Chapter 102

Timing of Synthetic Activities in the Cell Cycle

CHARLES E. HELMSTETTER

INTRODUCTION

Cell division in Escherichia coli is an end and not a beginning. The biosynthetic steps leading to division begin well before the previous division at all growth rates. This chapter presents information on the timing and duration of events that the cell undergoes as it duplicates its constituents and eventually divides by binary fission to form two essentially identical daughter cells. During the course of this duplication process, there are four clearly discernible actions: initiation and termination of chromosome replication, and initiation and termination of cell division. The timing of synthetic processes will be related to these acts whenever possible.

THE DIVISION CYCLE

Slowly Growing Cells

The first portion of this chapter will describe the means to determine the relationship between chromosome replication and cell division for cells growing at any rate. It is essential to understand this relationship, and the manner in which it varies with growth rate, to fully understand the basis for the timing of the events to be described later. This relationship is determined by applying the "I + C + D rule" (66) to a hypothetical cell containing a single nonreplicating chromosome, where I is the time required to achieve the capacity for initiation of chromosome replication (i.e., the interinitiation time), C is the time for a round of chromosome replication, and D is the time between completion of chromosome replication and the subsequent cell division. According to this rule, cells start their duplication by performing the biosyntheses needed for initiation of chromosome replication, and when that is completed in I minutes, chromosome replication begins and the cells complete division C + D minutes later. An example of the construction of chromosome replication and division patterns for a case in which I equals 70 min, C equals 40 min, and D equals 20 min is shown in Fig. 1. The numerical values for I, C, and D were chosen arbitrarily to simplify the analysis, but these are realistic values for E. coli, as will be shown in subsequent sections. The construction begins at the far left of the figure with a hypothetical cell containing a single, nonreplicating chromosome. The sawtoothed line represents the time, I, required for this cell to progress through the steps necessary for initiation of chromosome replication. This is then followed by chromosome replication (solid line), and the cell divides D minutes after completion of chromosome replication (dotted line). Each division is indicated by a vertical line. The second (I + C + D) sequence is shown below the first sequence, with the cell beginning to prepare for the second initiation at the moment the first initiation takes place. The same sequence is then repeated a third time. Note that the sequences do not interact. The I + C + D rule simply states that once a cell initiates a round of chromosome replication, a new I + C + D sequence is inaugurated. Since the construction begins with a single hypothetical cell at the far left of the diagram, two cells are formed at the first division, four at the second division, and so on. Therefore, if the number of cells were plotted versus time, this construction results in an idealized, synchronously dividing population.

Fig. 1Construction of chromosome replication and cell division patters for E. coli with I = 70, C = 40,min, and D = 20 min. The construction is based on the I + C + D rule, beginning at the left (-130 min) with a hypothetical cell containing a single chromosome. P[reparation for chromosome replication requires I minutes (sawtoothed line), chromosome replication occupies C minutes (solid line), and the cell divides D minutes later (dotted line). At the time chromosome replications begins (-60 min), preparation for the next I + C + D sequence commences below the first. The long vertical lines indicate divisions. The chromosome configurations are shown across the top as linear structures. A filled circle at the left end of a chromosome indicates initiation of replication. The filled circle, corresponding to a replication fork, then proceeds along the chromosome. The configurations between 0 and 70 min, i.e., between the first two division, indicate the chromosome replication patterns during the division cycle of cells growing under these conditions. Only one cell is followed through the division cycle. The number of cells formed at successive division would be two, four, etc. The amount of DNA per cell is shown at the top in terms of the number of chromosome equivalents of DNA (G), where 1 G is the amount of DNA in one nonreplicating chromosome. The rate of chromosome replication during the division cycle is shown in the lower portion of the figure in terms of the number of replication forks per cell. Since replication is actually bidirectional with two forks proceeding in opposite directions, the values are twice the number of replication forks shown in the linear chromosome configurations.

Source:

Designations for additional periods of the duplication process are shown above the second I + C + D sequence. B is the time between cell division and the initiation of chromosome replication. U is the time between initiation of chromosome replication and the initial appearance of visible cellular constriction during the process of division. T is the time between initiation of visible constriction of the cell envelope and cell separation. The significance of these intervals and their durations will be discussed in subsequent sections of this chapter.

