Chapter 105

Nucleoid Segregation

MOLLY B. SCHMID and ULRIK von FREIESLEBEN

INTRODUCTION

Bacterial nucleoids segregate with great fidelity into the daughter cells that arise during cell division. Less than 1 in 30,000 wild-type Escherichia coli cells lacks DNA, as seen under a microscope (46). Faithful nucleoid segregation, therefore, depends on a precise mechanism that ensures the equipartition of nucleoids among daughter cells as well as the coordination between DNA replication and cell division (for details, see chapters 99, 101, 102). The process involves a number of distinct steps, namely, termination of DNA replication, decatenation of the daughter chromosomes, and nucleoid separation, the spatial repositioning of daughter nucleoids at both sides of the cell’s midline. The first two of these steps are better understood than the last one, but, in general, the study of this fundamental cell cycle event has proven to be quite difficult and our state of knowledge has proven to be limited. The genes required for this process have not been fully identified, and the role of several known proteins, while plausible, has not been definitely established. There are several recent reviews on this topic (28, 29, 31, 43, 44, 65, 90, 95).

One of the reasons why this fundamental subject has proven to be so elusive is that mutants with lowered segregation fidelity are not necessarily lethal. Mutants that lack highly faithful segregation remain viable in artificial media, although presumably they are at a disadvantage under natural conditions. This suggests that, in bacteria, multiple systems contribute to the overall fidelity of nucleoid segregation. This is in stark contrast to the situation in eukaryotes, in which a clearly "dedicated system," the mitotic apparatus, ensures faithful distribution of daughter chromosomes at cell division. Whether bacteria possess more than one functionally dedicated system that work in combination to assure highly faithful nucleoid segregation or accomplish this using multifunctional proteins with roles in other cell cycle functions remains uncertain. There is precedent for functionally redundant mechanisms in the faithful segregation of large, low-copy plasmids, such as F, R, and P1. These plasmids have both a primary segregation system (sopA, sopB, and sopC in F) and a mechanism to prevent host cell division (ccd) until proper partition of the plasmids is accomplished (71, 74).

In this chapter we review the process of bacterial nucleoid segregation as it is known in E. coli and Salmonella typhimurium (official designation, Salmonella enterica serovar Typhimurium). We will not explore nucleoid segregation in other bacterial species such as Bacillus subtilis or Caulobacter crescentus, where developmental programs can lead to dimorphic nucleoids and nonidentical daughter cells, even though such systems provide good insight into the segregation process (69, 118); nor do we review the faithful segregation mechanisms of the low-copy-number plasmids of E. coli and S. typhimurium. Proteins and DNA sites necessary for segregation of the F factors, R factors, and P1 phage have been extensively studied and reviewed (44) and may indeed share components with the nucleoid segregation apparatus. Especially interesting (and regrettably not included here) are the ccd genes which inactivate gyrase action until the F factor is appropriately partitioned to the daughter nucleoids (71, 74).

Unlike in the eukaryotes, segregation of daughter DNA in bacteria is achieved without involving an obvious cellular ultrastructure such as a mitotic spindle. The first attempt to address how the process is carried out in bacteria was included in the classic postulation of the replicon model by Jacob et al. (55). They suggested that the rigid bacterial cell envelope serves as the functional analog of the eukaryotic spindle and that the attachment of the daughter nucleoids to this structure suffices to separate them. They postulated that new cell envelope material is laid down between the nucleoids (and segregates in a "semiconservative" fashion), which, upon closer examination, turns out not to be the case (see chapter 104). This is a "passive" segregation model because it does not involve any dedicated proteins that can generate a mechanical force.

A variety of methods have been employed in the search for answers to the following fundamental questions about nucleoid segregation in E. coli and S. typhimurium. Which genes are required? What are the functions of the relevant gene products? How do these gene products interact? What are the steps of the nucleoid segregation process? How is coordination with DNA replication and cell division achieved? The analysis of nucleoid segregation has required techniques of cell biology—microscopy, cell sorting, and specific genetic selections and screens—in addition to the general tools of genetics, biochemistry, and molecular biology.

