Autolysis Control Hypotheses for Tolerance to Wall Antibiotics

Antibiotics are divided into two classes: bacteriostatic and bactericidal. The reasons for their classification are usually self-evident, although seldom proven. For example, chloramphenicol stops protein synthesis and is bacteriostatic, probably because when it is removed growth can recommence. Penicillin and congeners are bactericidal, probably because continued cytoplasmic synthesis should lead to increased cellular pressure and eventually to the rupture of the wall and to cell death. Lethality is not always the outcome, and certain strains may neither grow nor lyse in the presence of lactams (46–51, 55, 57, 60); they are called “tolerant.”

The conventional interpretation is that tolerant cells fail to mobilize or create the autolysins needed for enlargement and division. This idea dates back to the time of Shockman (51), Rogers et al. (48), and Tomasz and colleagues (56, 57). This interpretation has the problem that if cytoplasmic synthesis continues in the absence of autolysins, the turgor pressure should increase progressively because the volume of the cell cannot enlarge (16, 19, 21, 22) and lethality from physical forces seems inevitable. Additionally, because autolysins are essential for normal growth, cell expansion, and division, an autolytic system must function before antibiotic application. In this commentary, the autolysin-control hypotheses will be considered further, although these considerations are also purely hypothetical in nature.

The key property for growth of bacteria is that they should increase the cell wall area only if they can do it safely, Mechanisms that can be used to accomplish this have been suggested (9, 11, 12, 16, 22, 23, 59, 62).

I start with consideration of the functions of autolysins. I will not explore the biochemical details of wall hydrolytic function because it has been well reviewed (3, 10, 13, 53). I begin with a description of autolysin function during vegetative growth of Bacillus subtilis with no antibiotic present. This case was chosen since it illustrates two normal, essential, nondestructive functions of autolysins and how the interplay of synthesis and autolysis with wall chemistry and physics can lead to normal bacterial growth.

ROLE OF AUTOLYSINS DURING GROWTH OF A GRAM-POSITIVE ROD

Side wall elongation.

The roles for autolysins in the growth of B. subtilis are now clear. For cylindrical elongation they must function throughout the cell cycle (15, 17, 18, 21, 23, 26, 27, 28) to cleave the outermost layer of the side wall. New layers of peptidoglycan (also called murein) are added just outside the cytoplasmic membrane and just inside the existing layer of peptidoglycan. As additional layers are added, a given layer moves outward and is stretched as the cell grows longer. Eventually, the cell's autolysins dissolve the outermost murein; they do so most effectively when the murein is stretched as far as its elastic limit will permit. This process is called the “inside-to-outside” elongation mechanism. Now, the increase in tension as the wall moves outward and elongates appears to be important in favoring the autolysis of the outermost layers, but other factors are probably involved (for a list, see reference 21). This action of autolysins for cleavage of the external layer of the side-wall growth is their best-known activity.

Septal splitting to create poles.

The new pole of a gram-positive organism is formed by construction of a septum across the cell and then splitting of the septum by autolysin action. As the cross wall is formed, no stresses act on the septal murein because it is entirely internal to the cell. On splitting, however, one wall surface is outside and one wall surface is inside the cell, and consequently, differential osmotic forces will act. This causes each half septum to become a new pole as it is stretched and becomes nearly hemispherical. This shape change occurs without the addition of any new murein (24, 25). This second constructive function of the autolysins of B. subtilis in the precise bisection of septa to create new poles is simply due to physical forces on the substrate (29). However, how can an enzyme distinguish where it should act from where it should not act? Besides the tension at the bottom of the splitting septum favoring autolysin action, pH cues, due to chemiosmosis, guide the cleavage by the autolysin (29). In B. subtilis cells deficient in autolysin, the growth rates, both in volume and in length, are substantially unchanged. However, septal splitting is markedly reduced, producing filamentous growth. This suggests that cocci generally may need very much less autolysin activity than gram-positive rod-shaped bacteria.

Inertness of the murein of poles once formed.

The established poles of B. subtilis are not subject to turnover like the side wall is (this was established by the extensive work from R. J. Doyle's and R. Archibald's laboratories; for a review, see reference 21), and the pole, therefore, is at most only slowly subject to the actions of the autolysins. Lack of turnover of poles is a general finding which also applies to gram-positive cocci and the poles of a gram-negative organism like Escherichia coli (2, 31). A possible explanation for the lack of turnover of the poles of gram-positive cells (rods and cocci) has been put forth (18), based on the geometrical rearrangement consequent to septal splitting. The cleavage plane of the septal wall during bisection by autolysin action is perpendicular to the cytoplasmic membrane of the annular ring of the closing septum, that is, to the plane of addition of murein to the ingrowing septum. Thus, the inertness in poles here could be explained if the autolysins of the cell cannot attack either the glycan or the peptide chains “end on” (23).