From the construction in Fig. 1 the pattern of chromosome replication during the division cycle can be determined. The simplest manner to determine the pattern will be described, in order to facilitate similar constructions for other growth rates not presented in this chapter. The chromosome is a circular structure which is replicated bidirectionally from a unique origin, oriC, located at 84 min on the genetic map (chapter 99). However, it is very difficult to draw these circular replicating structures, particularly at rapid growth rates. Since the goal of this presentation is to explain the means to easily determine cell cycle relationships at any growth rate, chromosomes will be shown as linear structures which replicate from left to right. In effect, one half of the actual chromosome will be considered. The chromosome replication and cell division patterns are shown along the top of the figure. These are obtained by following the construction from left to right and drawing the replicating chromosomes as indicated. At the extreme left (–130 min), the hypothetical cell contains a single chromosome. At the end of I (–60 min), replication of the chromosome is initiated as indicated by the filled circle at the left end of the chromosome. The replicating chromosome is then shown at 1/2 of the C period (–40 min). Upon termination of replication, two chromosomes are present and the cell divides D minutes later such that each daughter cell receives a single chromosome. Only one of the two daughter cells formed at each division is followed during each successive division cycle. All of the constructions to the left of the first division (at time 0) are totally hypothetical. After this first division, a true (albeit ideal) synchronous culture is developed as indicated by the cell outlines surrounding the chromosomes. Ten minutes after division, the cells have completed preparation for the next initiation event in the second duplication sequence, a new round of replication is initiated, and the cell divides again C + D (or U + T) minutes later. The relationship between chromosome replication and the division cycle is obtained by observing the chromosome replication pattern between the first and second (or second and third) divisions. In the example shown, chromosome replication initiates 10 min after division, the chromosome replicates during the next 40 min, and then there is a period devoid of DNA replication during the last 20 min of the division cycle. The amount of chromosomal DNA per cell, expressed as the number of chromosome equivalents of DNA (G), can be easily determined from the stick figure chromosomes as shown at the top.

DNA replication during the cycle can also be estimated from this construction in terms of replication forks per cell, as shown in the lower portion of the frame. The plot also indicates the rate of chromosome replication, if it is assumed that the rate of DNA chain elongation is constant during C (3). Although there may be some variation in chain elongation rate, particularly in slower growing cells (90) and in the terminus region (chapter 100), the I + C + D construction only defines the time for a round of replication and does not require specification of the rate of fork movement.

The important point to note from Fig. 1 is that the relationship between chromosome replication and the cell division cycle consists of a repeating series of linear sequences that overlap. The duplication of the bacterial cell is not cyclic because the duplication sequences do not repeat one after the other. On the other hand, the preparation for initiation of chromosome replication is continuous and cyclic since duplication can also be described as the cyclic achievement of the capacity to initiate chromosome replication, followed each time by cell division C + D minutes later. Thus, the synthetic events required for division begin before the previous division, and the interdivision time (τ) is determined by the interinitiation time (I).

Rapidly Growing Cells

The relationship between chromosome replication and cell division in cells growing at any rate can be determined by similar applications of the I + C + D rule. Figure 2 shows a second example, in which I equals 30 min, C equals 40 min, and D equals 20 min. The construction is performed exactly as shown in Fig. 1, beginning with a hypothetical cell containing a single chromosome. Again, the chromosome replication patterns and the rate of chromosome replication during the division cycle are shown at the top and bottom of the figure, respectively. Since I is less than C, a new round of replication begins before the previous round has terminated, resulting in the appearance of multiple levels of replication forks per chromosome. This situation obtains because each time cells have completed preparing for initiation, a new round of replication begins, irrespective of the status of the cell with respect to ongoing chromosome replication or division. There is a minimum interval allowable between replication forks (the eclipse period), but this will not be considered until later in this chapter.

Fig. 2Construction of chromosome replication and cell division patterns for I=30 mins,C=40 min, and D=20 min (see legend to Fig.1 for details).

Source:

There is one final important issue that distinguishes fast-growing from slow-growing cells. When τ is less than C + D, there is more than one chromosomal origin at the time of initiation of replication. The model described to this point assumes that at the time of initiation, replication starts at all origins essentially simultaneously. Based on the distribution of initiation in the cell cycle (30, 32, 58, 61) and on studies with flow cytometry (reviewed in reference 143), this is a fair assumption for most cells. However, it is not always the case. Some mutations, such as dam (4, 10, 99), dnaA (98, 143), and seqA (101), result in asynchronous initiation.