MICROSCOPIC OBSERVATIONS

Light Microscopy

Nucleoids of living cells can be seen under a phase-contrast microscope, which allows observation and recording of the gross morphological changes that occur during growth (72). For visualization of the nucleoid, the refractive index of the medium must be altered to nearly match that of the cytoplasm, or a fluorescent DNA-binding dye, such as 4',6-diamidino-2-phenylindole (DAPI) or Hoechst no. 33342 (16), can be used. Fluorescent dyes are generally taken up after fixation (46) or permeabilization (97), but they can also be incorporated into living cells (116). Advances in fluorescent probe technology and digitized microscopy have allowed some greater resolution of intracellular compartments in bacteria (63, 118).

Time lapse photomicrographs of living E. coli cells (72) show that nucleoids change shape and separate without undergoing the cycles of condensation and decondensation characteristic of mitotic chromosomes in eukaryotes. Eventually, the sister nucleoids separate completely and move to more polar positions in the cell. Repositioning of the sister nucleoids takes 6 to 8 min of the 30-min generation time of fast-growing cells and occurs before the onset of septum formation. The timing of the microscopically visible events in nucleoid separation does not vary greatly from cell to cell and has a coefficient of variation of about 20% (96). A reason why the timing of nucleoid segregation may be quite precise is that all cells terminate nucleoid replication at nearly the same time, at least in synchronized cultures (13).

The actual temporal order of nucleoid separation, termination of replication, and formation of the septum depends on the growth rate of the cells. Under conditions of slow growth, cell constriction (septum initiation) often occurs before visible nucleoid separation, while under conditions of rapid growth, most cells reverse the order, with nucleoid segregation preceding cell constriction (113). Termination of replication cannot be visualized but in slowly growing cells is presumed to occur before the time of visible nucleoid separation and in rapidly growing cells is presumed to occur concurrent with nucleoid separation (113).

Electron Microscopy

The development of electron microscopic methods to study cellular structures and nucleoid shape in bacterial cells is reviewed in chapter 4. Fixation procedures can alter the cellular and nucleoid morphology (see chapter 4). A recent advance in fixation methodology relies on ultralow temperature rather than chemical cross-linking. The cryofixation freeze-substitution technique has resulted in images believed to closely resemble the unfixed cells (51). However, this point is still being discussed (6).

In thin sections of cryofixed bacteria, nucleoids do not appear as the compact body seen with conventional fixation but as a multilobate structure with projections extended over a considerable distance (see chapter 12). There is little in this morphology to directly suggest aspects of the segregation process. A few observations carried out with an electron microscope may, however, be relevant to nucleoid segregation. Although there is no cytologically observable spindle in wild-type E. coli or S. typhimurium cells, mutants that overproduce certain proteins (e.g., CafA [see below]) result in intracellular filaments (85). In addition, other species have filamentous structures that may prove to have functional homology with the eukaryotic spindle (8). The mesosome, a structure previously thought to be associated with the nucleoid, is considered to be a fixation artifact (51; see chapter 4).

Introduction of Macromolecules into Cells

Small molecules and small proteins can be introduced into viable cells (99). It has been possible to show, for example, that the histone-like protein HU, when added externally, is distributed throughout the ribosome-free nucleoid space, whereas non-DNA binding proteins localize in the ribosome-filled space (99). While this technique has not been widely exploited thus far, it may provide an important tool for the complementation of nucleoid segregation mutants by purified proteins.

CELL SORTING TECHNIQUES

Flow Cytometry

The distribution of DNA content in members of a population can be studied by flow cytometry, using a fluorescent DNA-specific dye, such as mithramycin (14). Single cells are counted and measured on the basis of both DNA content (fluorescence intensity) and cell mass (light scattering). This important new technique has proven to be valuable for assessing the heterogeneity of DNA content in populations of exponentially growing wild-type cultures (101, 102), as well as in the characterization of mutant strains (64, 110, 112). This technique has not yet been used to determine whether segregation of the nucleoids of multinucleated, fast-growing cells takes place with the same degree of synchrony as the initiation of their replication (101).

"Baby Machine"

A uniform population of newborn cells can be obtained from an exponentially growing culture by using membrane elution techniques (see chapter 102 for details). Such a baby machine relies on the adsorption of cells to a nitrocellulose membrane and the release of only one of the daughter cells formed upon division (40). Newborn cells are detached from the membrane and eluted into the medium flowing through the membrane. This method has been used in conjunction with a microscopic technique to determine the cellular polarity of the segregating nucleoids (see chapter 104). The results generally indicate that there is a preference toward nonrandom segregation toward one of the cell poles.

