradually change shape, becoming large spheres which eventually lyse (57). In the presence of mecillinam, before becoming fully spherical most cells divide once while enlarging around the division site so that they assume a pearlike shape, which suggests that some murein continues to be formed by the septation pathway. PBP 2 has been shown to have transpeptidase activity (34). Surprisingly, this enzyme can initiate elongation in stationary-phase cells given fresh medium, and the elongation proceeds at a normal rate for one generation even when all other PBPs are inactivated (47). A dilemma must be noted here. Although PBP 2 seems to be responsible for most murein synthesis during elongation, there appear to be only 20 or so molecules of this protein per cell as measured by binding studies with radioactive penicillin (56). In the discussion of elongation above, reasons were given for believing that 200 enzyme complexes were inserting new murein strands at separate sites. Clearly 20 PBP 2 molecules cannot be in 200 complexes simultaneously. Either there are only 20 elongation strand factories, or many more PBP 2 molecules exist which cannot be detected by binding studies. Quantitation of the number of molecules with anti-PBP 2 antibody would clarify this situation.

While it is clear that PBP 2 is necessary to preserve the cylindrical shape of the sacculus, the way in which this is accomplished is far from being understood. Yet another surprise regarding PBP 2 is that it is essential in wild-type E. coli but not under conditions where ppGpp is overproduced (61). It has been suggested that PBP 2 has a nonessential function involving elongation and maintenance of rod shape plus a vital function that is in some way related to cell division (61). Thus, PBP 2 remains an enigma and may well hold the key to eventual understanding of the biochemistry of the elongation process.

"Recycling," i.e., Reutilization of Murein Sacculus Components

In 1985 Goodell (25) reported the unexpected finding that E. coli breaks down almost 50% of its murein in each generation and reutilizes, i.e., recycles, the intact tripeptide, l-Ala-d-Glu-DAP, to form more murein. It was subsequently shown that an oligopeptide permease (Opp) was required for the uptake of the tripeptide and that, following transport into the cell via Opp, the tripeptide was incorporated into the sacculus without prior degradation (26). Then in 1993 it was reported that recycling took place just as efficiently in the absence of Opp (45)! Therefore, some other uptake mechanism must be involved in the recycling pathway. Components of the alternative pathway, though not recognized as such, had already been studied since two proteins involved, AmpD and AmpG, are required in an unknown way in the induction of β-lactamase (35). As illustrated in Fig. 6, lytic transglycosylases present in the periplasm hydrolyze up to 50% of the murein each generation to form GlcNAc-anhMurNAc-tripeptide, presumably with the aid of specific periplasmic endopeptidases and an l,d-carboxypeptidase (21). This muropeptide is transported into the cell by AmpG, a membrane protein required for β-lactamase induction. AmpD is a cytoplasmic protein needed in order for β-lactamase to be inducible. It was found to be an anhMurNAc–l-Ala amidase (31, 35, 36). Thus, tripeptide is released in the cytoplasm by AmpD and presumably reintroduced into the biosynthetic pathway for murein by direct addition to UDP-MurNAc. Another known cytoplasmic enzyme, β-N-acetylglucosaminidase (65), probably releases GlcNAc from the muropeptide so that it also can be reutilized. The fate of the anhMurNAc is unknown.

Fig. 6Proposed interconnected pathway for recycling muropeptides and for their involvement in ?-lactamase induction is illustrated. Symbols: , GlcNAc; , anhMurNAc; , MurNAc; , Ala; , Glu; , DAP.(A) The recycling pathway. As shown, murein is degraded by known enzymes in the periplasm to the fragments listed. The muropeptide, GlcNAc-anhMurNAc-tripeptide, is transported into the cytoplasm through AmpG. It is then degraded into GlcNAc-anhMurNAc and free tripeptide by AmpD. Alternatively, the ?-N-acetylglucosaminidase, a cytoplasmic enzyme, removes the GlcNAc of the GlcNAc-anhMurNAc-tripeptide, and the resulting anhMurNAc-tripeptide is hydrolyzed by AmpD to release tripeptide. Free tripeptide can then be added directly to UDP-MurNAc by an as yet unidentified enzyme, thereby reintroducing it into the biosynthetic pathway for murein synthesis. (B) Muropeptides as inducers of ?-lactamase. Intracellular accumulation of GlcNAc-anhMurNAc-tripeptide as a result of inactivation of ampD triggers production of Citrobacter freundii AmpC ?-lactamase. The muropeptides presumably bind to the transcriptional regulator AmpR and convert it into an activator for ampC expression. (C. freundii ampR and ampC are expressed from a plasmid.)

Source: Reprinted by permission from reference 35.

Recycling is not an essential function, since ampG mutants grow normally. The fact that E. coli possesses AmpG and AmpD even though it is noninducible and that these proteins are part of the β-lactamase induction pathway in certain bacteria suggests that recycling may play a role in sensing the condition of the murein sacculus. For this purpose, the concentration of muropeptide in the cytoplasm would reflect the status of the sacculus under various growth conditions. The possible regulatory role of recycling remains to be investigated.

UNFINISHED BUSINESS

In this chapter I have described the murein sacculus of E. coli and the various changes it undergoes during growth. The murein is a very complex molecule composed of a heteropolysaccharide with unique peptide side chains that are cross-linked to each other to form a covalently closed sacculus. Yet this polymer is in a very dynamic state. As the cell doubles in length, about 600,000 peptide cross-links are cleaved to allow intercalation of murein strands containing over 2 × 10⁶ new muropeptide units. The formation of the new poles of the daughter cells during the septation process involves the addition of another 10⁶ muropeptides. In addition to this activity, the cell is constantly degrading almost half of its wall material each generation only to recycle it to make more murein.

With the exception of the chemistry of murein, most aspects of the murein sacculus story need further investigation before they can be fully understood. If 20% of the sacculus is triple layered, is this in the poles or in patches of randomly dispersed material? Which enzyme forms the DAP-DAP cross-links? Which enzyme forms the covalent link between murein and lipoprotein? Given that PBP 2 and PBP 3 play the central roles in elongation and septation, what are the roles of the other glycan polymerase-transpeptidase two-headed enzymes, PBP 1A and PBP 1B? It should be noted here that the cell cannot survive if both PBP 1A and PBP 1B are nonfunctional (58). How does PBP 2 contribute to maintenance of the rod shape of the cell? Are there 200 or only 20 sites for incorporation of new murein during elongation? How is PBP 3 activated and inactivated during the cell division cycle? Does PBP 2 undergo regulation so as to be active only during elongation? How are the multitude of autolytic enzymes, the transglycosylases, endopeptidases, and amidases, controlled during growth and stationary phase so that the cell survives? Why does the cell break down 50% of its murein during every generation, only to reutilize most of its building blocks? Is recycling part of an elaborate signaling device? All of these questions and many more must be answered before it can be said that we know how E. coli grows.