I AND B PERIODS

The remainder of this chapter surveys measurements of the timing of the division cycle periods described schematically in Fig. 1 and 2, and the kinetics of biosynthetic processes during these periods. Table 1 shows a list of measured values for the various cell cycle periods for E. coli strains B/r and K-12. The B/r strain was subdivided into the A, K, and F substrains when it was observed that stocks maintained in different collections possessed distinctly different cell division properties (65, 161). Since inclusion of every reported value for each period is beyond the scope of a useful table, the listed values show the results of a majority of representative measurements performed in different laboratories. Values for I are not given because it has been infrequently measured and, as indicated in Fig. 1 and 2, during balanced growth the interdivision time (τ) is equal to, and determined by, the interinitiation time. It should be noted that unlike the ideal cultures represented in Fig. 1 and 2, individual cells in a growing population possess variances in the lengths of the periods of the division cycle (see chapter 103). It is important to consider the extents and distributions of these variances when evaluating the significance of measured parameter values. The numerical values for the periods will differ depending upon whether they represent averages over an entire batch culture, averages of individual cells in a culture, or averages from synchronous cultures (11, 61). These differences, which are likely to be 10% or less, are not specifically indicated in the data since the accuracy with which cell cycle period measurements can be performed does not warrant such distinctions.

Table 1Values for cell cycle periods in E. coli growing at 37°C

Temporal Expression of Genes Involved in Chromosome Replication

The biosynthetic processes which take place during I, that is, the requirements for initiation of replication, are the subject of chapter 99 and will not be discussed in detail here except as regards the timing of synthesis of some components. RNA and protein synthesis are required throughout most of the I period under most growth conditions (14, 60, 66, 96, 97, 106), as expected for synthesis of the gene products that are involved in the formation of the complex of components which initiate DNA polymerization at oriC. The question is: are these required components synthesized continuously or periodically in the division cycle? Presumably, initiation timing could depend on the temporal regulation of formation of components of the initiation complex.

There have been several reports suggesting that certain genes in E. coli are expressed periodically with respect to either chromosome replication or the cell division cycle. Especially notable among these are genes that encode products involved in an aspect of chromosome replication (19, 121, 147, 148, 151). One of these is the dnaA gene. As described in chapter 99, the product of the dnaA gene acts early in initiation by binding to four sites (DnaA boxes) in oriC to participate in the opening of the origin to start replication. In fact, the timing of initiation of replication during steady-state growth may be determined by the binding of a fixed quantity of active DnaA protein to oriC (98). Thus, the timing of formation of the protein could conceivably participate in initiation timing. Expression of the dnaA gene has been found to fluctuate in the cell cycle (19, 121, 151). This fluctuation is due to an inhibition of transcription for an interval immediately after the gene replicates, which can last up to one-third of the interdivision time.

The inhibition of dnaA transcription after the gene replicates is due to the presence of a GATC sequence within the dnaA promoter. This is the recognition sequence for dam methyltransferase which methylates the N ⁶ position of adenine. It has been found that newly replicated, hemimethylated DNA binds to the cell membrane. Ogden et al. (122) were first to show this binding with oriC DNA. Originally it was thought that oriC bound to outer membrane, but it now appears that it binds to a unique membrane fraction (22), mediated through a protein, SeqA, which binds hemimethylated DNA (101). Campbell and Kleckner (19) subsequently reported that the time course of the period of hemimethylation varied for different genes, from about 1.5 to 12 min at 37°C. During this interval the genes are presumed to be sequestered in the membrane and unavailable for transcription.

Transcription of the gidA gene, located immediately to the left of the origin of chromosome replication, displays an oscillatory pattern identical to the dnaA gene (121, 151). The mioC gene, adjacent to oriC but on the right side, also exhibits a strong periodicity in expression, but with considerably different kinetics. mioC transcription is shut off prior to initiation of chromosome replication and apparently remains off for several minutes after initiation of replication (121, 151). The shutoff of the mioC gene is probably due to DnaA protein binding upstream of its promoter, and this repression of transcription may be necessary for initiation of replication. Transcription of the genes encoding ribonucleotide reductase has also been shown to fluctuate in the cycle, with a maximum in expression coincident with initiation of replication (147, 148). Finally, there are a few promoters within oriC that might also be involved in initiation (chapter 99). These transcripts could show periodicities in the cycle, but information on their expression has yet to be reported.

It is conceivable that these transcriptional periodicities play roles in regulating initiation of replication such that genes required for replication are maximally expressed just prior to replication, and then shut down after initiation as part of the mechanism to prevent premature reinitiation of replication. Conversely, genes whose transcripts might prevent initiation, such as mioC, might need to be shut down for replication to start at oriC. As regards formation of the DnaA protein itself, it is likely that there would be a decrease in its rate of synthesis during inhibition of dnaA transcription. However, this could be a small change in rate, and in fact, no dramatic change in DnaA formation was detected in synchronized cells (139). Furthermore, no dramatic cell cycle-dependent variations in synthesis of individual proteins in general have been detected (40, 105). Thus, the general conclusion is that most proteins required for initiation of replication are produced continuously with only minor changes in their rates of synthesis.