GENETIC SCREENS FOR SEGREGATION MUTANTS

Several methods have been used to identify E. coli and S. typhimurium mutants defective in nucleoid segregation. Two different cytological phenotypes have been used to determine deviations from faithful nucleoid segregation. One consists of cells with a centrally located nucleoid containing multiple genome equivalents, the so-called partitioning or par mutants (Fig. 1). The other phenotype leads to the production of DNA-less, anucleate cells. In addition, a genetic selection based on the resistance of polyploid cells of many species to camphor vapors (84) has been used to identify mutations in nonessential genes that cause increased DNA content.

Fig. 1Fluorescence photomicrographs of S.typhimurium partitioning-defective mutants stained with DAPI. (A) Heat-sensitive strain SE5171 carrying the parC171 mutation,grown under nonpermissive conditions.Two photographs allow comparisonof the phase-contrast image and the DAPI fluorescence image.(B and C) Photomicrographs taken while both the fluorescence and phase-contrast images provides increased photographic sensitivity. (B) Heat-sensitive strain SE7784 carrying the parC281 mutation, grown under nonpermissive condition. (C) Certain mutations, such as those causing partial inhibition of ftsZ, can cause increased cell size (107), which provides additional resolution. The figure shows strain SE5758, which carries parE206 and at least one additional suppressor mutation.

Source:

"Partitioning" Mutants

Mutants with nucleoid partitioning defects (par) have been found within the collections of conditional lethal mutants of E. coli (48, 49, 60, 91) and S. typhimurium (97), using microscopic screening of cells placed under nonpermissive conditions. These mutants are relatively rare members of the filamentous temperature-sensitive (fts) class that continue DNA replication but leave the newly replicated daughter nucleoids centrally located in filamentous cells (50). In some cases the mutant populations contain cells that are highly varied both in cell dimensions and in nucleoid morphology (97, 120).

The genes identified thus far by screening conditional lethal mutant collections do not encode proteins involved in the generation of mechanical force. However, the set of genes likely to be involved in nucleoid segregation is not yet complete. The initial search through the E. coli collection of temperature-sensitive mutants (49) did not identify the parC and parE mutants, which were found within the same collection 20 years after the initial screening (59, 60). Additional essential genes with a role in nucleoid segregation may still remain unidentified. Microscopic screening through a conditional lethal mutant collection of S. typhimurium has identified several additional par mutants whose mutations are other than the known parA to parF genes (67, 97; J. Fukayama, D. Sekula, and M. B. Schmid, unpublished data).

DNA-Less Cell Producers

Some conditional lethal mutants produce anucleate, DNA-less cells, indicating at a minimum a loss of coordination between nucleoid segregation and cell division (50, 103). The proportion of anucleate cells in a mutant population may be difficult to determine because, even in the absence of a mechanism that ensures equipartition, there may be physical or mechanical constraints to the localization of two nucleoids in one daughter cell.

One class of mutants (in the min genes) leads to the production of "minicells," small, spherical DNA-less cells (2). It would appear that, since nucleoid segregation proceeds otherwise normally in such mutants, these are not directly relevant to nucleoid segregation. As discussed below, this has turned out not to be the case.

Another class of mutants produces a high percentage of DNA-less normal-sized cells. An ingenious method to specifically identify nonlethal mutants that produce anucleate cells has been developed by Hiraga and coworkers (46). The starting strain does not express the lacZ gene because the high level of lac repressor represses the plasmid-encoded lacZ gene. In DNA-less cells the loss of the chromosomally encoded lac repressor allows expression of the plasmid lacZ gene. Thus, mutants that frequently segregate DNA-less cells have a blue or mixed colony color on 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside plates. As a control for the screening procedure, Hiraga showed that min mutants make blue colonies with this method. These colonies arose because ∼50% of the cell number in a minB colony are DNA-less, although they account for only ∼5% of the cell mass in the colony, since they are of smaller than average size. On the basis of this result, it was estimated that the screen could detect mutants that segregate as few as 5% of normal-sized DNA-less cells, which is the actual frequency observed with the mutants. This genetic method identifies nonlethal mutants that produce anucleate cells.

Polyploidy

Trun and Gottesman (108, 109) and Trun et al. (110) used resistance to camphor vapors to identify mutants that have lost the ability to maintain only one chromosome per cell. The rationale for this genetic selection remains elusive; however, polyploid mutants in several species have been identified by this procedure (84). Several E. coli genes that can mutate to give an Mbr (mothball resistant) phenotype have been identified. By several criteria, many of these mutant strains show increased ploidy. Whether these mbr mutations cause a primary defect in the nucleoid segregation process or in the regulation of replication initiation remains uncertain.

