Control of Cell Division in Bacteria

Transcription

1 BAnrEmoIwOOIcAL REVIEWS, June 1974, p Copyright American Society for Microbiology Vol. 38, No. 2 Printed in U.S.A. Control of Cell Division in Bacteria MARTIN SLATER AND MOSELIO SCHAECHTER Department of Molecular Biology and Microbiology, Tufts University School of Medicine, Boston, Massachusetts INTRODUCTION A Digression into Yeast PHYSIOLOGICAL ASPECTS OF CELL DIVISION Division Process Site of Envelope Synthesis Sensitivity of Cell Division to Interfering Conditions GENETIC STUDIES OF CELL DIVISION General Comments DNA Metabolism Protein Synthesis "Division Potential" Placement of the Site of Division Cell Wall Small-Molecule Metabolism CONCLUSIONS INTRODUCTION Because in one sense bacteria are structurally and genetically simple and amenable to study, it might be expected that their division depends on relatively few events which are interrelated in simple patterns. Our principal conclusion is that this expectation is essentially wrong. We will concern ourselves with the regulation of division, rather than with the process itself. A recent review by Higgins and Shockman (52) deals extensively with the present knowledge on the process of division in bacteria. Here we will discuss briefly the operational definitions of bacterial division, the sequence of events leading to its initiation, and how these events are interrelated. Our emphasis will be on genetic studies. In our minds, there are major operational problems in working with bacterial cell division. These are: (i) determining if inhibitors or mutations which affect division are specific; (ii) deciding whether closely linked mutations are in different sites of the same cistron or in different cistrons; (iii) recognizing the existence of bypasses and possible artifacts produced near restrictive conditions; and, (iv) finding biochemical means of studying the synthesis of the cross wall and membrane separately from the synthesis of the peripheral envelope. Recent work indicates that these problems appear to be less troublesome in the yeast Saccharomyces cerevisiae (47). We will describe some of this work and use it as a guideline for determining the kind of questions that may be asked of cell division in bacteria. A Digression into Yeast Two facts about yeast indicate some of the advantages of this system. First, the primary wall septum seems to consist of a single specific substance, chitin, which is not found in the rest of the cell wall (18). Thus, the study of the biochemistry of chitin synthesis may constitute a direct study of the regulation of cell division in yeast. Second, the regulation of division seems to be more specific and less sensitive in yeast than in bacteria. In bacteria a wide range of inhibitors, many of which stop growth at high concentrations, preferentially stop division at low concentrations. In yeast, the spectrum of preferential inhibitors of division is both more narrow and more specific (89). A remarkably rapid and straightforward study of the genetic control of the yeast cell cycle is being conducted by Hartwell's group (47). From an array of temperature-sensitive mutants, "cell division cycle" (cdc) mutants were selected for study initially on the basis of two defining properties: (i) when shifted from 23 to 36 C, all of the cells of an asynchronous population accumulate with a uniform morphology characteristic of a block in one step of the cell cycle (this is called the terminal phenotype). (ii) After a discrete point during the cell cycle, the remainder of the cycle can be carried out to completion even at 36 C (this is called the 199

2 200 SLATER AND SCHAECHTER BACTERIOL. REV. execution point). Cells which have passed the execution point at the time of the shift complete the ongoing cycle, divide once, and accumulate terminal phenotypes in the next cell cycle. Because mutations affecting the synthesis rather than the activity of proteins would not necessarily confer a relevant execution time on the mutant, in some cases cdc mutants are selected on the basis of uniform terminal phenotypes. However, the execution point remains an important criterion in selecting and grouping specific mutants. Blocks at each of the major observable events in the "deoxyribonucleic acid (DNA)-division" cycle in yeast are found among these mutants. It has also been possible to determine the time at which cells "decide" to start a mitotic cycle rather than enter meiosis or conjugation. This is essentially a differentiation step subject to regulation (in this case by mating factors) which changes one type of cell cycle to another type (in this case from haploid cycles to mating and diploid cycles). The mutations are recessive and, by complementation tests and tetrad analysis, were found to fall into 35 nuclear cistrons. In all cases, different mutations in a given cdc cistron resulted in the same phenotype (same terminal phenotypes and execution points). Apparent, exceptional cases were shown to be double mutants. The terminal phenotypes and execution points are, then, cistron rather than allele specific. These results purport to show that a cdc gene specifies a function which is required at a particular time to mediate one and only one step in the cell cycle. These studies have permitted determination of the sequence and relationship between all the observable events in the cell cycle of yeast. PHYSIOLOGICAL ASPECTS OF CELL DIVISION Division Process Gram-positive bacteria divide by forming a cross-septum consisting of both membrane and wall (22) (Fig. 1). In some gram-negative bacteria such a structure is usually not apparent (as an example, see 46a). Dividing Escherichia coli cells, for example, appear constricted at the site of division, showing a gradually increasing inward curvature of both membrane and wall (Fig. 2). (Morphologically speaking, it is erroneous to refer to this configuration as "septum." Nonetheless, we are forced to make a general use of this term since there is no other convenient way to refer to the wall and membrane directly involved in cell division). The presence or absence of a morphologically distinct septum may be a real difference between species and may reflect possible differences in the timing of various steps in the division process (123) (see Fig. 2 and 3). Thus, a septum would exist only fleetingly if the inward growth of envelope is followed very rapidly by cell separation. On the other hand, it is possible that the reason why septa are usually not seen in gram negatives is that they are destroyed during cytological fixation. This seems to be the more likely case, because septa can be seen when E. coli are fixed with 5% Acrolein and 0.5% glutaraldehyde (I. D. J. Burdett and R. G. E. Murray, personal communication) (Fig. 4). They are also seen when other fixation procedures are carried out at certain temperatures or when cells are grown at 45 C (138). This question needs further attention because precise knowledge on the mode of division of E. coli is required to define the phenotypes of mutant cells. These uncertainties make it difficult to estimate the time required for various steps in the division process. The only step whose timing has been determined with precision is the interval from the initiation of production of a septum to its expression in a division in Bacillus subtilis. Paulton showed that the time for this step is constant over a wide range of growth rates and is 138 min long at 30 C (109). At rapid growth rates, this step is initiated repeatedly before the original cells separate, thus producing multiseptate cells. This is analogous to the constancy of the time required for a round of DNA replication at various growth rates and with the multifork pattern of chromosome replication which exists at rapid growth rates (e.g., 48). It must be emphasized that 138 min is the time from the inception of a septum to its participation in division and not the time required to make the septum. In the case of gram negatives, comparable studies have not been carried out, but there are suggestions that the time required for division is also not a fixed fraction of the generation time. The proportion of cells with a visible constriction was measured in cultures of E. coli growing at different rates (84). This proportion was found to be greater in cultures growing at faster growth rates. In cultures growing with a doubling time of 30 min, it reached a value of 25% of the total cell population. Because cells with visible constrictions have obviously initiated "septum" synthesis sometimes before, this measurement is an underestimate of the fraction of the cell cycle required for "septum" synthesis. The magnitude of this underestimate is unknown, but can

3 VOL. 38, 1974 BACTERIAL CELL DIVISION CONTROL 201 FIG. 1. The septum in gram-positive rods. Electron micrographs of thin sections of Bacillus cereus showing septa (A) and their thickening by exposure to chloramphenicol (B); reproduced from reference 23 by permission of K. L. Chung and the National Research Council of Canada. be approximated from another study. When cultures of E. coli or Salmonella typhimurium, also doubling every 30 min, were transferred to starvation media, they continued to divide until the cell number nearly doubled (128). If, by analogy with DNA replication, starvation prevents the initiation, but not the actual formation of the "septum," then this extent of residual division indicates that "septum" synthesis was initiated in the previous cell cycle. It should be noted, however, that the assumption concerning the effect of starvation on initiation is hitherto unproven. The most important physiological step in cell division may be the functional partitioning of the cell, that is, the process whereby the two daughter cytoplasms become separate compartments. The timing of this step has been studied by using physiological criteria for the partitioning of daughter cells. The principal method has been to determine if the kinetics of killing of cells vary during the cell cycle by using phage infection or sonication of synchronized cultures (25). Clark (25) found that within a defined time in the cell cycle, before a morphological partition is seen, the number of hits required to kill a cell changed from one to two. This would correspond to a time of "physiological" rather than physical separation between sister cells. Extensive work on this subject was carried out by Onken and Messer (108), who studied several activities expected to be affected by separation of sister cytoplasms. They found the following sequence of events in synchronously dividing cells whose doubling time was 45 min: DNA replication ends at 25 min, the synthesis of murein increases at 30 min, sensitivity to colicin E2 increases between 30 and 35 min, inactivation by phage T4 changes from one to two hits at 35 min, and the ability of two mutant phages to recombine and complement each other continues until shortly before division. They favored the interpretation that some of these results are due to changes in properties of the cell envelope and not to compartmentalization of the cytoplasm. They concluded that cells become both physically and functionally separated only shortly before they actually divide. Little is known about the mechanisms involved in the physical separation of divided

4 202 SLATER AND SCHAECHTER BACTERIOL. REV. FIG. 2. Constriction in gram-negative rods: cell division in E. coli D21. All layers of the envelope participate in the concentric invagination process. x62,400; reproduced from reference 104 by permission of S. Normark and the publishers of Acti Pathologica Microbiologica Scandinavica. Site of Envelope Synthesis It has been proposed repeatedly that cell division is under the control of the activities of a "membrane growth site" which is responsible for both the extension and the division of the cell. According to this model, the site feeds wall in the peripheral direction for cell extension and in the centripetal direction for cell division. We will consider the evidence for the existence of a "membrane growth site," as well as the available data regarding its location. In gram-positive cocci the existence of such a site is firmly established, thanks to a detailed analysis of the morphological changes associated with the function of new cell wall. This work has been recently reviewed and will not be further treated here (52). In rod-shaped bacteria, the existence of a membrane growth site rests on two types of evidence. (i) New cell wall appears to be deposited in cells of all ages at the center of the rod, as bacteria. On the level of single cells, it is thought that the variability in the timing of separation is mainly responsible for the differences in generation times of individual cells observed in many species (e.g., 82, 112, 129). In B. subtilis and pneumococci there is evidence that hydrolases function in the separation step (34, 38a, 142a). The extent of cell separation is correlated with the hydrolase activity of various mutants and of cultures growing under different conditions (34, 35). There are mutants of E. coli, called enva, which form chains ( ). Their murein cross wall is completed without the outer lipopolysaccharide-containing layer, which later is placed between the layers of the murein septum. A mutant of S. typhimurium, 4a, forms filaments at 42 C, but at 25 C in the presence of high concentrations of yeast extract FIG. 3. The septum in mutants of gram-negative it forms chains of completely partitioned cells. rods: cell division in the E. coli enva mutant D22. A In these chains, the lipopolysaccharide layer is septum separates individual cell units. Centrally included in the cross walls between cells (7). In located in the septum is a thin structure of moderately electron-dense material. Close to one invagination this mutant, in otheraoniss point this structure is split into two components. the variability in the time of separation of x60,000; reproduced from reference 104 by permisindividual cells may be due to the sticky properties of a slime layer. sion of S. Normark and the publishers of Acta Patho- logica Microbiologica Scandinavica.