Interinitiation Intervals

One of the earliest findings with regard to I was the apparent relationship between chromosome replication and cell mass in Salmonella typhimurium and E. coli (56, 106, 141). It was suggested that the mean cell size per chromosomal origin at initiation of replication is constant (39) and that initiation takes place when this critical size is reached (39, 59, 62, 131). Measurements of cell size, usually cell mass as assayed by absorbance at 450 or 460 nm, indicated that the average size per chromosomal origin at initiation was essentially the same in cells of a given strain growing with generation times between about 20 and 60 min at 37°C, but that it may differ in more slowly growing cells (26, 60, 129) or after a period of protein synthesis inhibition (51). Recently, variations in cell mass at initiation have been reported based on measurements by flow cytometry (160a), but again there has been little change in cell size per oriC detected at initiation for cells growing with doubling times in the 20- to 60-min range. Furthermore, there is a positive correlation between cell size and initiation in a given population, and initiation takes place within a distribution of cell sizes which is decidedly narrower than the distribution of cell ages at initiation (12, 84, 87, 88, 90, 117). This fixed relationship between cell size and initiation, when it exists, is likely a reflection of the specific macromolecular synthesis that fixes the length of I and could be indicative of a constant active DnaA protein concentration at initiation.

The information in Table 1 deals with cell cycle parameters in normal cells in which chromosome replication initiates from oriC. Chromosome replication can also take place, and cells can divide, when sites other than oriC are employed as origins. In one example, the requirement for oriC can be circumvented by mutations in rnh which eliminate RNase activity and lead to initiation at alternative sites designated oriK (86, 157). As a second example, chromosome replication can also be driven from integrated plasmid replication origins when oriC is nonfunctional (6, 7, 100, 120, 153, 158). C and D appear longer in some plasmid F-controlled replication (153) and in rnh (156) mutants. When plasmid R1 was integrated in oriC, and replaced it as an origin, division seemed fairly normal, as did DNA content, consistent with little change in cell cycle parameters (7). Finally, C is also longer when DnaA concentration is increased above normal levels in cells replicating from oriC, due in part to slow replication close to the origin (3, 98).

The minimum I is the time between two successive initiations when all positive-acting macromolecules which could normally be rate-limiting for initiation are present in excess, and all trans-acting negative elements are disengaged. Theoretically, this minimum interval might establish the minimum I, and possibly the minimum τ, in a given strain. Studies on this subject generally involve analyses of mutants which are temperature sensitive for initiation at oriC, especially dnaA and dnaC mutants, which produce gene products that are nonfunctional at nonpermissive temperature. When such mutants are shifted from permissive to nonpermissive temperature, initiation of chromosome replication ceases but potential for initiation accumulates in the cells (20, 46, 47, 57, 64, 125, 152). When employing mutant alleles in which the activity of the gene product is thermoreversible, this potential can be expressed by shifting to permissive temperature, resulting in multiple rounds of reinitiation of chromosome replication. The interval between the first and second initiations is approximately 25 min at 30°C and 30 min at 25°C (46, 47, 64). Similarly, when potential is accumulated during thymine starvation of a thy mutant, the interval between subsequent initiations is about 12 min at 37°C (165).

This is the minimal interval between successive initiations, also referred to as the "eclipse" period. Part of this interval is accounted for by the time required for the newly synthesized, hemimethylated oriC to become fully methylated since it is buried in the membrane and inert for initiation while in that state. As expected, overexpression of dam methyltransferase decreased the interval (109). However, the period of hemimethylated oriC only accounts for about half of the eclipse. The explanation for the remainder is unknown, although it has been suggested that it is the time needed to shut down mioC transcription (151). In any case, this eclipse period, when oriC cannot function, probably prevents premature reinitiation of replication when DnaA activity may be very high in the vicinity of oriC (10, 94, 110). It should be noted that oriC reportedly also binds membrane before initiation and that this association is involved positively in the initiation event (50, 80, 81).

At the opposite extreme, when the interdivision time is long, i.e., greater than C + D, a gap in DNA synthesis exists between cell division and initiation of chromosome replication, which is designated B. As seen from Table 1, the B period can occupy a very large fraction of the division cycle in very slowly growing cells. The values given are those in the referenced reports, but B can also be obtained for the other growth rates listed by subtracting (C + D) from τ. All three substrains of B/r possess B periods during slow growth, although the length of the period during moderately slow growth (T < 150 min) is often less in strain B/r A than in substrains B/r K and B/r F. The considerable variation in the measured values for B in substrain B/r A is probably a real effect of different growth conditions, i.e., glucose-limited chemostat versus exponential or synchronous growth. The B period, however, does not have a unique physiological significance since it is simply a variable portion of the I period designated by I – (C + D). The obvious similarity between the B period in the cell cycle of slowly growing bacteria and the G1 phase found in most eukaryotic cells has led to a number of theoretical and experimental comparisons of slow-growing bacteria and animal cells (29). As with the B period in bacteria, the entire G1 phase in eukaryotes is not an essential part of the mitotic cycle, and a portion of it, at least, simply represents the growth period required between successive initiations of chromosomal DNA replication (the S phase).