GENES INVOLVED IN SEGREGATION

Many genes with potential roles in nucleoid segregation are known through genetic analysis of mutants. However, since some of these mutants produce only a small percentage of DNA-less cells, it is uncertain whether these genes play a primary role in nucleoid segregation or merely relax the normally tight coupling between nucleoid segregation and cell division.

The first phase of nucleoid segregation, the resolution through decatenation and monomer formation, depends on genes with a known biochemical function, such as topoisomerases and resolvases. The remaining genes cannot be so definitively classified. Necessary to the understanding of nucleoid segregation will be the biochemical characterization of these gene products. The genes identified in this section remain contenders for playing functional roles in nucleoid segregation.

DNA Replication Genes

Several conditional lethal DNA replication mutants, dnaA (50, 80), dnaB (54), dnaG (39), dnaK (16), dnaX (80), and dnaZ (117), accumulate normal-sized DNA-less cells under nonpermissive growth conditions. DNA-less cells are thought to arise in these mutants because they undergo cell division without nucleoid segregation. The DNA-less cells are not the result of degradation of the DNA (52). This uncoupling between cell division and nucleoid segregation takes place despite the existence of several mechanisms that normally inhibit septation when DNA replication stops (26). The existence of these multiple systems for inhibiting septation makes the appearance of DNA-less cells in certain conditional dna(Ts) mutants perplexing. Most likely, DNA-less cells result from the release of cell division inhibition. However, the case of dnaG suggests that, in certain genes, only specific alleles will have nucleoid segregation defects. Two mutant alleles of dnaG lie within a small region in the C-terminal domain of DnaG and have a partitioning-defective phenotype (39).

Cell Division Genes

Mutations in essential cell division proteins have been isolated as temperature-sensitive (fts) mutants that filament at the nonpermissive temperature. Most of these mutants have well-spaced nucleoids and no apparent nucleoid segregation defect. However, mutants in two distinct fts loci, ftsB and ftsC, display slightly disturbed nucleoid segregation (88), producing a small percentage of DNA-less cells. Further work showed that these two fts mutants have replication-related defects. The ftsB mutant is an allele of nrdB, which encodes the B₁ subunit of ribonucleotide reductase which converts ribonucleotides to deoxyribonucleotides (106). Mutants in the second subunit of ribonucleotide reductase, encoded by nrdA, were originally isolated as dnaF mutants. Thus, the nucleoid segregation defect in ftsB mutants may be another case of a replication block causing abnormal septum formation. An ftsC mutant also segregates normal-sized DNA-less cells and has been shown to replicate DNA at a decreased rate.

Tetart et al. (107) proposed a model whereby nucleoid segregation occurs as the result of a decrease in the levels of a protein directly involved in septum formation, FtsZ (see below). The amount of DNA/nucleoid appears to be increased in strains with decreased FtsZ expression. It is proposed that FtsZ may inhibit nucleoid segregation when free in the cytoplasm, before it acts to initiate septum formation. This conflicts with previous results suggesting no role for FtsZ in nucleoid segregation on the basis of results using a temperature-sensitive allele (23). However, FtsZ is a complex protein, and a specific missense allele may have a phenotype dissimilar to the lack of the protein altogether.

Mutations in the min genes (see chapter 101) cause the formation of DNA-less minicells. In addition, populations of these mutant strains include filamentous cells, 10 to 20% of which have abnormal nucleoid distributions, indicating loss of nucleoid segregation fidelity (4, 56). Certain minB mutant alleles produce DNA-less cells that are heterogeneous in size (56), whereas other min mutants lead to the production of normal-sized cells lacking any DNA (57). The MinD protein has homology with the plasmid-encoded SopA protein (27, 76), which is required for faithful partition of mini-F plasmids. In addition, loss of Min protein function causes loss of negative superhelicity, while overexpression of Min proteins causes increased negative superhelicity (79). This correlation between the level of Min proteins and DNA supercoiling may be related to the SopA-dependent changes in mini-F plasmid superhelicity that has been observed (10).

The divA, divB, and divC mutations of E. coli cause production of DNA-less cells but only in fts and dna mutants (49). These mutations by themselves cause no discernible phenotype. Mutants with analogous phenotypes were found in S. typhimurium (98, 103).