5 VOL. 38, 1974 BACTERIAL CELL DIVISION CONTROL 203 A c r. FIG. 4. The septum in wild-type E. coli seen as the result of special fixation. Cells grown in glucose minimal medium were fixed with 5% Acrolein and 0.25% glutaraldehyde in 0.05 M sodium cacodylate buffer, ph 7.5. A, E. coli B, x60,000; B, E. coli Bir, x 121,000. Micrographs were provided by courtesy of I. D. T. Burdett and R. G. E. Murray (submitted for publication to J. Bacteriol., 1974). shown radioautographically by the location of peripheral bulges produced by weakening of the diaminopimelic acid newly incorporated into wall. It has been proposed (30) that these bulges murein (126). (ii) E. coli treated with low correspond to the areas of murein synthesis and concentrations of penicillin shows characteristic to the sites where division would have taken

6 204 SLATER AND SCHAECHTER BACTERIOL. REV. place but for the inhibition by penicillin. The location of these bulges was determined in cells whose size was altered by growth at different rates (30). The bulges were always formed at a constant distance from one of the poles of the cells. Thus, in small cells growing at a slow growth rate, this distance is smaller than onehalf of the cell length. In such cells the site is therefore asymmetrical and closer to one pole. The contradiction between this result and that obtained by radioautography has not yet been resolved, but in either case it seems that the murein component of the cell wall is synthesized at a unique site. After being synthesized, murein is distributed rapidly over the cell surface (88). This points out at least one of the difficulties in the interpretation of studies of segregation of parental wall or membrane material. Thus, the finding that wall or the membrane subunits are distributed randomly among progeny cells does not permit differentiating between synthesis at many sites in each cell or synthesis at one site followed by rapid randomization of the newly made material. It is not known with certainty how membrane growth sites originate. One study with mutants of B. subtilis indicates that new sites may originate de novo and are not formed by the duplication and segregation of preexisting sites (98). The converse mechanism has been proposed for E. coli (30). Sensitivity of Cell Division to Interfering Conditions An unusual feature of cell division in bacteria is that it can be selectively inhibited by a great variety of chemical or physical agents. Many substances which stop growth at high concentrations will, at lower concentrations, not inhibit growth, but will stop cells from dividing (61, 89, 120). The result is that rod-shaped bacteria continue to grow and form nonseptate filaments. The list of agents capable of this is remarkably varied. It includes antibiotics, dyes, detergents, disinfectants, antimetabolites (e.g., 114), high and low temperature (133), ultraviolet light (75), the presence of ind--x prophage (142), deprivation of certain essential nutrients, and excesses of other nutrients (7, 147). Although some of these substances or physical agents selectively inhibit the synthesis of constituents of cell envelopes (e.g., penicillin), others are not known to do so. Few cases have been studied in detail, and it is not possible to make definite statements concerning the reasons for this unusual sensitivity of bacterial division. However, some tentative generalizations concerning conditions which favor either filament formation or division can be made. Generally, filament formation is: (i) caused by a low dose of an agent which inhibits growth; (ii) enhanced by fast growth in rich media (e.g., 2, 40, 59, 144, 149, 152); (iii) reversed by suddenly slowing growth (chloramphenicol treatment; 17, 43, 152), nutritional shift down (144), or "liquid holding" (73, 74). In fact, it seems possible that any chemical at some concentration, whether attainable in the laboratory or not, could cause filament formation. (When this was told to Rollin Hotchkiss, he muttered: "How depressing!") Agents that restore the ability of filaments to divide make up another varied list. Thus, filaments of various species which result from the expression of various mutations or nutritional conditions divide after the addition of sodium chloride (43, 66, 93), pantoyl lactone (3, 44), spermine (46), dimethyl sulfoxide (63), lysolecithin (63), sodium oleate (63), or short-chain alcohols (63). Cell division also requires magnesium in Aerobacter aerogenes (78) and Clostridium (146) and calcium in Lactobacillus bifidus (83). A division-promoting, lipase-sensitive particulate fraction has been isolated from E. coli (3, 4). Small-molecular-weight substances isolated from the blue-green alga Agmenellum quadriplicatum stimulate the division of filamentous mutants (63). Although some of these agents can be expected to act on the cell membrane, others influence a large variety of cell functions. Some recent findings emphasize the need for caution in interpreting the reversion of division mutants by nonspecific agents such as salt. Thus, 80% of randomly collcted temperature-sensitive mutants of E. coli can be reversed at the restrictive temperature by raising the osmotic pressure of the medium (12, 119a, 125). Salt dependence of a division mutant of B. subtilis was traced to its effect on enzymes of glutamine metabolism. In fact, the osmotic protection of temperature-sensitive mutants may be explained by a phase change of membrane constituents at 40 C (72, 95). At this temperature wild-type cells may be in a precarious position which can be upset by any of a number of mutations and affect the most sensitive event in the cell cycle-cell division. GENETIC STUDIES OF CELL DIVISION General Comments Before discussing genetic studies on cell division, we must introduce several notes of cau-

7 VOL. 38, 1974 BACTERIAL CELL DIVISION CONTROL 205 tion. In order to make such studies meaningful, three types of operational criteria should be met. (i) It is necessary to determine whether complex phenotypes are due to pleiotropic effects of a single mutation or to multiple mutations. (ii) It must be determined whether two mutants with the same apparent phenotype have a lesion in the same cistron. This can be definitely established only by complementation tests. Unless this test is carried out, it is not possible to decide whether in closely linked mutants allele- or cistron-specific differences account for different phenotypes. Alleles of genes regulating division or DNA synthesis may elicit different phenotypes. Mutations affecting, say, allosteric properties, might result in phenotypes different from those resulting from mutations in the same gene affecting catalytic sites. Complementation tests have not been reported extensively with bacterial division mutants, and the interpretation of work done with them is limited. (iii) It is necessary to determine the specificity of the mutation for cell division. As pointed out by Mendelson and Cole(96 (97), all conditional lethal mutants are, in some manner, division mutants. In order to study cell division, one must choose from a wide spectrum of conditional lethals those believed to be involved specifically in cell division. Generally, mutants that under some conditions continue to grow without dividing are considered to be division mutants. The fact that division is preferentially inhibited by a great many disparate agents indicates the need for caution. Many mutations controlling functions related to growth may appear to be "division mutants" because they are just "leaky" enough to allow exponential growth without division. Many such filament formers may have a primary genetic defect that is reflected in many metabolic processes other than those directly related to the division process. It is not obvious that much will be learned from their study before the complexities of their causal relationships are known. Mutations relevant to studies on cell division may affect (i) functions in the division process itself (i.e., septum synthesis), (ii) functions preceding the division process, but specifically related to the preparation or triggering of the initiation of division, or (iii) functions related to the separation of sister cells. Besides the criteria discussed above, there are other difficulties in classifying division mutants into the first two of these groups. At first glance, conditional division mutants may be grouped (in a fashion analogous to that used with DNA mutants) into those which cannot initiate division (initiation type) and those defective in the process of septum formation (elongation type). Mutants of the first type should continue to divide if they initiated the process before being placed at the restrictive condition, whereas mutants of the second type should stop division immediately. Elongation-type mutants should be frozen at whatever stage in division they were at the time of the conditional shift. Mutants that stop division immediately have, in fact, been found. Unfortunately, they have not been shown to possess incomplete cross walls (6, 24, 101, 115, 117). One must conclude that either the time for septum synthesis is very short or that incipient septa are resorbed (possibly by hydrolases concentrated near their site of synthesis) or destroyed during cytological fixation (139). To predict the quantitative behavior of "initiation-type" mutants one must know the time required for division. The extent of residual division at the nonpermissive condition depends on the time in the cell cycle when formation of the septum was initiated. Because this is not known with precision, it is not possible to define initiation-type mutants on the basis that they carry out some residual division (8, 15, 97). In the absence of any other useful criteria, we may say with assurance that no satisfactory assay for initiation of bacterial division exists as yet. The difficulties of this field are compounded by the fact that there is no known biochemical difference between the envelope components involved in cell division and those of the rest of the cell. This makes it very difficult to describe various division steps at the biochemical level and to assign mutants to these steps. We can expect progress in this area because differences in structure have been reported between the wall at the ends and at the sides of B. subtilis (37). However, these differences may arise subsequent to completion of the septum (35a). In 1968, Hirota et al. (58) summarized work done with a wide spectrum of temperature-sensitive mutants of E. coli. Many of the mutants showed alterations in the normal relationships between events in cell division, thus permitting dissection of this process. At that time, the various components of the cell division cycle could not be arranged in an ordered sequence. In the years that have followed, little progress has been made in formulating a causal order for these components. In 1973 the temperature-sensitive filament formers were genetically classified in seven groups, ftsa through G, and physiologically separated according to reversibility