C AND D PERIODS

Durations of C and D

The time for a round of chromosome replication (the C period) depends upon the growth medium and the growth temperature. Table 1 shows a summary of a number of measured values of C in cultures growing at a variety of rates at 37°C. In E. coli B/r A, B/r F, and B/r K, there is little variation in the length of the C period in cells growing with doubling times between about 20 and 60 min at 37°C, with an average value being approximately 42 min. It can also be seen that C increases in cells growing under steady-state conditions in batch cultures with longer generation times. This may not be the case under non-steady-state conditions such as in cells grown in chemostats. As a means to compare the change in C as a function of growth rate for E. coli B/r A, B/r K, and B/r F, the values of C for the three substrains reported in the references listed in Table 1 are plotted in Fig. 3. Although the data are from experiments in which C was measured by a variety of techniques in batch-grown cultures, the consensus shape of the curve showing C as a function of growth rate is clearly evident. There is little change in the C period for both strains as the growth rate decreases from 2.5 to 1.0 doublings per h, and then there is a gradual and continuous increase in the length of C. The data in Fig. 3 indicate that there is little detectable difference between the lengths of the C periods in the three strains.

Fig. 3Duration of the C period in E. coli B/r A, B/r K and B/r F as a function of growth rate at 37°C.The values for C were reported in the references listed in Table 1.

Source:

There is more variability in C for the K-12 strains listed in the table. This is because most entries are for different K-12 strains, or derivatives of strains, as indicated. Each of these could be very different as regards genetic and physiological make-up. This is in contrast to the listings for B/r, in which the same substrain was used for every study. However, if an individual K-12 strain were considered, it is likely that the constancy pattern of C with growth rate would be similar to that for the B/r substrains (see, e.g., AB1157). The actual values, however, could be different, especially in derivatives that have been subjected to numerous exposures to mutagens (chapter 133). Although there are a few measurements indicating a constancy of C during slower growth, based on gene frequency determinations, the accuracy of such methods for slow-growth studies has been questioned (24), and the bulk of the evidence indicates that C increases gradually as the doubling time is increased above 60 min at 37°C.

The table also shows that the D period does not change significantly in a given strain growing with doubling times between about 20 and 60 min at 37°C, similar to the constancy of the C period. The duration of the D period is partly determined by the time required for cell invagination, and the apparent constancy could be due to a compensatory relationship between cell diameter and the rate of invagination at different growth rates. In strain B/r A the D period is 22 to 25 min long over this range of growth rates, whereas in other strains it can vary between about 15 and 40 min. During slower growth, the D period increases, but again to a greater extent in some strains than others. Some of this variation can be accounted for by differences in techniques used to perform the measurements. For instance, division in the presence of inhibitors of DNA, RNA, or protein synthesis has sometimes been used. The possibility exists that in some strains and at some growth rates, protein synthesis might be required during the early stages of D for division to ensue (12, 161). Nevertheless, the majority of the division process in D can be completed in the absence of protein synthesis.

Equations Relating Cell Cycle Parameters to C and D

From values of C, D, and τ, it is possible to calculate reasonable values for various cell cycle parameters in a growing culture, and a number of equations have been derived for this purpose.

The average chromosomal DNA content per cell in an exponentially growing culture is given by:

The average cell age, in fractions of the division cycle, at replication of a specific gene on the chromosome is given by: a_x = (n + 1)τ – [(1 – x)C + D], where n is the smallest integer so that (n + 1)τ ≥ [(1 – x)C + D] and x is the fraction of the C period at which the gene replicates.

Consequently, average cell age at initiation of replication is: a_i = (n + 1)τ – (C + D), and average cell age at termination of replication is: a_t = (n + 1)τ – D.

The mean number of copies of a gene per cell in an exponential-phase culture is: X = 2^{[(1 – x )C + D]/τ}.

Consequently, the mean number of origins per cell is: X_O = 2^{(C + D )/τ}, and the mean number of termini per cell is: X_T = 2^{D /τ}.