Muk

The muk mutants have been identified through Hiraga’s screen that identifies colonies with a high percentage of anucleate cells. The authors called these mutants muk, from "mukaku" or "anucleate" in Japanese. Five muk genes have been described and partially characterized. These are mukA (tolC), mukB, mukC, mukD (43, 44, 45, 46), and mukF (S. Hiraga, personal communication).

The MukB protein has been most extensively characterized. This 177-kDa protein has two globular domains connected by an α-helical coiled-coil region. The N-terminal globular region has an ATP-binding consensus motif and weak homology with rat dynamin. The C-terminal globular region has three putative zinc finger motifs that are hypothesized to bind DNA (81). MukB protein shows ATP/GTP binding activity in the presence of Zn²⁺ and adsorbs to a calf thymus DNA-cellulose column (44). Electron microscopy of purified MukB protein has shown that it has a structure similar to that of the heavy chains of eukaryotic myosin and kinesin (44). It has been suggested that MukB is a motor that drives the nucleoids along a cytoskeletal system of filaments spanning the length of the cell (44).

MukB null mutants form normal-sized cells, 5% of which are anucleate and, for reasons not known, are viable at a low temperature only. DNA replication and timing of initiation are normal in these mutants. F plasmids partition normally into both the nucleate and anucleate cells, indicating that mukB affects chromosomal partitioning only (35). Multicopy suppressors of the high-temperature lethality phenotype of the mukB106 mutation have been identified (msmA, msmB, and msmC) (119). Two of these also suppress anucleate cell formation (msmB and msmC). None of them could suppress both phenotypes in a mukB null mutant.

Mbr

Four unlinked mbr loci have been identified by the selection for polyploidy (108, 109, 110). The mbrA to mbrD mutants have an increased DNA/protein ratio and altered cytology, including both DNA-less cells and heterogeneous nucleoid distribution. These characteristics suggest that the wild-type mbr genes play a role in nucleoid segregation. Many of the mbr mutants showed conditional growth at a high temperature or on enriched medium, which allowed further characterization of the mutations in these strains. The mbrD mutation may be allelic with rpoB on the basis of genetic mapping. The mbrA mutant is not viable when coupled with a mutation in RNase H (rnh), suggesting that initiation from the alternate replication origins, oriK (which occurs in rnh mutants), causes inviability.

RecA

A direct role for recA in segregation has been suggested by Zyskind et al. (120), who found a high number of normal-sized anucleate cells in recA and recA recD double mutants. These anucleate cells were assumed to arise from a direct segregation defect, contrasting with conclusions of previous experiments (18, 19, 20, 21). However, Skarstad and Boye (100) have reported that the high number of anucleate cells in a recA recD double mutant may still be explained by selective degradation of individual chromosomes because recD mutants exhibit substantial exonuclease activity (89). Supporting the original interpretation of Zyskind et al. is the finding that a mutant in four genes (recA recD recJ xonA) lacking demonstrable exonuclease activity still produces anucleate cells (W. Wackernagel and J. Zyskind, personal communication).

HU

The major "histone-like" proteins in E. coli and S. typhimurium are HUa and HUb. Together with H-NS and other DNA binding proteins, these proteins bind DNA, affecting the ability of DNA to function as a substrate for transcription, replication, and recombination. A strain carrying mutations in both hupA and hupB, encoding the two HU subunit proteins, produces a high percentage of DNA-less cells and aberrant nucleoid structures (53). As suggested in a recent review by Løbner-Olesen and Kuempel (65), this points to the importance of nucleoid structure in proper segregation of daughter chromosomes, a point that deserves further attention.

The following three genes encode proteins that may form filamentous structures involved in nucleoid segregation.

FtsZ

FtsZ is essential for septation and cell division (for details, see chapter 101). FtsZ is uniformly distributed in the cytoplasm of nondividing cells but becomes concentrated in a ring-like structure at the site of septum formation just prior to cell division (9). FtsZ protein shows several similarities to eukaryotic tubulin: it binds GTP and GDP and hydrolyzes GTP to GDP (25, 77, 87). FtsZ contains a tubulin-like signature motif and assembles into filaments and tubules (15, 78). At the moment, FtsZ is the best candidate for a prokaryotic homolog of a cytoskeletal protein.