8 206 SLATER AND SCHAECHTER BACTERIOL. REV. and the effect of chloramphenicol on reversibility (119a). Again, no ordering was attempted. Accordingly, in this article we will not attempt to do so, but will present separate data on the relationship between various physiological processes and cell division. We thus manifest the opinion that the formulation of causal connections in bacterial cell division seems premature. DNA Metabolism In this section we will present evidence that: (i) the timing and the relationships within the DNA-division cycle are complex and subject to change; (ii) some alternative mechanism is required for cells to divide in the absence of DNA synthesis; (iii) the control of cell division is affected by several facets of DNA metabolism (synthesis, repair, and recombination) rather than just DNA synthesis; the DNA repair capacities of the cell act as negative controls on division; (iv) DNA synthesis is related to properties of the cell envelope; the factors linking DNA synthesis and the envelope are possibly also involved in linking DNA synthesis to cell division; and (v) different mutations in the same "gene" affecting DNA synthesis give rise to different phenotypes. Point (i): the timing and the relationships within the DNA-division cycle are complex and subject to change. The normal temporal order and how it can be changed. On a gross level, the sequence of events in the DNA division cycle has been worked out for E. coli. Over a wide range of growth rates at 37 C, a round of chromosome replication begins 60 min before a division and ends 20 min before that division (48). Division is correlated with the attainment of a cellular "unit of mass" (28). The orderly sequence of morphological changes, division, and segregation of nuclei in living cells has been studied by several authors, notably Adler and Hardigree (3). There is considerable variation in the timing of those processes in individual cells (129). In rapidly growing E. coli, division of the nuclear bodies takes place, on the average, about halfway through the cell cycle, before any signs of wall constriction (5, 129). The timing of these events can be altered. For instance, lowering the temperature increases the time required for a round of DNA replication (C period), but does leave the 20-min lag between termination of the round of replication and completion of division (D period) relatively unaffected (91). The D period in E. coli p245 apparently is shorter than in E. coli B/r. The length of the C period in E. coli 15T - can be varied by changing the thymine concentration (151). Heat shocks synchronize division, but not DNA synthesis, thus changing the D period (135). Division takes place 5 to 10 min after termination of rounds of replication (short D) when E. coli is subjected to alternate blocks of ribonucleic acid (RNA) and DNA synthesis, followed by reversal of the DNA block. The C period is not changed (71). Increasing the generation time beyond 60 min alters both C and D, but, according to Pierucci (110), does not change the relationship C = 2D. On the other hand, according to Kubitchek and Freedman, the age at initiation of a round of replication varies with the growth rate, but C and D do not (85). This contradiction has not been fully resolved, but may be attributable to differences between strains (C.- Helmstetter, personal communication). In conclusion, the time between initiation of rounds of replication can be changed without affecting C, D, or the mass per new growing point, the C period may be changed without affecting D, the D period may be altered without affecting C, and both may be altered. Normal relationships and how these can be changed. In E. coli, cell division does not take place in the absence of DNA synthesis. Adding inhibitors of DNA synthesis (25) or transferring most conditional mutants defective in DNA synthesis (dnai) to restrictive conditions (58) leads to filament formation. Although several models for this connection will be discussed below, it is not known if the signal or signals act positively or negatively, or if the biochemical steps involved are closely or distantly related. Under special circumstances, this normal connection is changed and cells can divide without DNA synthesis. To date this bypass has been observed only with specific mutations, and no nutritional conditions or chemicals or physical agents have been found which allow wild-type cells to divide without synthesizing DNA. In B. subtilis the situation is different, because wildtype cells can divide in the absence of DNA replication (31). In E. coli, temperature-sensitive mutations in DNA synthesis have been mapped in seven chromosomal positions (designated A through G). The dna- A and C mutants are defective in initiation of DNA replication, whereas the others are involved in the elongation of the molecule (e.g., 55). E. coli T46 (a dna- A mutation) and S. typhimurium 11G (probably dna- C) are examples of DNA initiation mutants that have been studied intensively in connection with cell division. The presence of a mutation, diva, allows E. coli T46 (54) to divide at the restric-

9 VOL. 38, 1974 BACTERIAL CELL DIVISION CONTROL 207 tive condition, whereas S. typhimurium l1g divides without any known "div" mutations (138). E. coli T43 (56) and ts27 (42, 64, 66) are dna-b mutants, defective in chain elongation. DivB allows E. coli T43 to divide, whereas E. coli T27 divides in the absence of known "div" mutations. These examples show that the requirement of DNA synthesis for division of gram-negative rods is not absolute and that nearly all possible combinations have been found: usually division stops if DNA synthesis is stopped; sometimes division continues in mutants defective in initiation or continuation of DNA synthesis; some dna - mutants defective in initiation or continuation of DNA synthesis require additional "div" mutations, whereas in others such a requirement is not evident. Point (ii): some alternative mechanism seems necessary for division to occur in the absence of DNA synthesis. The bypass mutants just described do not divide if DNA synthesis is blocked by means other than the effect of the mutation. In all cases tested. (E. coli T46-divA [54], S. typhimurium 11G [131, ], and E. coli ts27 [64, 66]), when inhibitors of DNA synthesis were added at the permissive temperature, the cells did not divide. Similarly, when DNA synthesis in dnamutants is blocked by thymine starvation, cells do not divide (66). These results suggest that in these mutants the relationship between DNA synthesis and cell division has been altered and operates by a different mechanism than in normal cells. We use the term "mechanism" in the broadest sense, because we do not know if it reflects new biochemical steps or new insensitivity to the consequences of inhibiting DNA synthesis. Further evidence for an alternate mechanism is obtained from studies on conditional DNA initiation mutants. Cell division in these mutants has a lag of about 1 h when cells are placed under the restrictive condition. This lag is related only to the time spent at the restrictive condition and is not related to the termination of rounds of replication or events triggered thereby (56, 132). Thus, division in the absence of DNA synthesis requires the specific expression of certain mutations. How this is done is as yet unknown. Point (iii): DNA metabolism, not just synthesis, affects the control of cell division. There is evidence that the DNA repair mechanisms are involved in the regulation of cell division. This is based on studies of a certain class of mutants such as E. coli tif, which do not divide at 42 C (20, 21, 80). At the restrictive temperature these cells do not show defects in DNA or DNA synthesis, but have increased repair capacities. Suppressors of tif- are defective in repair abilities. The tif gene may be a regulator for the repair systems, because at the restrictive temperature tif causes an increase in these systems, even when DNA synthesis is not inhibited (20, 21). This suggests that when DNA synthesis is inhibited in the wild type, lesions are produced in the DNA which "induce" an increased level of repair. This, in turn, stops cell division. A similar case was uncovered when studying revertants of an E. coli DNA repair mutant, lex -. A suppressor of this mutation, tsl, leads to inhibition of cell division (99). This was also interpreted to mean that the tsl mutation causes a gratuitous increase in repair enzyme synthesis which inhibits cell division. These may be examples of two separate mechanisms since the tif- and tsl- mutations map at different sites, and the tsl- mutant does not exhibit the same responses to added purines as tif- (80, 81, 124). In this connection we may mention a classical, but poorly understood, observation that after conjugation the recipient cells show a lag in division. Perhaps the DNA fragment introduced conjugally induces repair systems which inhibit division. RecA - mutants of E. coli continue to divide when DNA synthesis is inhibited chemically or by the expression of conditional dna - mutations. Here the normal relationship between DNA synthesis and division appears to depend on the reca gene product, which mediates a recombination function (65). RecA - also suppresses tif (99) and lon- (41) mutants. The lon - mutant, which will be discussed in detail below, is not altered either in DNA synthesis or in the repair functions tested. This mutant is defective in the recovery of division from transient unbalanced growth during which the DNA-mass ratio is low. (It also appears to be defective in the response to other regulatory systems such as the gal operon control and X repressor). Because tif- is involved in the coupling of DNA synthesis to division, the suppression of tif- mutants by reca- also implicates the reca gene product in the link between synthesis and division. Point (iv): the role of DNA synthesis in division and properties of the cell envelope. To begin with, several envelope properties have been found to be altered by inhibiting DNA synthesis. Hirota et al. (57) and Inouye and Pardee (68) have reported that when DNA synthesis is inhibited, certain membrane pep-

10 208 28SLATER AND SCHAECHTER tides are missing when analyzed by acrylamide gel electrophoresis. Conditional DNA mutants as well as wild-type cells exposed to chemical inhibitors of DNA synthesis have increased sensitivity to sodium desoxycholate at high temperatures, suggesting increased permeability of the outer membrane (57, 137). A variety of dna - conditional mutants also exhibit generalized changes in permeability (49, 57). The shape determination of cells is also affected by DNA synthesis, because spherical cells of mutants of E. coli (103) and B. subtilis (118) elongate when DNA synthesis is inhibited. This is also true for wild-type Streptococcus faecium (52). These findings are consistent with the proposal that during inhibition of DNA synthesis the synthesis of peripheral wall is favored over centripetal growth of the wall or thickening of the wall (52). Another possible connection between DNA metabolism and properties of the envelopes can be derived from studies on the partial inhibition of DNA synthesis. In E. coli decreasing the rate of DNA chain elongation by lowering the thymine concentration results in change in volume due to changes in cell width rather than in cell length (151). Because the envelope is the agent of division it would not be surprising to find that the relationship of DNA synthesis to the envelope is mediated by the same factors mediating the link of DNA synthesis to division. There are, in fact, indications that this is the case. The lonmutant in E. coli cannot recover the ability to divide after transient, unbalanced growth which lowers the ratio of DNA to mass. It is not defective in DNA synthesis itself, but is believed to be altered in the production of a factor TABLE 1. linking DNA synthesis to division (69, 75, 87, 145, 152). It is not yet known whether this alteration is due to a decrease in a division activator, or an increase in an inhibitor, or a change in a receptor site for an activator or inhibitor specified by other genes. There are some indications that favor the notion that the mutant is defective in a receptor site and that this receptor site is located on the envelope (69). Lon mutants are unusually sensitive to penicillin and other wall synthesis inhibitors (87), they have increased permeability (38), and they exhibit other envelope-related abnormalities (see Table 1). As yet, no defects in structure or synthesis of the wall were found in exhaustive studies. The Ion mutation is suppressed by the lex (33) (called exra in E. coli B) and sul (32) mutations, which are themselves involved in the control of envelope properties (69). Point (v): different mutations in the same "gene" may result in different phenotypes. In most E. coli dna- A mutants, a shift to the restrictive condition allows rounds of replication to terminate and cells to divide until each completed genome segregates into a daughter cell. However, in one mutant in this gene, CRT83, division stops immediately while rounds of replication continue until termination (79). If the dnaa gene product affected only initiation of rounds of replication, it would be difficult to explain why its inactivation stops division immediately. It is possible, then, that the gene product also changes the activity of other proteins involved in control of division. Increasing the temperature of CRT83 would then inactivate its ability to initiate a round of replication and its ability to affect division- Envelope-related characteristics of Ion- mutant' BACTERIOL. REV. Wall suscep- Accumulation Division Known Osmotic tibility to of uridine recovery in gross Mutant Penicillin pressure- sodium do- diphosphate Division presence of chemical swelling decyl sulfate muramyl chlor- wall plus trypsin pentapeptide amphenicol defects Wild type, control Normal Yes Normal No N/A N/A No Ion-, control High No _ No N/A N/A No Penicillin-induced filaments Wild type N/A - _ No Normal Yes No Ion N/A - _ Normal No No UV-induced lonfilaments - _ - No Growth in FIM Wild type - _ Normal - No Ion- High Yes No a N/A, not applicable;-, not done; FIM, filament-inducing medium in Ion- cells; does not induce filaments in the wild type.