Highly accurate calculations depend upon information on the variability in the measured values (25), but quit reasonable calculations can be made by simply employing the average values given in Table 1 or Fig. 3. The duration of the D period varies considerably among individual cells in a culture, and cell cycle variability in general, at least when τ is less than (C + D), is a consequence of this distribution in the length of D (11, 12, 13). In slow-growing cultures, τ > (C + D), it has been shown that the B period is also broadly distributed (87). Thus, the D and B periods of the cell cycle may be the most variable phases, although they are not independent and have compensatory variations (118).

U AND T PERIODS

The next issue to consider is the division process and its relationship to chromosome replication. It is often convenient to describe division timing, and the relationship between chromosome replication and initiation of division, with respect to the timing of visible cell invagination. However, the steps in the division process appear to actually start long before the visible binary fission of the cell begins. Since U is the time between initiation of chromosome replication and initiation of visible cell constriction, and T is the time between initiation of visible constriction and cell separation, then C + D = U + T by definition. T is reasonably constant at 10 to 12 min over the same range of growth rates in which D is also constant (Table 1), possibly for the same reason. T increases in length with increasing generation time, but to different extents in different strains. It has generally been observed that the duration of the T period is less than, or at most equal to, the duration of the D period. As shown in the table, however, there are a few instances in which T appears to be somewhat longer than D. The significance of the relationship between T and D will be considered later when the coupling between chromosome replication and cell division is discussed.

Kinetics of Cell Growth

The kinetics of cell growth, particularly in relation to U and T, require consideration of changes in cell dimensions during growth (i.e., volume, surface area, length, and diameter) and the rate of total macromolecular syntheses (i.e., total cell mass, cytoplasm, membrane, and peptidoglycan). Many early analyses of growth focused on measurements of the kinetics of cell elongation during the division cycle, since the diameter of the rod-shaped E. coli cell changes little if at all during growth (31, 155), but more recent work has centered on the kinetics of growth of total cell mass. Findings and proposals on this topic can generally be fit into two categories (see references 30 and 116). The first possibility is a constant rate of increase, with a doubling in rate at a specific time in the division cycle (a bilinear pattern). Indeed, models for cell cycle control have often been based on the idea that sites of cellular growth might double at a specific time in the division cycle (33, 42, 45, 53, 92, 126, 130, 136, 167, 168). The second possibility is an exponential increase during the cell cycle. There now appears to be a consensus that the cytoplasm and/or total mass of the cell increases exponentially during the division cycle (30, 31, 83, 116). This may also be the case with cell volume during the division cycle, and in fact, there has been a recent suggestion that volume growth responds directly to cytoplasmic growth and thus would be indistinguishable from exponential (48). Length growth could well have a more complex pattern due to the interruption of side wall growth by pole formation during invagination.

In principle, the kinetics of synthesis of individual components of the cell should be easier to determine than the kinetics of growth of the entire structure, since rates of synthesis can be measured directly, thereby permitting accurate information on synthetic patterns. Measurements of the rate of protein synthesis during the cycle, based on uptake of a radioactive amino acid, have clearly shown that it increases exponentially (30, 48). Similar approaches have been used to investigate cell cycle timing of the synthesis of the major components of the cell envelope: peptidoglycan and membrane. The peptidoglycan sacculus of E. coli, consisting of a single network of long glycan chains and short peptide chains covalently linked together, completely encloses the entire cell between the cytoplasmic membrane and the outer membrane (chapter 6). Synthesis of this covalently closed structure during the division cycle is particularly intriguing because it involves the breakage and reformation of the covalent bonds between the polysaccharide units. Based on measurements of the incorporation of radioactive peptidoglycan precursors (usually diaminopimelic acid), the rate of synthesis during side wall formation increases continuously, probably exponentially, in direct proportion to protein synthesis (48, 69, 111, 124, 160). Although there had been some early evidence for zonal growth of peptidoglycan in the cell surface (138, 142), it now is evident that it takes place diffusely at multiple sites over the cell (15, 16, 17, 36, 85, 115, 156, 160, 162). When invagination begins, there is increased synthesis at this position and decreased side wall synthesis (31, 48, 160, 162), with the main incorporation at the leading edge (160).

It is difficult to precisely define the pattern of membrane synthesis since there has been some disparity in results, probably due to the use of synchronously dividing cultures for many of the experiments and to the well-known disturbances of growth properties which can take place upon synchronization. Furthermore, it is often difficult to distinguish between a stimulation in synthesis at a specific time in the cycle (e.g., a doubling in rate of synthesis) versus a repression in synthesis at an earlier time. The general finding is that membrane phospholipid synthesis is higher toward the end of the cycle (21, 27, 48, 55, 77, 78, 123, 127, 129). However, no consistent relationship has been found between rates of membrane synthesis and initiation of I, U, or T periods, although there could be an increase at initiation of invagination. It may be that membrane synthesis and peptidoglycan synthesis go hand-in-hand, both in response to cytoplasmic growth (31, 48). Similar considerations apply to the synthesis of proteins in the envelope of E. coli. A number of envelope proteins, particularly from the outer membrane, have been reported to increase in rate of synthesis (or activity) at discrete times in the division cycle (9, 27, 55, 129). Again, however, no consistent relationships have been established with either initiation of chromosome replication or the initial stages of cell division.