CafA

Overexpression of a gene, cafA, located at 71 min on the E. coli chromosome map causes formation of cytoplasmic axial filament (85). The amino acid sequence of the CafA protein has similarity to members of the eukaryotic myosin and kinesin family, suggesting a cytoskeletal nature. However, mutants defective in cafA do not show any phenotype.

FtsA

The appearance of axial filamentous structures has been reported when a mutant FtsA protein is overproduced (37). A structural homology between FtsA and actin has been suggested (12, 92). Mutants in FtsA do not show defects in segregation, although they have a clear defect in appropriate septum formation.

CELLULAR STRUCTURES INVOLVED IN SEGREGATION

Cell Membrane

The classical model for chromosome segregation of Jacob et al. (55) proposes that the daughter nucleoids attach to the cell membrane on both sides of the incipient septum. Subsequent growth of the membrane and elongation of the cell between these attachment sites would ensure proper segregation. Lamentably, the membrane is not synthesized in such a semiconservative mode. Rather, it has been shown that the membrane of E. coli is assembled by insertion of new material at many sites (115). However, this does not exclude a role of the membrane in chromosome segregation.

There is considerable evidence that the chromosome is attached to the membrane at the replicative origin, but probably for a portion of the cell cycle only. It has been shown that the replicative origin, oriC, is attached to the membrane in vivo prior to initiation (58). In vitro, oriC DNA binds specifically to an outer membrane fraction (83). This specific affinity is limited to newly made origin DNA and due to its unique state of methylation. For a substantial period after replication, newly made origin DNA is hemimethylated, that is, the new DNA strand has not had a chance to become methylated by the major methyltransferase of E. coli, the Dam enzyme. This enzyme methylates the adenine of the sequence GATC, which is found in unusual abundance in the origin of E. coli and other enteric bacteria (typically, 11 GATCs in 245 bp [121]). Binding of hemimethylated oriC DNA takes place mainly with a specific fraction that is separable on sucrose density gradients (22). Methylation of the origin has been shown to be significantly delayed when compared with the average time required to methylate GATCs elsewhere on the chromosome (17, 83). This delay has been ascribed to sequestration of oriC in the membrane. While this sequestration appears to be important for the proper timing of replication by preventing premature initiations, it could also serve as an initial step in segregation (for more details on the role of methylation in the timing of replication initiation, see chapter 99). Sequestration of oriC lasts about 10 min, which could be sufficient time to give the daughter chromosomes their sense of direction and ensure that each becomes destined to occupy one of the cell halves.

Attachment to the membrane is not limited to the replicative origin. Early studies have shown that there are several sites, perhaps situated randomly along the DNA, that are attached to the membrane. The number of these sites has been estimated to be between 20 and 80 per nucleoid (for a review, see reference 62). It has been proposed that this kind of membrane attachment may be longer lasting than that of the origin and that it may play a role in segregation by ensuring that each incipient daughter nucleoid remains in its own cell half until septum formation is completed (93).

Several problems arise, however, when trying to imply a role of oriC membrane attachment in segregation. E. coli dam mutants, which do not contain hemimethylated DNA and hence do not sequester oriC, do not show segregation defects such as anucleate cell formation (112). Also, if oriC acts as a centromere-like site, the presence of high numbers of minichromosomes (plasmids using oriC as their only origin of replication) would be expected to interfere with segregation of the chromosomal origin. This is not the case, and moreover the minichromosomes themselves segregate in a random fashion.

Do centromere-like loci exist on the E. coli chromosome? We do not know. Every attempt so far to identify such sites has failed. One approach has been to try to clone sequences from the chromosome that would stabilize otherwise unstable plasmid replicons such as partitionless mini-F (D. Lane, personal communication). Centromere-like sites are known to exist for unit-copy plasmids such as P, F, and R1 (5, 24, 44). The cis-acting DNA sites parS in P1, sopC in F, and parC in R1 are required for accurate plasmid partitioning. These sites are also capable of stabilizing minichromosomes if the plasmid-encoded partition proteins are provided in trans.

Because termination of replication is a key event in segregation, the terminus region is a good candidate for containing a centromere-like site. Strains with deletions that remove up to 330 kb from the terminus region exhibit the phenotype associated with the absence of the dif locus, but other partitioning errors are not apparent (41). Thus, aside from the dif site’s role in chromosome resolution (see below), no locus in the terminus region has been shown to be important for nucleoid segregation.