11 VOL. 38, 1974 BACTERIAL CELL DIVISION CONTROL related protein. When CRT83 is returned to permissive conditions, division resumes after 10 min. The recovery becomes insensitive to chloramphenicol after 5 min, but remains sensitive to reshift to 42 C for the full 10 min. A similar situation in the temperature-sensitive division mutant E. coli ts2o (to be discussed later) was interpreted as suggestive of complex protein interaction. Difference in alleles of dnaa mutants have been found with respect to the time at which division stops after shift to restrictive conditions, recovery of division in the presence of chloramphenicol at the time of reshift to permissive conditions, reinitiation of rounds of replication in the presence of chloramphenicol, the number of growing points introduced upon reversal of inhibition, susceptibility to salt suppression, and response to integrative suppression (42, 56). Much caution must be exerted in interpreting the existence of different phenotypes in various dnaa (or other dna-) mutants, because it has not always been proven that they are mutants in a single locus. Nitrosoguanidine, the commonly used mutagen, induces multiple closely spaced mutations, which may lead to misinterpretations of apparent pleiotropic effects (e.g., 11). Protein Synthesis Although protein synthesis is, of course, required for the continued progress through the cell cycle which leads to division, division can take place in its absence (15, 71, 96, 101, 111, 141). In fact, in some cases division may be stimulated by blocking protein synthesis (43, 152). Pierucci and Helmstetter (111) found that there is a transition (execution) point in the cell cycle after which division can occur in the absence of protein synthesis. They found that 40 min of protein synthesis (normally coincident with period C) is required for division and that cells in the D period can divide in the absence of protein synthesis. The required 40-min period of protein synthesis can be dissociated from the normally concurrent DNA synthesis. It is not known if the equal length of both periods is more than a coincidence. Protein or RNA synthesis appears to be needed for some final processing step of the chromosome which is required for cell division (23). Jones and Donachie (71) recently suggested that special proteins may be involved in connecting DNA synthesis to cell division. They found that the synthesis of certain proteins cannot be dis- 209 sociated from DNA synthesis. These appear to be triggered by termination of rounds of replication and have been termed "termination proteins." Their results also suggest that some of the events in the normal D period are triggered by the 40-min period of protein synthesis. These events require neither previous termination of rounds of replication nor concurrent protein synthesis. Jones and Donachie proposed a model in which DNA synthesis and the synthesis of "division proteins" occur in parallel pathways which are normally concurrent. When the round of replication terminates, "termination protein" synthesis is triggered. Normally at this time enough "division proteins" have been made to trigger a series of events needed for division which do not require concurrent protein, RNA, or DNA synthesis. At a late stage of this series of events the preformed "termination proteins" are utilized and division follows. Marunouchi and Messer (91a) reported that a short terminal segment of the chromosome is not replicated during amino acid starvation. (This segment is replicated when a conditional dna A- mutant is placed under restrictive conditions.) When amino acids are added, cells divide provided the terminal segment is allowed to replicate. This result does not disprove the conclusion reached by Jones and Donachie (71), that termination is required for the synthesis of proteins needed for division, but it does contradict some of their data. Both groups inhibited protein synthesis and then inhibited DNA synthesis after relieving the inhibition of protein synthesis. Marunouchi and Messer (91a) found no cell division, which is the basis of their report, whereas Jones and Donachie (71) found that such cells divided. However, these differences may be attributed to the fact that Jones and Donachie (71) inhibited DNA synthesis by thymine starvation and Marunouchi and Messer (91a) by adding nalidixic acid. Dix and Helmstetter (27) reported that, in analogous experiments, the two treatments differ in that thymine starvation, but not nalidixic acid treatment, allows cells to divide after relieving the inhibition of protein synthesis. The matter, therefore, awaits further clarification. Evidence that protein synthesis is required for the initiation, but not for the continuation of the division process, has been obtained from studies with mutants of B. subtilis 35ts (97) and tms 12 (15), which are thought to be defective in the initiation of septum formation. If protein synthesis is inhibited at the time of reshift to permissive condition, recovery of division is

12 210 SLATER AND SCHAECHTER BACTERIOL. REV. blocked. If a short time elapses before protein synthesis is ihibited, division occurs some time later. Apparently a brief interval of protein synthesis is required to initiate division but once it is initiated, it continues in the absence of further protein synthesis. Similar observations have been made with temperature-induced filaments of E. coli ts2o (101), where they have been interpreted to be due to interactions between protein subunits, and with penicillininduced filaments of E. coli B (Ion-) (141), where they are thought to be due to alterations in properties of the envelopes. "Division Potential" We will now consider a speculative notion regarding the activity of proteins directly involved in division. Let us imagine that some of these proteins are under a special regulatory control, that they are concentrated at or near the site of division, and that conditions may dictate their utilization later than in the normal cycle. We will call these concentrations of relevant substances the "division potential." This is an extension of the use of that term as proposed by Reeve and Clark (115, 116). It is stimulated in good part by their studies on the kinetics of division of a temperature-sensitive filament former, E. coli BUG 6, by the proposals for the involvement of the membrane in initiation of cell division by Ingram and Fisher (62, 63) and by the pioneering studies of Adler et al. (2) Ṫhe case for "division potential." There are indications that some membrane and periplasmic proteins are specifically related to cell division. Examples of these indications are as follows: it was found by Shen and Boos (134) that E. coli BUG 6 is deficient in some periplasmic proteins and that there is a correlation between cell division and the synthesis of several of these proteins. Inhibition of cell division has been correlated with changes in membrane proteins (46, 57, 68, 140) and membrane-bound enzymes (107). The appearance of several proteins in the membrane is more resistant to different antibiotics than protein synthesis in general (53). In addition, several membrane proteins are related to DNA metabolism (57, 68) Ẇhen filaments are placed under permissive conditions, division is sometimes very rapid, suggesting that "division potential" may have accumulated while division was inhibited (29, 87, 116). The case for "division potential" is obviously weak, and we believe that it rests more on its plausibility than on actual data. What "division potential" may explain. The concentration mechanism involved in building up "division potential" may help explain two puzzling aspects of cell division: (i) the general sensitivity of division to different stresses, and (ii) the fact that slow growth tends to favor cell division. The concentration mechanism of "division potential" may be a cellular feature unique to division and may be sufficiently delicate to be the target of many unrelated stresses. Any alteration of the organization of the cytoplasm or the cell membrane may affect it, lowering its concentration at its site of action and preventing cell division. The formation of filaments occurs more readily in cells growing rapidly than in cultures growing at slow rates. Filaments of several mutants divide when shifted to media supporting slow growth rates. It has been proposed that filament formation in these mutants is due to lowering of production of division promoting substances (2). Rapid growth, by greater production of cell mass, may lower the chances of achieving their required concentration at the division site. Slowing growth by treatments affecting the synthesis of cytoplasmic proteins selectively may increase the chance of achieving the proper local concentration of divisionrelated proteins. In certain permeability mutants, substances comprising "division potential" may be lost by leakage through membranes. An example of this may be filament-forming mutants of the blue-green alga Agmenellum quadruplicatum which leak a division-related effector (61, 62). Erwinia sp. filaments produced by ultraviolet light or addition of fl-alanine and D-serine leak proteins into the medium (45). These proteins originate from the periplasm and perhaps the cell wall, but not from the cytoplasm. The filaments also have decreased amounts of murein. Adding pantoyl lactone or Carbowax prevents leakage and inhibition of cell division, but does not affect the decrease in the amount of murein. These results suggest that division is prevented by leakage of proteins from the peripheral area of the cell. This phenomenon would explain the observation that cell division is impaired in some E. coli mutants which require high sucrose concentrations for survival (90). Placement of the Site of Division (i) Loss of control. A mechanism which places sites for division at regular cell length intervals continues to operate whether or not

13 VOL. 38, 1974 BACTERIAL CELL DIVISION CONTROL 211 divisions actually occur. In several cases (e.g., E. coli BUG 6 [116] and S. typhimurium DivA [24]) shifting filaments to permissive conditions leads to rapid expression of all of the divisions missed at the restrictive condition. Normalsized cells are eventually produced. As mentioned earlier, addition of low concentrations of penicillin prevents division and causes formation of bulges at potential division sites. These results indicate that during filament formation, the ability to measure normal cell lengths for placement of the division site is not lost. This ability can be abolished or altered by other mutations or treatments. Thus, penicillininduced bulge formation can be prevented without affecting general sensitivity to penicillin (130) when DNA synthesis is inhibited by nalidixic acid or when an E. coli dna - B mutant is placed at the nonpermissive temperature (130). In E. coli BUG 6 it was shown that placement of the site of division in filaments depends on completion of synthesis or segregation of chromosomes (115, 117). Cells produced in the absence of DNA synthesis in the dna - B. subtilis mutant ts151 (95) and the E. coli dnab mutant ts27 (64) are not of uniform size. In B. subtilis div 355ts (96), nuclear bodies are irregularly spaced in filaments, and irregularly sized cells are produced when they are transferred to permissive conditions. These results suggest that placement of the division site and the completion or segregation of chromosomes are related. However, as we have discussed above, this relationship can be bypassed by specific mutations because uniformly sized cells are produced by E. coli mutants defective in chromosome segregation or in the initiation or continuation of DNA synthesis, provided that secondary "div" mutations are present (54, 56). In S. typhimurium 11G, where no div mutations are found (138), uniformly sized cells as well as uniformly spaced penicillin-induced bulges (131) are produced in the absence of initiation of DNA synthesis. Nutritional shifts-up at the restrictive temperature cause increases in the rate of division, but do not change the size of the cells (131, 132). Apparently, the uniform placement of the division site is intact, but the mechanism which alters the cell size in response to nutritional conditions is not. It is not clear if bulge formation is related to DNA initiation. In S. typhimurium, bulge formation occurs in the DNA initiation mutant 11G (131), which divides at the nonpermissive temperature, but not in the E. coli DNA elongation mutant, CR34-43 (130), which does not. Bulge formation has not been tested in the dnaa or C (initiation) mutants which do not divide, or in the dnab (elongation) mutants, which do divide. In B. subtilis the loss of control of proper placement of the septum occurs with or without alterations in the structure of the septa. B. subtilis divb mutants do not form proper septa and have irregularly spaced membranous protrusions (143). In B. subtilis div 355ts and diva, septa are misplaced, but their structure appears to be normal in ultrathin sections (97, 119, 143). Thus, the control of septum placement can be lost without disturbing the division process itself. The results discussed above indicate that: (i) the proper placement of potential division sites can occur under some conditions where division is prevented, (ii) other division-inhibiting conditions abolish the control of placement, (iii) abolishing the placement control does not necessarily alter the proper orientation or structure of the misplaced septa, (iv) normally, inhibition of proper chromosome synthesis and/or segregation inhibits proper placement, but this dependency can be overcome, and (v) the determination of cell size by the division process can be dissociated from the mechanism that alters that size in response to nutritional conditions. (ii) Alteration of control. We have concerned ourselves with the production of cells of variable length, indicating that they have lost the control in the placement of division site. In other cases, cells of abnormal but uniform size are produced, indicating that this control is not destroyed but is altered. Cells of E. coli P exhibit both normal (median) and "minicell"-producing (terminal) divisions at an average ratio of 1: 2 (1, 5). Extensive studies with this strain suggest a fairly explicit model, that a division site close to a pole becomes available, in addition to the normal median site. Either site can be activated by the cell's normal signal to initiate division. The choices are random, but mutually exclusive. Once it is triggered, the division process at either site is normal. Because minicells lack chromosomal DNA, in this mutant the normal position of DNA relative to the division site is not required for the synthesis of a properly oriented and structured septum. Thus far, the surface features of the minicell are indistinguishable from those of normal cells (2). This is an important fact because minicells, by their geometry, contain a high percentage of septum material. The production of minicells in this strain involves mutations in at least two genes. Their role is not yet known, but this situation may be