Initiation of Cell Division

Several years ago, Rothfield and coworkers (28, 107, 137) proposed the existence of zones of envelope differentiation which could be involved in division timing and placement. The zones, called periseptal annuli, were reported to consist of rings of adhesion of outer membrane, peptidoglycan, and inner membrane which flank the developing constriction site during the process of division. These adhesion zones separate the division site from the rest of the cell. It was further proposed that the annuli form well before division actually starts, such that a newborn cell already contains the annuli centrally located in the cell to be used in the next division. New annuli appear to form at the site of existing annuli and move toward the cell quarters as they develop. Thus, a cell late in the cycle has one set bracketing the ongoing constriction, and two new annuli at the centers of the nascent daughter cells. This continuous formation of the zones is consistent with the finding that membrane fractions enriched for membrane-peptidoglycan adhesion zones are formed continuously in the cell cycle (79). If the model proves correct, then initiation of annuli synthesis could represent the first stage of initiation of division, long before the beginning of the observable T period. Furthermore, initiation of chromosome replication and initiation of annuli synthesis could be associated, and this association could represent the biosynthetic definition of U + T.

The preceding idea was based on the appearance and organization of spaces or "bays" between the cytoplasmic membrane and the cell wall during hyperosmotic shock. Whereas Rothfield and coworkers consistently found an organized pattern to these plasmolysis bays (28, 28a, 107, 137), in agreement with their model, Mulder and Woldringh (114) reported random localization of the bays, which is not consistent with the model. Woldringh (160b) has proposed an alternative model based solely on the physical properties of the cell envelope to explain the positioning of plasmolysis bays. Additional work will be needed to resolve these differences and validate the involvement of these structures in an early stage of constriction-site localization.

Another identifiable event in the bacterial cell cycle is the initiation of cell invagination during the division process. Shortly after completion of a round of chromosome replication, invagination begins in midcell, leading to cell fission about 20 min later. Among the numerous gene products required for the division process (41, 44, 108), the product of the ftsZ gene is essential for division, acts early in the initiation of division (5, 34, 72, 102), and is found localized in a ring around the cell at the constriction site (8). It remains at the leading edge of the constriction and then dissociates when division is completed. Initiation of cell division could be set by the periodic formation of the FtsZ ring, and it is suggested that the level of FtsZ dictates the frequency of cell division (103, 104).

If FtsZ ring formation is critical for division, then it is of considerable interest to inquire as to the mechanism that might control the timing of its formation. It is conceivable that temporal expression of the gene plays a role in this periodicity of FtsZ action. The ftsZ gene resides within a contiguous group of cell division genes and is transcribed from several promoters located in upstream genes (1, 37, 134, 135, 159, 164). A number of laboratories have examined cell cycle expression of the gene. Recently, Garrido et al. (49) and Zhou and Helmstetter (169) examined total ftsZ expression in the cell cycle and found a two- to threefold fluctuation in transcript levels. Although maximal expression occurred near the time of initiation of replication, the periodicity is most likely due to inhibition of transcription at the time of gene replication, similar to the dnaA and gidA genes, and not to a coupling of transcription to initiation of chromosome replication. Earlier studies on ftsZ transcription employing transcriptional fusions to lacZ reached conflicting conclusions. It was reported that transcription was restricted to the time of cell division (38) or that it occurred throughout the cycle with a doubling in rate at the time of initiation of replication (133), but neither study measured total, intrinsic ftsZ transcription. In any case, the periodicity in transcript levels would not be expected to cause a major change in FtsZ protein in the cycle. It appears most likely that ring formation is due to timed self-assembly of the structure from preexisting molecules in the cell (104).