Periseptal Annuli

The periseptal annuli are future sites of septum formation that can be observed microscopically. These membrane regions extend around the cell and are found at the cell midpoint and at the cell poles. The periseptal annuli are believed to create a separate compartment of the periplasm at these regions (36). The periseptal annuli are observed in electron micrographs, as well as by phase-contrast microscopy, when plasmolyzed cells are used. Observing plasmolysis bays, the periseptal annuli were hypothesized to move to the 1/4 and 3/4 positions within the cell as the cell elongated. Thus, these structures move by an unknown mechanism to similar positions within the cell as the nucleoids do. Further evidence comes from ftsZ mutants. In ftsZ mutant strains, the periseptal annuli are randomly distributed (26), and in cells underexpressing FtsZ, the nucleoids fail to separate consistently (107). However, the specific location of plasmolysis bays, hence the existence of periseptal annuli, has been called into question (114).

Cell Poles

The cell poles are former sites of cell division; one of the poles is the site of the most recently completed cell division. As such, there may be remnants of a mechanoforce-generating push mechanism from a previous cell division or active participants in a mechanoforce-generating pull mechanism. There are unique periplasmic and cell membrane proteins in the cell poles of E. coli (33, 69). These proteins include the maltose-binding protein and chemoreceptor proteins (70) and the Era protein (3). Era protein is an essential GTP-binding protein that is localized at the cell poles and the cell midpoint and halfway between the poles and midpoint (1/4 and 3/4 positions [38]). The cellular localization makes it interesting; however, mutants with a nucleoid segregation defect have not been described. It is possible that finding proteins at the poles is due to previous localization at the septum from which the cell poles are derived and may thus be involved in septation rather than nucleoid segregation.

STEPS IN NUCLEOID SEGREGATION

After termination of replication, the two daughter chromosome face two problems in order to segregate: they must resolve dimeric chromosomes produced by homologous recombination between daughter chromosomes and must unlink the catenanes that result from replication of the DNA duplex.

Decatenation

To achieve physical separation after replication of the right-handed DNA duplex, the two circular daughter chromosomes must undergo complete topological unlinking. Four topoisomerases capable of aiding this topological unlinking are known in E. coli, allowing for the participation of several mechanisms. Much of the unlinking probably occurs during the elongation phase of replication, presumably accomplished by DNA gyrase. Any remaining interlocks must be resolved at or after the termination phase of replication. Intertwined completed duplex molecules require a type 2 DNA topoisomerase for resolution. E. coli and S. typhimurium have two type 2 topoisomerases, DNA gyrase (encoded by gyrA and gyrB) and topoisomerase IV (encoded by parC and parE). The unlinking of gapped molecules, on the other hand, can additionally be carried out by type 1 topoisomerases, either topoisomerase I (encoded by topA) or topoisomerase III (encoded by topB). The type 1 topoisomerase, topoisomerase III, is required to achieve resolution of replicated daughter plasmids in vitro (30, 42, 75).

Mutant alleles of the genes encoding the type 2 topoisomerase DNA gyrase, gyrA (parD) and gyrB (parA), were identified among the original par mutations (48, 49, 82, 91). The role for DNA gyrase as the enzyme responsible for decatenation was further strengthened by finding that isolated nucleoids from a gyrB mutant strain grown under nonpermissive conditions were doublets and that these doublet nucleoids could be resolved into singlets in vitro by the addition of purified DNA gyrase (105).

Recent data have cast doubt on the conclusion that DNA gyrase decatenates the newly replicated chromosomes. A new type 2 topoisomerase enzyme was discovered in a search through conditional lethal mutants (59, 60, 67, 97, 104). The enzyme, topoisomerase IV, is encoded by the linked genes parC and parE. The enzyme has been purified and has a strong decatenase activity (86). In vivo experiments demonstrated that small plasmids remain catenated after a shift to nonpermissive growth conditions in parC and parE mutants but not in gyrA and gyrB mutant strains (1). The catenated plasmids were shown to have right-handed interlinks, a critical feature resulting from the separation of daughter DNA strands during DNA replication. These experiments demonstrate that topoisomerase IV, not DNA gyrase, is the major replication decatenase in bacterial cells. However, they do not explain why gyrase mutants also have a cytological nucleoid segregation defect. Potentially, small plasmids are an imperfect model for nucleoid replication and segregation. The domain structure of the nucleoid might impose additional topological constraints not found in small plasmids. The mutants suggest that both gyrase and topoisomerase IV are required for complete resolution of daughter nuceloids. Two different unlinking reactions may be accomplished by gyrase and topoisomerase IV; both may be necessary for complete nucleoid decatenation. Alternatively, gyrase may play an active role in the segregation process after decatenation or may affect the expression of topoisomerase IV, causing the segregation defect as a secondary phenotype.