14 212 SLATER AND SCHAECHTER BACTERIOL. REV. compared to that seen in other cases such as dna- div (58) or dna- reca- mutants (65), where one mutation blocks DNA synthesis and the other uncouples division from its normal relationship with DNA synthesis. In the minicell former, one mutation may increase the frequency at which division sites are made and the other may destroy a mechanism which normally ensures that accidently (or prematurely) made extra sites are never used in one cell. The phenotype of B. subtilis div IV-B1 (119) is intermediate between the two types of septum-placement mutants discussed so far. The division site is neither completely randomized nor is it as strictly controlled as in the E. coli minicell producer. The spectrum of cell sizes in this B. subtilis mutant ranges from tiny spherical minicells to short rods to normal-sized cells. Some anucleate short rods divide, a unique finding which suggests either that a potential division site may occasionally be trapped in a short rod or that a new site is made by the anucleate cell. In this mutant, short rods and minicells are produced throughout the growth cycle, but their frequency increases during late log to stationary phase. Thus, their production is dependent on the physiological state of the cells (119). Stationary phase cultures of wild-type B. subtilis often contain very small "mini"-type cells (102). The mutant may be defective in the mechanism that prevents the abnormal divisions, which occasionally occur when nutritional conditions become suboptimal. Recent results (S. L. Coyne and N. H. Mendelson, personal communication) indicate that minicell divisions occur one-third of the time, instead of one-half of the time as in E. coli. The probability of a minicell division is greater at the site of the oldest cell pole. (The phenotype of this mutant may be due to an abnormal activation of genes for sporulation, which also takes place at the pole of the cell.) In addition, not all minicell divisions are normally oriented. For those reasons, the E. coli and B. subtilis minicell producers may be fundamentally different. In an earlier section it was stated that nearly all possible combinations for the relationship between DNA synthesis and division have been found. This also appears to be true for division and the placement of the division site. Certain treatments which pevent division do not affect the uniform placement of potential division sites, whereas others do. In some cases the mechanism is destroyed, whereas in others the placement is changed from one controlled state to another. The same type of statements can be made with respect to placement of the division site and DNA synthesis or segregation. In some cases, but not others, interference with DNA synthesis or segregation affects proper division site placement. The proper placement of division sites has also been dissociated from the mechanism which relates cell size to the growth rate. In all of the cases discussed in this section, the misplaced septa are properly oriented and structured. The site, but not the process, of division is affected in all of the mutants discussed so far. We turn now to cases where both site and process are affected and find that this is associated with generally defective wall synthesis. Cell Wall Logically, there must be a direct connection between cell division and the proper synthesis and functioning of the cell envelopes. We will not discuss here the mechanisms of wall synthesis, the association of envelope components with each other, or the structural role of the components in wall or cellular structure. We will deal here only with the dependence of division on the elastic properties of the wall and on its shapedetermining properties. Elastic properties of the wall. It has been proposed that division is initiated by the invagination of the cell membrane, as a result of changing the direction of the membrane synthesis from the peripheral direction to a bidirectional (peripheral and septal) direction. Such an invagination could result from a transient increase in membrane synthesis relative to wall synthesis and would depend on the proper rigidity of the peripheral wall (10, 52). Such dependence may be inferred from studies on the requirements for protoplast reversion (86, 127), on the effect of sudden changes in osmotic conditions on swelling and division (90), and on the effects of inhibiting murein synthesis on division (130). It is also indicated by studies with specific mutants. Four out of five E. coli mutants selected for dependence on high osmotic conditions (sud mutants) grow as filaments in the absence of high sucrose, whereas the fifth mutant lyses (90). Addition of wall constituent D-alanine allows one of the filaments chosen for study (sud 25) to divide in low-sucrose medium (90.) We discussed elsewhere the possibility that the E. coli Ion- might be defective in the control of wall synthesis and that the resulting altered envelope might decrease the affinity of an envelope receptor for a signal linking DNA

15 VOL. 38, 1974 BACTERIAL CELL DIVISION CONTROL synthesis to division. E. coli B (lon-) is among the mutants which fail to swell upon sudden shift in osmotic properties (148), indicating an alteration in the elastic properties of the wall. The studies with lon- and sud- mutants suggest that those wall characteristics which promote proper elasticity (and rigidity) are required for the initiation of division. Determination of cell shape and its relation to division. It may be expected that distortions in cell shape may affect the proper orientation or formation of the cell septum and thus affect cell division. Studies with various mutants of B. subtilis and E. coli have shown that different changes in cell shape affect cell division to different degrees. This type of relationship between the structural properties of the wall and cell division may be particularly fruitful because something is known about the chemical determinants of the shape of bacilli. Thus, in E. coli it has been long held that the shape is determined primarily by the murein sacculus (e.g., 113 and 120). Now, it appears that the outer membrane may also play a role in this process because "ghosts" lacking murein, but containing outer membrane, maintain the cell shape (50). A particularly propitious case for study may be Bacillus brevis, whose wall contains a protein whose purified subunits, it is claimed, reassociate to form cylinders with the diameter of the normal cells (16, 19, 51). Certain mutants of B. subtilis, called rod-, can change reversibly from a rod to a sphere. The spheres show morphological defects in the structure of the wall and especially of the cross walls which appear "chewed," misplaced, and misoriented. There is defective splitting of the septa, resulting in the formation of clumps of cells. Correcting the defect in wall synthesis by placing the cells at the permissive conditions usually restores proper cross wall formation (13, 14, 26, 36, 76, 77, 118, 120, 121, 122). However, during recovery some of the rod forms have normal-looking peripheral walls, but "motheaten" cross walls (122). Thus, defective wall synthesis can cause obstructions in formation of septa. Correcting the defect usually, but not always, restores septa to normal. The converse of this is also true, when the rod- mutant is placed in a different genetic background. Here the ultrastructure and shape of the wall remain defective, but properly placed and oriented septa are made (Fig. 5). Studies with several gram-negative mutants have shown that cells with altered cell shapes can divide (50, 92, 103) (Fig. 6). In none of the cases discussed here is the 213 biochemical site of the lesion known with precision. However, important hints on the processes involved have been obtained from studies with a mutant of Bacillus licheniformis (39). This mutant provides a poignant example of the ability of cells to bypass normal causal connections. When cells are grown in low-phosphate media, teichoic acid in their walls is replaced by teichuronic acid. When teichuronic acid synthesis is blocked by a mutational change in the enzyme phosphoglucomutase, cells exhibit a spherical phenotype. The requirement for phosphoglucomutase is lost when glycerol and galactose are added to the medium. Under these conditions, lytic activity increases and the rodshaped morphology is maintained (39). Not only does this study exemplify the possible alternate means and bypasses which bacteria use but, in an operational sense, it illustrates the difficulties in assigning primary regulatory roles to biochemical effectors. In conclusion, it is evident that septum synthesis depends on definable properties of the cell wall. Of these, the proper determination of cell shape is not absolutely required and can be bypassed in certain mutants. Small-Molecule Metabolism We will list here a variety of somewhat unrelated observations which indicate that interference with cell division can be traced to the metabolism of some small molecules. In keeping with previous discussions, it is not possible to infer the biochemical relevance of these connections. Amino acids and amines. Filaments induced by high temperature in E. coli ts 52 (152) or by D-serine in Erwinia (43) divide when treated with chloramphenicol. The proposed interpretation of this finding is that when protein synthesis is blocked, small molecules accumulate which, conceivably, stimulate division (152). Because rapid growth increases the tendency to form filaments, it may decrease the level of accumulation of these molecules. In some species, the proper balance of putrescine to spermidine is necessary for division (67). A mutant of E. coli (MA-135), conditionally deficient in putrescine synthesis, forms filaments in rich medium, but not in minimal medium. When putrescine levels are decreased further, growth stops (149). The level of polyamines depends on amino acid metabolism. It seems possible, therefore, that a block in protein synthesis makes amino acids increasingly available for polyamine synthesis. This, in turn, may stimulate division and account for the phenotype of E. coli ts 52.

16 214 SLATER AND SCHAECHTER BACTERIOL. REV. v,.,y:> K.., A,>t 00- "nt,1f,,-0-,,.,j :.s. J.S - l I.'4 ] s a OL Downloaded from on September 7, 2018 by guest FIG. 5 Grampositive cells: altered shape with normal division in B. subtilis mutant tag-1. This temperature-sensitive mutant was grown for 3 h at 45 C. OL indicates the dense and particulate outer layer of wall. Note the lack of inner layer and periplasm. The distance between arrowheads indicates the width of the membrane. Magnifications: high power, x49,000; low power, insert, x10,500. Both bars designate 1gm; reproduced from reference 118 by permission of N. H. Mendelson and the publisher, Springer- Verlag, Berlin-Heidelberg-New York.