COUPLING BETWEEN REPLICATION AND DIVISION

The last issue to consider is the control of the timing of cell division. Chromosome replication and cell division must be coupled, because chromosomeless cells are rarely detected, but the question is: how are they coupled and what actually times cell division? One important observation in this regard concerns the behavior of cells with respect to cell division when DNA replication is inhibited. When replication is inhibited in the absence of induction of the SOS response, which blocks division via inhibition of FtsZ (chapter 89), cell division continues for D minutes and then it stops, at least temporarily. From these and related experiments (e.g., references 18, 52, and 62), it has been concluded that termination of chromosome replication is normally required for division. However, division resumes after its initial blockage during inhibition of chromosome replication in certain temperature-sensitive DNA replication mutants (67, 68, 73, 146, 149) and in mutants defective in SOS functions (70, 71, 74, 75). Studies on the kinetics of division during inhibition of DNA replication showed that cells with completed chromosomes divided, and then there was a delay in cell division while the cells continued to elongate. After this delay, chromosomeless cells of roughly normal size began forming from the DNA-free ends of the elongated cells. This continued division in the absence of chromosome replication leads to the inescapable conclusion that cell division can proceed in the absence of concurrent chromosome replication. It is thus likely that the normal coupling between replication and division is mediated by a negative control such that division in midcell is prevented until the chromosome has completed replication and at least begun to segregate from the cell center (41, 43, 66). This could be a consequence of a veto-type inhibitory effect which prevents division by a negative influence of the chromosomal mass on invagination. Indeed, it has been shown that peptidoglycan synthesis is lower in the nucleoid-containing portion of filaments (113). The release of the veto would coincide with the appearance of a DNA-free zone in the cell center as the chromosomes vacate the site of the developing invagination. In some slowly growing cells, invagination can begin before nucleoid separation, suggesting that the veto may be exerted subsequent to initiation of constriction (43, 66, 150). Furthermore, cell pairs with one member containing about two chromosomal DNA copies and the other containing a small amount of DNA are observed in some mutants (e.g., reference 119), suggesting a guillotine-like behavior of constriction when it is completed near the periphery of a nucleoid.

Based on the preceding, and on similar information that has accumulated over the past two decades, the processes leading to division and chromosome replication probably follow separate but interactive pathways, as suggested by Jones and Donachie (76; reviewed in references 41 and 140), such that the steps leading to cell division progress in parallel with chromosome replication. In a recent study on this topic, division was examined in cells in which a portion of the left end of oriC was replaced with a temperature-controllable R1 plasmid (7, 120). When timing of initiation was varied, relative to cell mass, the cells divided at normal size, lending strong support to the concept of parallel pathways. The pathways could begin simultaneously (e.g., at the end of I) or one might be an offshoot of the other, and they are coupled due to the veto effect of the unsegregated chromosomes. But the question remains: how is the division site localized in the cell? As described in chapter 101, the minB operon encodes topological specificity factors that locate division to a site in midcell—but what determines the placement of this site? One view is that it is determined by the envelope/division pathway exclusively, i.e., the pathway localizes the upcoming division site at midcell, and the only function of chromosome replication/segregation is to prevent its full activity until the chromosomes have segregated. The periseptal annuli could serve this site localization function, for instance. A second view is that a positive division signal directly couples envelope synthesis and division to DNA replication and segregation. This "nucleoid occlusion" model (112, 163) states that in addition to the inhibitory effect of the DNA mass on division, there is a positive effect of completion of chromosome replication that stimulates division between the chromosomes. In support of this idea, the localization of division during formation of chromosomeless cells in filaments appears to be influenced by the positions of the chromosomes. When they were replicating, division was nearby, whereas when they were not, division was more random in the DNA-free regions (112, 113). Thus, the chromosomeless cells formed in filaments are not necessarily of normal "newborn" size, as had been reported earlier (35, 68, 75, 149), but their existence clearly shows that completion of chromosome replication does not "trigger" division. The chromosomes may simply fine-tune division site placement.

If the dual-pathway concept is correct, then the timing of U and T relative to C and D, and evidence for positive correlations between cell sizes at initiation of C and T (see references 82 and 88), might depend on the extent of overlap of the pathways in a given strain or growth state. In some instances invagination could be inaugurated before the end of the C period, but completion of division would await termination of replication and segregation of the chromosome. On the other hand, if the U period were longer than the C period, the timing and variations in initiation of the T period would be functions solely of properties of the cell division pathway. This hypothesis could explain a number of interactions between replication and division which have been summarized in this chapter. For instance, a lengthy interval of protein synthesis is required for division to take place (about 40 min at 37°C; 128), and this interval could define U. Thus, a requirement for protein synthesis at the beginning of the D period for some strains to divide is anticipated when U is longer than C. This would not be the case when C is longer than, or equal to, U. Conversely, the finding that certain strains divide at a smaller-than-normal size after completion of chromosome replication in the absence of protein synthesis (51) would be anticipated if C were longer than U. In any case, the majority of steps in the division process are probably timed by their own pathway, rather than the chromosome pathway, with the exception that segregation of completed chromosomes is normally a prerequisite for division to be completed.