A theoretical reason for this general inert nature of poles is that the poles form a scaffold to support the new side wall and nascent poles for rods and cocci, respectively. This is a key element in the surface stress theory (16, 21, 22).

ROLES OF AUTOLYSINS

The roles of functions of autolysins for the growth process are systematized elsewhere (p. 142–145 of reference 53). They are (i) the provision of new acceptor sites, (ii) enlargement of the peptidoglycan sacculus, (iii) remodeling functions, (iv) cell division and cell separation, and (v) peptidoglycan turnover. Above, these areas have been touched upon in a different way. In addition, it seems likely that peptidoglycan hydrolases play important roles in the liberation of the mature spore from the mother cell and the hydrolysis of cortex peptidoglycan during spore germination.

A presumed important role for autolysin is that of causing cell suicide, whence comes its name. Although it is hard to find a selective value to such a process, the long-term survival of the strain may be dependent on an active death process. If only one of a million or a billion cells in a colony propagates the strain, that would be enough (15, 34). Prokaryotic cell death might be the single-celled organism's analogue that corresponds to the phenomena of apoptosis and altruism considered for the cells of multicellular organisms under the heading of programmed cell death.

AUTOLYSIN CONTROL

It is almost self-evident that the cell's autolysins must be under careful control, especially in many tolerant strains. However, not all autolysins are lethal to the cells that created them, for example, the enzymes that finish the turnover and degradation of wall peptidoglycan. Not all walls can be attacked by particular autolysins; the fact that poles are typically inert was raised above. Attempts to produce completely autolysin-negative bacteria have failed; for example, Fein and Rogers (4) could easily find a mutant simultaneously deficient in the two major autolysins of B. subtilis but could not produce a totally autolysin-negative cell. Although the major autolysin in pneumococci that causes lysis of stationary cultures can be ablated, it may not be essential for vegetative growth.

Analogy of the effects of antibiotics to the stringent response has been made (42), but the analogy is far from perfect. In the original context (1a), stringent cells deprived of a needed amino acid had mechanisms to shut down RNA metabolism, but a cell treated with an antibiotic against the cell wall in principle has resources for the synthesis of protein, RNA, and DNA. Therefore, the claim of Ishiguro and Ramey (14) may not be valid, and control of peptidoglycan synthesis may be only a secondary effect of lowering of the cell's turgor pressure.

Currently, in studies in Tuomanen's laboratory with Streptococcus pneumoniae (40, 41, 42) and in studies in Bayles's laboratory with Staphylococcus aureus (1, 5, 6, 8), regulatory genes that affect autolysin function are being studied. Evidently, tolerant strains of these species regulate the autolysin function. While the evidence for regulatory gene function is clear in both cases, the role of the turgor pressure in the cells in general has not been studied. However, I briefly review the current studies of the regulation of the autolysins, even though the role of turgor pressure has not yet been covered.

Studies with S. pneumoniae

S. pneumoniae has been important in studies of molecular biology of a pathogen since the studies of Hotchkiss, Tomasz, and their colleagues. The findings of Moreillon et al. (35) may be considered an important change of paradigm. Studies from Tuomanen's laboratory with both penicillin (40) and vancomycin (41) have recently documented that the phenomenon of tolerance does depend on a two-component system that they have identified and studied. This VncR-VncS system is a two-component system of the sensor histidine kinase-phosphatase type. The signal sensed is a secreted peptide, Pep²⁷. This does prove that signal transduction is critical for the tolerance phenomenon (41). Yet, just what the two-component system controls is not clear, but see references 1 and 29. Does it control general metabolism, or are the permeability properties of the wall controlled by some global control?

Studies with S. aureus and virulence regulators.

Bayles (1) reviewed the work from his laboratory (5, 6, 8) on the regulators of autolysis and virulence. lytS-lytR is a novel two-component regulatory system. Downstream from it is the irgAB operon, which requires the two-component regulator system for action. It then inhibits extracellular murein hydrolase and increases the tolerance to penicillin. IrgA appears to be an antiholin. The viral holins (61) were formerly known only as murein lytic agents that oligomerize to form channels that allow bacteriophage-encoded murein hydrolases access to the peptidoglycan.