17 VOL. 38, 1974 When a S. typhimurium mutant constitutive in the histidine operon is placed in a medium containing a high concentration of glucose, it forms filaments (100). Another S. typhimurium mutant, diva (ts), forms filaments at the restrictive temperature in glucose, but not in succinate-containing medium (24). Cyclic adenosine 5'-monophosphate causes division of those filaments. Azide stimulates division in E. coli ts 52, but azide and iodoacetamide together block it (152). Purines and pyrimidines. E. coli tif- forms filaments at 40 C. Division takes place in the presence of added guanine, cytidine, hedacine (an inhibitor of adenylic acid synthesis), and pantoyl lactone. The protection is reversed by adenine. Mutant and wild-type cells exhibit the same changes in nucleotide pools at restrictive and permissive temperature, but the mutant may respond differently to these changes. The key small molecule effector appears to be a substituted furan moiety (80, 81, 124). BACTERIAL CELL DIVISION CONTROL CONCLUSIONS We conclude that bacterial cell division is a complex series of events which can be measured in different ways, and that its causal dependencies are diverse, interrelated, and variable. Division may be measured by the functional separation of sister cytoplasms, by the completion of the septum, or by separation of the sister cells. There are indications that sister cytoplasms become functionally separated only at the time of completion of the septum. Little is known about the timing of septum formation during the cell cycle. Cell separation in B. subtilis apparently requires the same time, regardless of the rate of growth. Studies with mutants of gram-negative bacteria indicate that the cell separation step is under genetic control and related to the relative timing of the synthesis of various envelope components of the "septum." There is reason to believe that gram-negative rods, like gram-positives, divide by formation of a morphologically distinct septum and not, as commonly held, by gradual constriction of the cell envelopes. The details of the mode of division and its timing must be known for the proper definition of mutant phenotypes. Specificity of division controls. In many bacteria, division is extremely sensitive to a wide range of growth inhibitors and gene mutations. Low doses of inhibitors, insufficient to slow growth, will preferentially stop division. This makes it difficult to define the degree of specificity with which physiological manipula- 215 FIG. 6. Gram-negative cells: altered shape with normal division. Electron micrograph of thin section of cells of the roda mutant of E. coli grown in L-broth at 37 C. Marker bar represents 0.2 Am; reproduced from reference 92 by permission of H. Matsuzawa and the publisher, The American Societ~y for Microbiology. tions or gene mutations are involved in the regulation of division. When conditional mutants are used to probe the causal connections of a cell cycle, it is usually assumed that the target of a restrictive condition is the sensitized product of a mutated gene. In several cases, such as E. coli Ion-, B. lichenformis mutants, and probably E. coli tif-, this is not so. Thus, while the target for ultraviolet light in E. coli ion- is DNA, the mutant is not defective in DNA synthesis or its regulation, but in the response of the cell to unbalanced decreases in DNA synthesis. Diversity of division controls. Division depends on many cellular activities, among which we can list the synthesis of macromolecules, the metabolism of small molecules, and certain properties of the cell envelope. More specifically, division is related to several aspects of DNA metabolism: replication, repair, and perhaps recombination functions, and the levels of precursor pools. Division also requires the synthesis of proteins made at or near the completion of rounds of DNA replication, as well as proteins normally made during the round of replication but dissociable from it. Division ap-

18 216 SLATER AND SCHAECHTER BACTERIOL. REV. parently depends on properties of the cell envelopes that contribute to the rigidity of the wall and probably to the determination of the proper shape of the cell. Although murein plays an important role in determining and maintaining the shape of the cell, recent evidence indicates that special envelope proteins may also be involved in this process. Despite uncertainties, the biochemistry of envelopes and chromosomes is more thoroughly understood than the myriad of other factors upon which division depends. It must be emphasized that at present there is no biochemical means to assay the initiation of division or to study septum synthesis separately from the synthesis of peripheral wall or membrane. Interrelationships of division controls. Factors related to division are related to each other. Thus, shape determination, permeability, and envelope composition are related to DNA metabolism-possibly by the same effectors that link DNA synthesis to division. In no case is there a molecular explanation for these relationships. Variability of division controls. While some of the basic assumptions in the use of inhibitors or mutants may be too simplistic for the complexity of the system being investigated, we may conclude that there is not a unique and obligatory set of causal connections in the cell cycle. When normal relationships are interrupted they may be bypassed or they may be modified to produce artifacts. Bypasses and artifacts have complicated studies of viral systems where the genes and the gene products can be defined. In bacterial division where such knowledge is not available and where a black box approach must be used, the possibility of such bypasses and artifacts renders the study even more intractable. There are several indications that normal causal relationships in cell division can indeed be bypassed. Thus, although division normally depends on DNA synthesis, this requirement is lost in some mutants. Sometimes the new pathway may require additional mutations for its expression. Division also appears to require normal cell shape and structure of the wall, but certain mutants with altered shape and/or wall structure can divide. The main event may be that, at restrictive conditions, there are possibilities for artifacts, and that what is observed under such conditions is not relevant to the "normal" cycle. We would like to end by stating that in cell division, as in other cellular phenomena, what can be learned from conditional mutants is restricted to what can be learned about their phenotype. Unfortunately, present knowledge of the morphological and biochemical events leading to cell division does not permit the detailed characterization even of the mutants that are available. A recent method for ordering the sequence of complex functions by using an alternation of different restrictive treatments has recently been developed for the study of the bacteriophage P22 (70). It is being used in the analysis of division mutants of yeast and may well be useful in studies with bacteria. The use of such genetic tools, along with further biochemical studies of cell functions likely to be related to division, remains still the most hopeful of the available approaches. The complexity in regulation of cell division in bacteria may be attributed in part to their relative structural simplicity. Bacteria have not evolved the complex machinery that ensures equipartition of the genome in higher cells. It seems likely that they utilize structures and regulatory mechanisms that may be used also for other cell functions. Therefore, it is not surprising that there is not a unique cell cycle, but rather that there is a "normal" sequence of events with alternate sequences, should the normal one be interrupted. Speaking of cell division in higher cells, Mazia (94) stated in 1960 that "complex" does not necessarily mean "hard to understand." It should mean, simply, "composed of many parts and of definite relationships between the parts." In 1974, this statement is no less true, but, in the case of bacterial cell division, its optimism may be tempered by recognizing that the structural simplicity of bacteria is deceptive. ACKNOWLEDGMENTS We are grateful to B. Beck, D. Boyd, W. D. Donachie, E. A. Grula, R. James, 0. Maal0e, M. H. Malamy, L. A. McNicol, N. H. Mendelson, S. Normark, J. T. Park, H. J. Rogers, R. Rowbury, B. M. Shapiro, G. Shockman, S. Torti, and A. Wright for their helpful comments. They are not responsible for what we say because we did not always take their advice, a fact we may live to regret. LITERATURE CITED 1. Adler, H. I., W. D. Fisher, A. Cohen, and A. A. Hardigree Miniature Escherichia coli cells deficient in DNA. Proc. Nat. Acad. Sci. U.S.A. 57: Adler, H. I., W. D. Fisher, and A. A. Hardigree Cell division in Escherichia coli. Trans. N.Y. Acad. Sci. 31: Adler, H. I., and A. A. Hardigree Post irradiation growth, division and recovery in bacteria. Radiat. Res. 25: Adler, H. I., W. D. Fisher, H. A. Hardigree, and G. W. Stapleton Repair of radiation-

19 VOL. 38, 1974 BACTERIAL CELL DIVISION CONTROL 217 induced damage to the cell division mechanism of Escherichia coli. J. Bacteriol. 91: Adler, E. I., and A. A. Hardigree Biology and radiobiology of minicells. p In S. Bonotto (ed.), Biology and radiobiology of anucleate systems. Academic Press Inc., New York. 6. Ahmed, N., and J. Rowbury Antibiotics and cell division in a filament-forming mutant of Salmonella typhimurium. Microbios 4: Ahmed, N., D. G. Smith, and R. J. Rowbury Effect of growth medium on the division of a filament-forming mutant of Salmonella typhimurium. Submitted to Arch. Mikrobiol. 8. Allen, G., J. A. Smith, R. C. Knudsen, and J. R. Walker Initial characterization of temperature-sensitive cell division mutants of Escherichia coli. Biochem. Biophys. Res. Commun. 47: Anton, D. N Osmotic-sensitive mutant of Salmonella typhimurium. J. Bacteriol. 109: Bazil, G. W Lethal unbalanced growth in bacteria. Nature (London) 216: Bell, R. M., R. D. Mavis, and P. R. Vagelos Altered phospholipid metabolism in a temperature sensitive mutant of Escherichia coli CR34T46. Biochim. Biophys. Acta 270: Bilsky, A. Z., and J. B. Armstrong Osmotic reversal of temperature sensitivity in Escherichia coli. J. Bacteriol. 113: Boylan, R. J., and N. H. Mendelson Initial characterization of a temperature-sensitive rod- mutant of Bacillus subtilis. J. Bacteriol. 100: Boylan, R. J., N. H. Mendelson, D. Brooks, and F. E. Young Regulation of the bacterial cell wall: analysis of a mutant of Bacillus subtilis defective in biosynthesis of teichoic acid. J. Bacteriol. 110: Breakefield, X. O., and 0. E. Landman Temperature-sensitive divisionless mutant of Bacillus subtilis defective in the initiation of septation. J. Bacteriol. 113: Brinton, C. C., J. E. McNary, and J. Carnahan Purification and in vitro assembly of a curved network of identical protein subunits from the outer surface of a Bacillus. Bacteriol. Proc., p Brown, D., and N. E. Gillis The relationship between filaments, killing and restoration in irradiated Escherichia coli strain B. J. Gen. Microbiol. 70: Cabib, E., V. Farkas, and R. E. Ulane Yeast septum formation as a model system for morphogenesis. In Proceedings of the third international symposium on yeast protoplasts. Academic Press Inc., New York (in press). 19. Casper, D. L., and A. Klug Bacillus subtilis T. antigen. Cold Spring Harbor Symp. Quant. Biol. 27: Castellazzi, M., J. George, and G. Buttin Prophage induction and cell division in E. coli. Mol. Gen. Genet. 119: Castellazzi, M., J. George, and G. Buttin Prophage induction and cell division in E. coli. II. Mol. Gen. Genet. 119: Chapman, G. B., and J. Hiller Electron microscopy of ultrathin section of bacteria. I. Cellular division in Bacillus cereus. J. Bacteriol. 66: Chung, K. L Influence of cell-wall thickness on cell division: electron microscopic studies with Bacillus cereus. Can. J. Microbiol. 19: Ciesla, Z., M. Bagdasarian, W. Szczurkiewicz, M. Prygonska, and T. Klopowski Defective cell division in thermosensitive mutants of Salmonella typhimurium. Mol. Gen. Genet. 116: Clark, D Regulation of deoxyribonucleic acid replication and cell division in Escherichia coli B/r. J. Bacteriol. 96: Cole, R. M., T. J. Popkin, R. J. Boylan, and N. H. Mendelson Ultrastructure of a temperature-sensitive rod- mutant of Bacillus subtilis. J. Bacteriol, 103: Dix, D. E., and C. E. Helmstetter Coupling between chromosome completion and cell division in Escherichia coli. J. Bacteriol. 115: Donachie, W. D Relationship between cell size and time of initiation of DNA replication. Nature (London) 219: Donachie, W. D Control of cell division in Escherichia coli: experiments with thymine starvation. J. Bacteriol. 100: Donachie, W. D., and K. J. Begg Growth of bacterial cell. Nature (London) 227: Donachie, W. D., D. T. Martin, and K. J. Begg Independence of cell division and DNA replication in Bacillus subtilis. Nature N. Biol. 31: Donch, J. J., Y. S. Chung, M. H. L. Green, J. Greenberg, and G. Warren Genetic analysis of sul mutants of Escherichia coli B. Genet. Res. 17: Donch, J., M. H. L. Green, and J. Greenberg Interaction of the exr and Ion genes in Escherichia coli. J. Bacteriol. 96: Fan, D. P Autolysin(s) of Bacillus subtilis as dechaining enzyme. J. Bacteriol. 103: Fan, D. P., and M. M. Beckman Mutant of Bacillus subtilis demonstrating the requirement of lysis for growth. J. Bacteriol. 105: a. Fan, D. P., and M. M. Beckman Structural difference between walls from hemispherical caps and partial septa of Bacillus subtilis. J. Bacteriol. 114: Fan, D. P., M. M. Beckman, and W. P. Cunningham Ultrastructural studies on a mutant of Bacillus subtilis whose growth is inhibited due to insufficient autolysin produc-