Resolution of Dimeric Chromosomes

A reciprocal recombination between two circular molecules results in a single dimeric molecule; if this process were neither prevented nor resolved, nucleoid segregation could not proceed. Since homologous recombination between sister chromosomes can create these dimers, a mechanism is necessary to ensure ultimate resolution to two monomeric circular molecules. This subject is described in more detail in chapter 100. The site-specific recombination accomplished by the XerCD proteins, the ArgR and PepA proteins, and the dif site is required for proper cell division and faithful segregation in strains that are capable of homologous recombination (11, 61, 73). Cultures of recombination-proficient strains that lack the dif site include filamentous cells, 4% DNA-less cells, and 10% of cells with abnormal nucleoid morphology (61). The interpretation is that only the abnormal cells have failed to properly resolve the replicated chromosome to monomers.

In addition to the site-specific recombination event, a region of high homologous recombination neighbors the terminus, potentially aiding in the separation of sister chromosomes (66).

Nucleoid Separation

Two independent experiments have shown that nucleoids do not separate if protein synthesis is inhibited (32, 47). These experiments are variously interpreted to show that the synthesis of a specific new protein is necessary for nucleoid separation or that protein synthesis per se is required. This hypothesis has been further modified to suggest a protein synthesis-mediated link between DNA and the membrane. Membrane proteins in the process of simultaneous transcription-translation-protein localization could provide a DNA-membrane bridge, serving a role in proper nucleoid separation (68).

There is still controversy about whether the daughter nucleoids separate abruptly (32, 47) or gradually (111). Microscopic observations have suggested that nucleoids move in a gradual manner during cell elongation (111). The distance between the edge of the nucleoid and the cell pole remained constant, supporting gradual movement of the nucleoids rather than abrupt movement. These experiments are consistent with a passive separation mechanism, in which attachment of the DNA to a fixed cellular location accomplishes nucleoid separation as the cell elongates. However, nucleoids that were prevented from segregating by inhibition of protein synthesis, upon release of this inhibition, segregate more rapidly than the cells divide (32, 47). These experiments support the notion of an active movement of nucleoids, which requires a machinery for accomplishing nucleoid separation. Begg and Donachie (7) suggested as an alternative that this rapid movement could be accomplished by releasing the nucleoids from a possible membrane attachment and subsequent repulsion between the daughter nucleoids. The small size of cells and fixation-dependent changes in nucleoid shape make these very difficult experiments. In addition, it should be kept in mind that inhibition of protein synthesis results in rearrangement of the nucleoid shape, leading even to the fusion of individual nucleoids within a cell (34, 94).

CONCLUSION

The best-understood steps in the process of nucleoid segregation have proven to be the early ones that are required for decatenation and dimer resolution. The subsequent steps of faithful nucleoid partitioning have proven to be more difficult to study, to the point that it has been difficult to postulate even moderately detailed models. A number of promising findings have lacked subsequent substantiation. For example, there is neither genetic nor physiological evidence to validate the early postulation that attachment of the nucleoid to the cell membrane plays a direct role in the segregation process. This simply reflects our lack of knowledge and does not by itself mean that the cell envelopes may not be involved.

Evidence for the involvement of structural proteins is beginning to accumulate and to suggest an active process for bacterial nucleoid segregation. The strongest evidence arises from the production of normal-sized anucleate cells in mutants defective in MukB, a protein with similarities to eukaryotic proteins involved in the movement of organelles, and from the biochemical and cytological characterization of the FtsZ protein. The identification of additional proteins involved in nucleoid segregation is necessary for further understanding of nucleoid segregation and its interrelationship with events of the bacterial cell cycle. The characterization of these proteins may come from genetic methods as well as biochemical studies. The further biochemical characterization of the identified proteins should provide additional information that will allow the formulation of specific molecular models of nucleoid segregation. A great deal of hard work is still required to provide a molecular understanding of the process by which bacteria segregate their genomes with such high fidelity.