There are two additional points of interest concerning the staphylococci. The first point concerns lysis of these cells by autolysin. Consistent with ideas presented above, the part of the cell autolyzed is the site where the murein of the cross wall is actively being formed (54). The second point concerns the special bodies laid down within the cross wall. These are the “murosomes.” When these are activated by the cell, their hydrolases allow the walls to split evenly (7, 33).

TOLERANCE CAUSED BY AUTOLYSINS NOT ACTING TOGETHER WITH LIMITATION ON THE INCREASE IN TURGOR PRESSURE

The section on gram-positive rod growth was presented with no consideration of antibiotics active against the cell wall. This was done to point out the roles of autolysins in order to contrast those roles with the consequences of antibiotic action. In normal strains, if there is an increase in turgor pressure due to beta-lactam blockage of the enlargement process, extant autolysins may act to rupture regions of the walls of most cells, killing them. With tolerant strains, some special mechanism acts so that autolysins do not rupture the cell because of either qualitative or quantitative differences in the wall, the autolysins, or the inhibition of wall synthesis. However, the apparent logical inconsistency or incompleteness of this idea remains, in that with continued cytoplasmic growth, the turgor pressure should increase and eventually the wall should rupture or tear either with or without autolysin action.

Tolerant cells, evidently, would need to stop cell growth in some indirect way. Such action must be indirect because penicillin-type compounds bind to penicillin-binding proteins (PBPs) but do not interfere directly with DNA, RNA, protein, and membrane biosynthesis. Consequently, the Shockman-Rogers-Tomasz hypothesis of lack of autolysin action cannot be saved simply by imagining that such cells somehow regulate or inhibit (or fail to form) autolysins and thus prevent their attack on the stress-bearing wall. Four categories of possibilities for the preservation of the cells under such circumstances are considered below.

(i) Leaky walls.

The Shockman-Rogers-Tomasz hypothesis of a lack of autolysin action in tolerant cells could be saved simply by imagining that in such tolerant cells, as the turgor pressure increases, the membranes and/or wall become leaky enough to keep the turgor pressure from increasing excessively. This upper limit of turgor pressure would occur because salts and small molecules, etc., would leak out of the cell and prevent the tension in the peptidoglycan layer from becoming too great. In this tolerant state, small molecules must leak out as fast as they are pumped into the cell. This model requires that the tolerant cell have two changes: (i) failure of autolysin action upon antibiotic challenge and (ii) structural changes in the wall to make the wall specifically more permeable.

(ii) High turgor pressure indirectly stops cellular growth.

Another hypothesis for tolerance starts in the same way as the first hypothesis, i.e., that upon antibiotic challenge of the tolerant cells the autolysins are specifically inhibited from acting. The wall, however, remains strong and contains the turgor pressure of the cell, and it, together with the cytoplasmic membrane, does not passively leak salts and the cell's intermediary metabolites. The pressure, however, does not rise indefinitely because when it reaches a certain value water cannot enter, despite the osmotic pressure difference, and water may even passively leak out of the cell. Also, facultative and active transport mechanisms may be forced by the pressure to operate backwards to cause efflux instead of influx. So far this hypothesis is not much different from the leaky cell hypothesis, but its major aspect is the assumption that the changes due to pressure on the structure of cell water and cellular dehydration lead to a direct inhibition of synthesis of all classes of macromolecules. Such changes would be reversed when the lactam or other antibiotic with activity against the wall is removed. A possibility is that potassium loss, as well as water loss, would occur and could lead to an inhibition of the synthesis of proteins on ribosomes. In this model, the cell (i) must have a strong wall, (ii) prevent in some manner autolysin function, and (iii) block macromolecular synthesis at the biophysical level.

(iii) Active mechanisms controlling macromolecular synthesis.

One could imagine the existence in tolerant cells of global regulatory mechanisms which are sensitive to turgor pressure and which act to turn off cellular synthesis of macromolecules. Such mechanisms might be two-component systems, typical of common regulatory systems in bacteria. The sensor component could be fixed in the murein or cell membranes and could be activated by the stretching of the surrounding wall elements. In a variant model the sensor might respond to specific chemical agents like lactams. The idea that such a system was the well-known “stringent response” has some merit, but the response must be quite different from that expressed during a nutritional deficiency. There is no evidence that the original stringent response acts in response to turgor pressure.

(iv) Rapidly reverting gene mutations.

The experiments described below suggest that a recently cloned population of bacteria contains mutant individuals that can successfully respond to environmental challenges. After the challenge is over, the original dominant population results from rapid mutation and selection.