20 218 SLATER AND SCHAECHTER BACTERIOL. REV. tion. J. Bacteriol. 109: Fan, D. P., M. C. Pelvit, and W. P. Cunningham Structural difference between walls from ends and sides of the rod-shaped bacterium Bacillus subtilis. J. Bacteriol. 109: Farnsworth, P. A., and R. James The effect of the ton phenotype in Escherichia coli as indicated by growth of amoebae of Dictyostelium discoideum. J. Gen. Microbiol. 73: a. Forsberg, C., and J. J. Rogers Autolytic enzymes in growth of bacteria. Nature (London) 229: Forsberg, C. W., P. B. Wyrich, J. B. Ward, and H. J. Rogers Effect of phosphate limitation on the morphology and wall composition of Bacillus licheniformis and its phosphoglucomutase-deficient mutants. J. Bacteriol. 113: Gellert, M., and M. L. Bullock DNA ligase mutants of Escherichia coli. Proc. Nat. Acad. Sci. U.S.A. 67: Green, M. H. L., J. Greenberg, and J. Donch Effect of a reca gene on cell division and capsular polysaccharide production in a Ion strain of Escherichia coli. Genet. Res. 14: Gross, J. D DNA replication in bacteria. Curr. Top. Microbiol. Immunol. 57: Grula, E. A., and M. M. Grula Cell division in a species of Erwinia. V. Effect of metabolic inhibitors on terminal division and composition of a "division" medium. J. Bacteriol. 84: Grula, E. A., and M. M. Grula Cell division in a species of Erwinia. III. Reversal of inhibition of cell division caused by D-amino acids, penicillin, and ultraviolet light. J. Bacteriol. 83: Grula, E. A., and R. L. Hopfer Cell division in a species of Erwinia. XIII. Leakage of proteins from the peripheral area of filamentous cells. Biochem. Biophys. Acta 255: Grula, E. A., and R. D. King Changes in cell membranes in dividing and non-dividing cells of Micrococcus lysodeikticus dis II p+. Biochem. Biophys. Res. Commun. 44: a. Grula, E. A., and G. L. Smith Cell division in a species of Erwinia. J. Bacteriol. 90: Hartwell, L. H Saccharomyces cerevisiae cell cycle. Bacteriol. Rev. 38: Helmstetter, C. E., S. Cooper, 0. Pierucci, and E. Revelas On the bacterial life sequence. Cold Spring Harbor Symp. Quant. Biol. 33: Henning, U., K. Rehn, V. Braun, H. Bohn, and U. Schwarz Cell envelope and shape of Escherichia coli K12. Properties of a temperature sensitive rod mutant. Eur. J. Biochem. 26: Henning, U., K. Rehn, and B. Hoehn Cell envelope and shape of Escherichia coli K-12. Proc. Nat. Acad. Sci. U.S.A. 70: Henry, C. M Ph.D. thesis, Univ. of Pittsburgh, Pittsburg, Pa. 52. Higgins, M. L., and G. D. Shockman Procaryotes cell division with respect to wall and membrane. Crit. Rev. Microbiol. 1: Hirashima, A., G. Childs, and M. Inouye Differential inhibitory effects of antibiotics on the biosynthesis of envelope proteins of Escherichia coli. J. Mol. Biol. 79: Hirota, Y., F. Jacob, A. Ryter, G. Buttin, and T. Nakai On the process of cellular division in Escherichia coli I. J. Mol. Biol. 35: Hirota, Y., J. Mordoh, I. Scheffler, and F. Jacob Genetic approach to DNA replication and its control in Escherichia coli. Fed. Proc. 31: Hirota, Y., and M. Ricard Production of DNA-less bacteria, p In S. Bonotto (ed.), Biology and radiobiology of anueleate systems. Academic Press Inc., New York. 57. Hirota, Y., M. Ricard, and B. Shapiro The use of thermosensitive mutants of E. coli in the analysis of cell division. Biomembranes 2: Hirota, Y., A. Ryter, and F. Jacob Thermosensitive mutants of E. coli affected in the processes of DNA synthesis and cellular division. Cold Spring Harbor Symp. Quant. Biol. 33: Hirshfield, I. N., H. J. Rosenfeld, Z. Leifer, and W. K. Maas Isolation and characterization of a mutant of Escherichia coli blocked in the synthesis of putricine. J. Bacteriol. 101: Hughes, A. F. W In E. T. C. Spooner and B. A. D. Stocher (ed.), Bacterial anatomy, p th Symp. Soc. Gen. Microbiol., Cambridge; Cambridge University Press. 61. Ingram, L. O., and W. D. Fisher Novel mutant impaired in cell division. J. Bacteriol. 113: Ingram, L. O., and W. D. Fisher Mechanism for the regulation of cell division in Agmenellum. J. Bacteriol. 113: Ingram, L. O., and W. D. Fisher Stimulation of cell division by membrane-active agents. Biochem. Biophys. Res. Commun. 50: Inouye, M Unlinking of cell division from deoxyribonucleic acid replication in a temperature-sensitive deoxyribonucleic acid synthesis mutant of Escherichia coli. J. Bacteriol. 99: Inouye, M Pleiotropic effect of the reca gene of Escherichia coli: uncoupling of cell division from deoxyribonucleic acid replication. J. Bacteriol. 106: Inouye, M Reversal by sodium chloride of envelope protein changes related to DNA replication and cell division of Escherichia

21 VOL. 38, 1974 BACTERIAL CELL DIVISION CONTROL coli. J. Mol. Biol. 63: Inouye, M., and A. B. Pardee Requirement of polyamines for bacterial division. J. Bacteriol. 10: Inouye, M., and A. B. Pardee Changes of membrane proteins and their relation to deoxyribonucleic acid synthesis and cell division of Escherichia coli. J. Biol. Chem. 245: James, R., and N. E. Gillies The sensitivity of suppressed and unsuppressed lonstrains of Escherichia coli to chemical agents which induce filamentation. J. Gen. Microbiol. 76: Jarvik, J., and D. Botstein. A genetic method for determining the order of events in a biological pathway. Proc. Nat. Acad. Sci. U.S.A. 70: Jones, N. C., and W. D. Donachie Chromosome replication, transcription and the control of cell division in Escherichia coli. Nature N. Biol. 243: Kaback, H. R Transport. Annu. Rev. Biochem. 39: Kantor, G. J., and R. A. Deering Ultraviolet radiation studies of filamentous Escherichia coli B. J. Bacteriol. 92: Kantor, G. J., and R. A. Deering Recovery of division ability in ultraviolet-irradiated Escherichia coli induced by photoreactivation, photoprotection, and liquid holding treatment. J. Bacteriol. 94: Kantor, G. J., and R. A. Deering Effect of nalidixic acid and hydroxyurea on division ability of Escherichia coli fil+ and lon- strains. J. Bacteriol. 95: Karamata, D., M. McConnell and H. J. Rogers Mapping of mutants of Bacillus subtilis disturbed in morphology, cell division and in some instances in cell-wall chemistry, p. vi. In Soc. Gen. Microbiol. Proc., 64th Meet. London. 77. Karamata, D., M. McConnell, and H. J. Rogers Mapping of rod Mutants of Bacillus subtilis. J. Bacteriol. 111: Kennel, D., and H. Kotoulas Magnesium starvation of Aerobacter aerogenes. IV. Cytochemical changes. J. Bacteriol. 93: Khachatourians, G. G., and D. J. Clark Regulation of cell division and DNA replication in E. coli. Bacteriol. Proc. 1970: Kirby, Edward, P., F. Jacob, and D. A. Goldthwait Prophage induction and filament formation in a mutant strain of Escherichia coli. Proc. Nat. Acad. Sci. U.S.A Kirby, E. P., W. L. Ruff, and D. A. Goldthwait Cell division and prophage induction in Escherichia coli: effects of pantoyl lactone and various furan derivatives. J. Bacteriol. 111: Koch, A. L., and M. Schaechter A model for statistics of the cell division process. J. Gen. Microbiol. 29: Kojima, M., S. Suda, S. Hotta, K. Hamada, and 219 A. Suganuma Necessity of calcium ion for cell division in Lactobacillus bifidus. J. Bacteriol. 104: Kubitschek, H. E Growth during the bacterial cell cycle. Analysis of cell size. Biophys. J. 9: Kubitschek, H. E., and M. L. Freedman Chromosome replication and the division cycle of Escherichia coli B/r. J. Bacteriol. 107: Landman, 0. E., and S. Halle Enzymically and physically induced inheritance changes in Bacillus subtilis. J. Mol. Biol. 7: Leighton, P. M., and W. D. Donachie Deoxyribonucleic acid synthesis and cell division in a Ion- mutant of Escherichia coli. J. Bacteriol. 102: Lin, E. C. C., Y. Hirota, and F. Jacob On the process of cellular division in Escherichia coli. J. Bacteriol. 108: Loveless, L. E., E. Spoerl, and T. H. Weisman A survey of effects of chemicals on division and growth of yeast and Escherichia coli. J. Bacteriol. 68: Mangiarotti, G., D. Apirion, and D. Schlessinger Selection of sucrose-dependent Escherichia coli to obtain envelope mutants and fragile cultures. Science 153: Marunouchi, T., and W. Messer Deoxyribonucleic acid replication and cell division in Escherichia coli at 33 C. J. Bacteriol. 112: a. Marunouchi, T., and W. Messer Replication of a specific terminal chromosome segment in Escherichia coli which is required for cell division. J. Mol. Biol. 78: Matsuzawa, H., K. Hagakawa, T. Sato, and K. Imahori Characterization and genetic analysis of a mutant of Escherichia coli K-12 with rounded morphology. J. Bacteriol. 115: Matthieu, R., and Y. Hirota Effet des sels sur le processus de division cellulaire d'e. coli. C.R. Acad. Sci. Paris 268: Mazia, D The analysis of cell reproduction. Ann. N.Y. Acad. Sci. 90: Melchior, D. L., H. J. Morowitz, J. M. Sturtevant, and T. W. Tsong Characterization of the plasma membrane of Mycoplasma laidlawii. Biochim. Biophys. Acta 219: Mendelson, N. H Division site regulation in a temperature-sensitive mutant of Bacillus subtilis. J. Bacteriol. 111: Mendelson, N. H., and R. M. Cole Genetic regulation of cell division initiation in Bacillus subtilis. J. Bacteriol. 112: Mendelson, N. H., and J. R. Reeve Growth of the Bacillus subtilis cell surface. Nature N. Biol. 243: Mount, D. W., A. C. Walker, and C. Kosel Suppression of lex mutations affecting deoxyribonucleic acid repair in Escherichia coli