EXPERIMENTAL SYSTEMS

Studies with S. mutans.

Work from Shockman's laboratory (39, 52) showed that Streptococcus mutans is normally tolerant to lactams that block its growth. Experimental data that show that the tolerance resulted from the high level of strength and the integrity of the murein wall were presented. It was found that after treatment with a suitable lactam, formation of DNA, RNA, and protein gradually slowed and stopped. Once stopped, however, the synthesis reinitiated when lysozyme was subsequently added. The lysozyme caused the cell wall to dissolve, but of course, of itself it should have little to do with the stimulation of synthesis of diverse macromolecules. Therefore, it was surprising that the formation of the several kinds of macromolecules recommenced. I can only take this as evidence that the strong wall of S. mutans led to the high turgor pressure that had blocked the formation of macromolecules because the physical and hydrostatic forces interfered with the cell's synthetic activities. This physical inhibition was relieved when the cell's integrity was abridged. (This was the second hypothesis suggested above.)

Studies with A. aquaticus

While there are no good routine ways to measure the turgor pressure of bacteria, the best available one is that described by Walsby (58). His technique was further improved so that measurements could be carried out quickly and more accurately (30, 44, 45). However, this approach has the inherent disadvantage that it can work only with cells that possess gas vacuoles (sometimes called vesicles). There are known nonphotosynthetic organisms that have gas vesicles, and Ancylobacter aquaticus was chosen (32) for studies of antibiotic action.

A special apparatus was constructed to obtain measurements (30, 44, 45). One part served to allow a sample of cells placed in the pressure system to be exposed to hydrostatic pressure in a programmed way. Another part of the apparatus was a detection system based on a laser beam. The light scattered at 90° from the laser beam was measured. When the pressure on a gas-filled vesicle became high enough, it collapsed the gas vesicle, which then reduced the light scattering.

By this technique it could be shown that the turgor pressure of the growing cell was about 200 pN or 2 atm. This allowed, for the first time, measurement of the effect of a lactam on turgor pressure. What was anticipated was that on application of a penicillin derivative (such as ampicillin) the turgor pressure would quickly rise in seconds or minutes as cytoplasmic synthesis continued under conditions such that the wall volume could not increase. The expectation was that it would fall to zero when the wall integrity was breached. However, this sequence was not observed for all cells. Although this prediction was found to be true for three-fourths of the cells followed, with the turgor pressure dropping to zero in different cells over a 30-min period, for the remainder it was not; instead, for one-quarter of the cells, the turgor pressure did not fall but, rather, rose about 100 kPa and remained there for many hours. These cells did not lyse. This behavior was typical for cells treated with most of antibiotics including ampicillin, penicillin G, cephaloridine, and cephalexin. On the other hand, application of amdinocillin (which in E. coli inhibits PBP 2 and prevents rod-shaped growth) did not distinguish the two distinct components seen with the other lactams and did not lead to a collapse pressure that was as high (and, therefore, a low turgor pressure). This behavior would be expected if the cell became somewhat leaky.

The likely interpretation is that the cells have an active mechanism that is called into play in a fraction, but not all, of the organisms in the culture (20, 38). This causes the cell's global uptake and macromolecular synthesis mechanisms to arrest. This protects at least some cells from damage. However, the bulk of the cells do rupture, although not all cells rupture at the same time.

CONCLUSIONS

All or some of the hypotheses presented could apply to different tolerant bacteria. It is important to find out the actual basis of tolerance for pathogens because this could be the basis for a new chemotherapeutic approach.

This commentary focuses on two relatively new concepts. The first is the probable importance of cellular turgor pressure. This has been emphasized in a 1995 book that has appeared in a new edition (21). The second is the realization that bacteria possess ways to favor mutation at certain loci. In the last few years, the case has been building that even recently cloned cultures of bacteria are not genetically and physiologically uniform (20, 36–38). It appears that the many genetic regions of many bacteria are capable of rapid mutation. These are not ordinarily detected because the back mutation is rapid, too. A well-documented mechanism for rapid mutation is for particular genes to have a series of tandem repeats where each repeat length is different from a multiple of three. This arrangement allows a miscopying or misrepairing event to cause a frame shift and to alter protein production. The relevance for lactam antibiotics is that tolerant strains may be capable of this facile type of frameshift mutation to permit some cells to survive. These changes, in the absence of the antibiotic, can rapidly recover by the same type of frameshift mutation under a different selection pressure.