22 220 SLATER AND SCHAECHTER BACTERIOL. REV. K-12 by closely linked thermosensitive mutations. J. Bacteriol. 116: Murray, M. L., and P. E. Hartman Overproduction of his H and his F gene products leads to inhibition of cell division in Salmonella. Can. J. Microbiol. 18: Nagai, K., and G. Tamura Mutant of Escherichia coli with thermosensitive protein in the process of cellular division. J. Bacteriol. 112: Neal, E. K., and G. B. Chapman Effect of low temperature on the growth of Bacillus subtilis. J. Bacteriol. 104: Normark, S Mutation in Escherichia coli K-12 mediating spherelike envelopes and changed tolerance to ultraviolet irradiation and some antibiotics. J. Bacteriol. 98: Normark, S., H. G. Boman, and G. D. Bloom Cell division in a chain-forming enva mutant of Escherichia coli K12. Acta Pathol. Microbiol. Scand. Sec. B. 79: Normark, S., H. G. Boman, and E. Matsson Mutant of Escherichia coli with anomalous cell division and ability to decrease episomally and chromosomally mediated resistance to ampicillin and several other antibiotics. J. Bacteriol. 97: Normark, S., and B. Westling Nature of the penetration barrier in Escherichia coli K-12: effect of macromolecular inhibition on penetrability in strains containing the enva gene. J. Bacteriol. 108: Ohki, M Correlation between metabolism of phosphatidylglycerol and membrane synthesis in Escherichia coli. J. Mol. Biol. 68: Onken, A., and W. Messer Cell division in Escherichia coli. Septation during synchronous growth. Mol. Gen. Genet. 127: Paulton, R. J. L Analysis of the multiseptate potential of Bacillus subtilis. J. Bacteriol. 104: Pierucci, Chromosome replication and cell division in Escherichia coli at various temperatures of growth. J. Bacteriol. 109: Pierucci, O., and C. E. Helmstetter Chromosome replication, protein synthesis and cell division in Escherichia coli. Fed. Proc. 28: Powell, E An outline of the pattern of bacterial generation times. J. Gen. Microbiol. 18: Previc, E. P Biochemical determination of bacterial morphology and the geometry of cell division. J. Theor. Biol. 27: Previc, E., and S. Richardson Growthphysiological changes in Escherichia coli and other bacteria during division inhibition by 5-diazouracil. J. Bacteriol. 97: Reeve, J. N Cell division in a temperature sensitive mutant of Escherichia coli. Ph.D. thesis, Univ. of British Columbia, Vancouver, B.C., Canada Reeve, J. N., and D. J. Clark Cell division of Escherichia coli BUG-6: effect of varying the length of growth at the nonpermissive temperature. J. Bacteriol. 110: Reeve, J. N., D. J. Groves, and D. J. Clark Regulation of cell division in Escherichia coli: characterization of temperature-sensitive mutants. J. Bacteriol. 104: Reeve, J. N., N. H. Mendelson, and R. M. Cole Cell morphology of Bacillus subtilis: the effect of genetic background on the expression of a rod- gene. Mol. Gen. Genet. 119: Reeve, J. N., N. H. Mendelson, S. I. Coyne, L. L. Hallock, and R. M. Cole Minicells of Bacillus subtilis. J. Bacteriol. 114: a. Ricard, M., and Y. Hirota Process of cellular division in Escherichia coli: physiological study on thermosensitive mutants defective in cell division. J. Bacteriol. 116: Rogers, H. J Bacterial growth and the cell envelope. Bacteriol. Rev. 34: a. Rogers, H. J., and M. McConnell The role of L-glutamine in the phenotypic change of a rod mutant derived from Bacillus subtilis 168. J. Gen. Microbiol. 61: Rogers, H. J., M. McConnell, and I. D. J. Burdett Cell wall or membrane mutant of Bacillus subtilis and Bacillus licheniformis with grossly deformed morphology. Nature (London) 219: Rogers, H. J., M. McConnell, and I. D. J. Burdett The isolation and characterization of mutants of Bacillus subtilis and Bacillus licheniformis with disturbed morphology and cell division. J. Gen. Microbiol. 61: Rowbury, R. J Bacterial cell division: its regulation and relation to DNA synthesis. Sci. Progr. 60: Ruff, W., E. P. Kirby, and D. A. Goldwarth Cell division and prophage induction in Escherichia coli: Studies of nucleotide levels. J. Bacteriol. 106: Russel, R. R. B Temperature-sensitive osmotic remedial mutants of Escherichia coli. J. Bacteriol. 112: Ryter, A., Y. Hirota, and V. Schwarz Process of cellular division in Escherichia coli. Growth pattern of E. coli murein. J. Mol. Biol. 78: Ryter, A., and D. E. Landman Electron microscope study of the relationship between mesosome loss and the stable L state in Bacillus subtilis. J. Bacteriol. 88: Schaechter, M Patterns of cellular control during unbalanced growth. Cold Spring Harbor Symp. Quant. Biol. 26: Schaechter, M., J. P. Williamson, J. R. Hood, Jr., and A. L. Koch Growth, cell and nuclear division in some bacteria. J. Gen. Microbiol. 29: Schwarz, U., A. Asmus, and H. Frank Autolytic enzymes and cell division of Esche-

23 VOL. 38, 1974 BACTERIAL CELL DIVISION CONThOL richia coli. J. Mol. Biol. 41: Shannon, K. P., and R. J. Rowbury Alteration of the rate of cell division independent of the rate of DNA synthesis in a mutant of Salmonella typhimurium. Mol. Gen. Genet. 115: Shannon, K. P., B. G. Spratt, and R J. Rowbury Cell division and the production of cells lacking nuclear bodies in a mutant of Salmonella typhimurium. Mol. Gen. Genet. 118: Shaw, M. K Formation of filaments and synthesis of macromolecules at temperatures below the minimum for growth of Escherichia coli. J. Bacteriol. 95: Shen, B. H. P., and W. Boos Regulation of the beta methylgalactoside transport system and the galactose-binding protein by the cell cycle of Escherichia coli. Proc. Nat. Acad. Sci. U.S.A. 70: Smith, H. S., and A. B. Pardee Accumulation of a protein required for division during the cell cycle of Escherichia coli. J. Bacteriol. 101: Spratt, B. G., and R. J. Rowbury A mutant in the initiation of DNA synthesis in Salmonella typhimurium. J. Gen. Microbiol. 64: Spratt, B. G., and R. J. Rowbury Cell division in a mutant of Salmonella typhimurium which is temperature-sensitive for DNA synthesis. J. Gen. Microbiol. 65: Spratt, B. G., and R. J. Rowbury Physiological and genetical studies on a mutant of Salmonella typhimurium which is temperature-sensitive for DNA synthesis. Mol. Gen. Genet. 114: Steed, P., and R. G. E. Murray The cell wall and cell division of gram-negative bacteria. Can. J. Microbiol. 12: Starka, J Cell envelope proteins of dividing and nondividing cells of E. coli. FEBS Lett. 16: Starka, J., and J. Moravova Cellular division of penicillin-induced filaments of 221 Escherichia coli. Folia Microbiol. Prague 12: Strack, H. B., and J. H. Cox Lambdoid prophages modifying the kinetics of host cell division. Virology 43: a. Tomasz, A., A. Albino, and E. Zanati Multiple antibiotic resistance in a bacterium with suppressed autolytic system. Nature (London) 227: Van Alstyne, D., and M. I. Simon Division mutants of Bacillus subtilis: isolation and PBS1 transduction of division-specific markers. J. Bacteriol. 108: Walker, J. R., and A. B. Pardee Conditional mutations involving septum formation in Escherichia coli. J. Bacteriol. 93: Walker, J. R., and J. A. Smith Cell division of the Escherichia coli lon- mutant. Mol. Gen. Genet. 108: Webb, M The influence of magnesium on cell division. J. Gen. Microbiol. 3: Weinbaum, G Characteristics of cell walls from morphological variants of Escherichia coli. J. Gen. Microbiol. 42: Wu, P. C., and A. Pardee Cell division of Escherichia coli: control by membrane organization. J. Bacteriol. 114: Young, D. V., and P. R. Srinivasan Regulation of macromolecular synthesis by putrescine in a conditional Escherichia coli putrescine auxotroph. J. Bacteriol. 112: Young, F. E., and G. A. Nelson Genetics of Bacillus subtilis and other gram-positive sporulating bacilli, p In H. 0. Halvorson, R. Hanson, and L. L. Campbell (ed.), Spores V. American Society for Microbiology, Washington, D.C Zaritsky, A., and R. H. Pritchard Changes in cell size and shape associated with changes in the replication time of the chromosome of Escherichia coli. J. Bacteriol. 114: Zusman, D. R., M. Inouye, and A. B. Pardee Cell division in Escherichia coli: evidence for regulation of septation by effector molecules. J. Mol. Biol. 69: