Research Signpost 37/661 (2), Fort P.O. Trivandrum-695 023 Kerala, India

Escherichia coli and Bacillus subtilis: The Frontiers of Molecular Microbiology Revisited, 2012: 45-59 ISBN: 978-81-308-0492-7 Editors: Yoshito Sadaie and Kouji Matsumoto

3-2. Nucleoid organization and chromosome segregation Daisuke Shiomi1 and Hironori Niki 1,2

1

Microbial Genetics Laboratory, Genetic Strains Research Center, National Institute of Genetics 1111 Yata, Mishima, Shizuoka 411-8540, Japan; 2Department of Genetics, Graduate University for Advanced Studies, SOKENDAI,1111 Yata, Mishima, Shizuoka 411-8540, Japan

Abstract. All life, including bacteria, must transmit genetic information from generation to generation. To do so, chromosomes must be precisely replicated and segregated to daughter cells. Recent advances in cell biology and developments in sequencing techniques have revealed commonalities and differences in the mechanisms of chromosome replication and segregation between eukaryotes and prokaryotes. Here, we describe the bacterial cell cycle and the replication and segregation of bacterial chromosomes and extrachromosomal entities, such as plasmids.

1. Introduction Many entire bacterial genomes have been sequenced. While eukaryotic genomes are diploid and generally comprise several linear chromosomes, bacterial genomes, including those of Escherichia coli and Bacillus subtilis, generally comprise a circular chromosome. However, genome sequencing has revealed that there are some exceptions. For example, the genome of Streptomyces coelicolor forms a chromosome with a linear structure, and Vibrio species have two circular chromosomes. Correspondence/Reprint request: Dr. Hironori Niki, Department of Genetics, Graduate University for Advanced Studies, SOKENDAI,1111 Yata, Mishima, Shizuoka 411-8540, Japan. E-mail: [email protected]

46

Daisuke Shiomi & Hironori Niki

2. Cell cycle of bacterial cells (1) The cell cycle of bacterial cells is different from that of eukaryotic cells. It takes 40 minutes to completely replicate the E. coli chromosome (C phase), and it takes 20 minutes to complete cytokinesis after termination of chromosome replication (D phase). However, under the nutrient-rich conditions, E. coli cells divide every 20 minutes. This apparent contradiction in the progression of cell cycle is possible because multiple rounds of chromosome replication can be initiated before cytokinesis occurs”. Under these nutrient-rich conditions, the time between termination of chromosome replication and completion of cytokinesis, that is D phase, is equivalent to the generation time of the cell (Figure 1). Because chromosome replication reinitiate before cytokinesis, multiple replication folks are formed on chromosome (multi-fork replication). Thus, in the bacterial cell cycle, chromosome segregation occurs simultaneously with chromosome replication. This situation is completely different from that in eukaryotic cells; during the eukaryotic cell cycle S phase, i.e., DNA replication, and M phase, i.e., chromosome segregation, are clearly separated in time. Thus, in

Figure 1. Bacterial cell cycle. During the cell cycle, the C phase (40 min) from initiation of chromosome replication (I) to termination of replication (T) is shown in slanted lines and the D phase (20 min) required for cytokinesis is shown in black. At the end of D phase, the cell is divided into two daughter cells (S). Progression of replication in each phase is illustrated. Cell cycle of cells dividing every 60 minutes (A) and 20 minutes (B). In cells that divide every 20 min, multifork replication takes place.

Nucleoid and chromosome

47

bacteria, under nutrient-rich conditions, replicated regions of bacterial chromosomes are segregated before replication of the whole chromosome is terminated; ultimately, completely replicated chromosomes are fully segregated. In contrast, when bacterial cells are growing slowly (i.e., the generation time is above 60 minutes), there is a lag time between completion of cytokinesis and the initiation of chromosome replication. This lag time is called B phase. Under natural conditions, circumstances for optimal growth are often difficult to maintain; consequently, B phase is often the longest stage of the bacterial cell cycle. As expected from the fact that replication and segregation of chromosomes can occur simultaneously in the bacterial cell cycle, there is no checkpoint system, like that in eukaryotic cells, to prevent improper progression of cell cycle. Thus, cell elongation proceeds even in mutant E. coli strains that are defective in chromosome replication. Bacteria can benefit from the absence of a checkpoint system because, under optimal growth conditions, cells can divide very rapidly because replication and segregation of chromosome can occur simultaneously and, consequently, the number of cells can increase suddenly. This cell cycle system would be most suitable for those bacterial cells that are smaller and have a more simple chromosomal structure than eukaryotic cells.

3. Replication of chromosome The E. coli chromosome is circular and about 4.6 Mb (Figure 2). Replication of this chromosome initiates from a specific region called the origin (oriC). Replication forks progress bidirectionally from oriC, and these replication forks meet at a specific chromosomal region, which is opposite to oriC”. This region is required for termination of replication and is called ter. Ter sequences which terminate progressions of replication forks are scattered in the ter region (2). Tus protein, which binds to Ter sequences, inhibits the helicase activity of the DnaB protein; DnaB is positioned at the front of each replication fork, and the interaction between DnaB and Tus stops progression of a replication fork. The Ter sequences are asymmetric; consequently, there is a polarity in binding of Tus protein to a Ter sequence. Thus, Tus inhibits replication in a uni-directional manner. Therefore, the Ter sequences are scattered throughout ter to prevent replication forks from progressing past the ter site, which is opposite to oriC. Studies on the subcellular localization of E. coli proteins involved in chromosome replication have revealed that the replication of the chromosome occurs in a specific place in the cell (3). The replication machinery is localized at the middle of each cell; it captures template DNA and replicates

48

Daisuke Shiomi & Hironori Niki

Figure 2. Chromosome of E. coli and inhibition of replication fork. Replication forks progress bidirectionally from the origin of replication (oriC) of a circular chromosome in E. coli. The replication forks meet each other at a position 180˚ opposite to the oriC site. The ter sequences are scattered in this region to halt the progression of replication forks (A). The ter sequences are asymmetric and can inhibit progression of forks from only one direction (B).

the DNA. In other words, the replication machinery does not move on DNA; rather, when chromosomal DNA passes through the replication machinery, the DNA is replicated. Thus, the force generated when replicated DNA is released from the replication machinery would contribute to segregation of DNA to a daughter cell.

4. Resolution of chromosomes There is a specific region, called the dif site, for site-specific recombination; this site reside at a site opposite to oriC. Site-specific recombination at dif site occurs to resolve replicated homologous chromosomes (Figure 3). If an even number of recombination events occurs between homologous chromosomes, the chromosomes are resolved into independent chromosomes and no dimer forms. However, if an odd number of recombination events occurs, the replicated chromosomes cannot be resolved into independent chromosomes because they are linked as a single circular dimeric chromosome. The dif site plays an important role in

Nucleoid and chromosome

49

Figure 3. Resolution of a dimer and a catenane.

resolution of dimers to monomers by recognizing dimers (4). A complex consisting of XerC and XerD has been identified as a factor that recognizes and resolves dimeric chromosomes. In addition, the XerCD complex functions with FtsK; FtsK is a transmembrane protein that localizes to the midcell and serves as DNA translocator to resolve dimers efficiently. It is thought that the cytoplasmic domain of FtsK pulls DNA into a daughter cell. The cytoplasmic domain of FtsK is highly homologous in function, as well as in sequence, to SpoIIIE which translocates DNA into forespores in B. subtilis. When a circular chromosome is replicated, DNA forms chains. Such DNA is called a catenane. Resolution of a catenane requires a topoisomerase, which can catalyze the cutting and ligation of double-stranded DNA. In E. coli, topoisomerase IV is a major enzyme responsible for resolving catenanes after the replication. Activated topoisomerase IV, along with the XerCD complex and FtsK, is localized to the division site.

5. Nucleoid The E. coli chromosome is 4.6 Mb and 1 mm, and this chromosome must be packed into an E. coli cell, which has a length of 1~1.5 µm. Therefore, E. coli DNA is not randomly compacted; rather, it is condensed with a highly organized structure. DNA forms supercoiled structures every 10 to several

50

Daisuke Shiomi & Hironori Niki

hundreds kb, and hence dozens of supercoiled domains are formed throughout the E. coli chromosome. Consequently, the chromosome is highly condensed in the cell. It is thought that the borders of each supercoiled domain are not fixed, but rather, that they are flexible and dynamic. Biochemical studies examining the extent of condensation of isolated chromosome have revealed that RNA is involved in formation of the supercoiled domains (6). However, the molecular mechanisms underlying the formation of the supercoiled domains are unknown; for example, the role of RNA in this process is not known. In E. coli, DNA gyrase and topoisomerase I play a critical role in formation of the supercoiled domains. The supercoiled domain is further folded and more condensed and then the condensed chromosome is called nucleoid (Figure 4). Although there are nucleosome structures in archaea, nucleosome structures have not been found in eubacteria. However, a histone-like protein, called HU, is a highly basic protein that binds to various sites in the E. coli chromosome. HU protein is a heterodimer comprising HupA and HupB. It has been identified DNA-binding proteins such as IHF and H-NS as homologous proteins to HU protein. These proteins would not be involved directly in folding chromosomes; instead, they induce local structural changes, including bending of DNA at their binding sites.

Figure 4. Supercoiled domains in chromosome. Several tens to hundreds of supercoiled domains are formed in E. coli chromosomes. Thus, the chromosome, which has a length that is a thousand times of the cell length, is folded in the cell.

Nucleoid and chromosome

51

6. Factors for chromosome condensation During the process of nucleoid formation, DNA is folded to some extent by topoisomerase. However, other factors that condense the chromosome further are required. Such a factor has been shown to be required for condensation of chromosome during M phase in eukaryotes, and this factor functions independently of topoisomerase. This factor was identified in Xenopus egg extract as a complex that condenses sperm DNA into a chromosomal structure in M phase (7). This protein complex is now known as condensin. Condensin in eukaryotic cells is a large complex. The core complex consists of a heterodimer comprising SMC (Structure Maintenance of Chromosome) proteins. Accessory proteins called non-SMC proteins bind to SMC proteins. SMC proteins are conserved from yeast to human and play important roles in maintaining the structure of chromosome and in chromosome segregation. The amino acid sequences of SMC proteins are conserved. Proteins in the SMC family have common structural characteristics (Figure 5). The N-terminal domain is a globular domain that has ATP hydrolysis activity, and the C-terminal domain has DNA binding activity. There is a coiled-coil structure between the N-terminal and the C-terminal domains. It is originally thought that the coiled-coil domain is formed between molecules as in myosin. However, a fold can form at middle of the coiled-coil domain of SMC monomers, and thus a coiled-coil domain is formed in the molecule. This domain also contributes to dimer formation (8). The heterodimer forms a ring-like structure. DNA is inserted into the ring, and then DNA is condensed. Non-SMC proteins play a role in closing the ring of SMC proteins. In recent years, an SMC family of proteins has been identified in bacteria; these findings indicate that SMC proteins are highly conserved from

Figure 5. Model of bacterial condensin. MukB can fold at the coiled-coil domain. The domain also contributes to assembly of SMC proteins. The C -terminal globular region binds to the MukEF protein, which are non-SMC subunits.

52

Daisuke Shiomi & Hironori Niki

bacteria to human and that they have important molecular functions. However, SMC proteins have not been identified in the E. coli genome. Nevertheless, MukB protein was isolated as a protein involved in chromosome segregation in E. coli, and it was found to be structurally homologous to SMC proteins. The globular N-terminal domain of MukB has ATP hydrolysis activity, and the C-terminal domain has DNA binding activity. There is a coiled-coil domain between the N and C terminal domains. In addition, MukB requires two other proteins, MukE and MukF, for its biological functions. These characteristics are common to eukaryotic condensin. But, there is no significant sequence homology between the proteins of the MukB complex and those of the condensin complex. E. coli cells lacking mukB gene can grow at low temperature; however, anucleate cells are produced at high frequency because chromosomes are not segregated precisely. When temperature becomes high, viability of the mukB mutant decreases and the mutant cannot grow at 42˚C. Because cell division of this mutant is also abnormal, most cells are elongated. Although overall chromosome volume increases in mukB mutants because of replication, the chromosomes form a cluster that is distributed throughout the cell, probably because resolution and segregation of the replicated chromosomes are deficient. A mutation in topoisomerase I can suppress the abnormal phenotypes of the mukB mutant (11). Topoisomerase I mutations can suppress the mukB mutant phenotype because negative supercoiling activity in the cells increase because of the mutation in topoisomerase I; consequently, the chromosomes are condensed. This phenotypes indicates that MukB is involved in condensation of chromosome in E. coli. The phenotype of B. subtilis cells lacking the smc gene is comparable to that of E. coli cells lacking mukB (11). Based on these observations, it is thought that the MukB complex is functionally homologous to the SMC family and it is a bacterial condensin.

7. Spatial organization of chromosome Techniques in cell biology, particularly visualization of fusion proteins with green fluorescence protein (GFP) and fluorescence in site hybridization (FISH), have enabled the visualization of the spatial organization of specific regions of chromosomes in a cell. These techniques have revealed the characteristic locations of oriC, the replication origin, and the Ter domain, the termination site, during the cell cycle; these important loci localize to specific regions in a cell (Figure 6). Under conditions of slow growth in which multifork replication does not occur, chromosome replication occurs once per cell cycle. In such cells, oriC

Nucleoid and chromosome

53

Figure 6. Subcellular localization of ori and dif sites in chromosome segregation.

and the Ter domain localize to the midcell before replication is initiated. After replication starts at oriC at midcell, a region including the newly replicated oriC sites migrates to the cell poles. The rest of the chromosome is still being replicated. When the replication forks reach the replication terminus region, replication is terminated. The Ter domain is replicated at last at the midcell and the newly replicated Ter domains are segregated to daughter cells during cytokinesis. Thus, rapid migration of oriC to cell poles and segregation of the replication terminus region at midcell are characteristic subcellular localization patterns in E. coli (13). Similar localization patterns for the origin and the Ter domains have been seen in B. subtilis and Caulobacter crescentus (14,15). Such characteristic localization patterns are not limited to oriC and the the replication terminus region. A 1 Mb region including oriC, called Ori domain, migrates to cell poles and a 1 Mb region including the replication terminus region, called Ter domain, is segregated at the midcell (16). There may be a region required for migration to the cell poles in the Ori domain. Initially, oriC sequences themselves play a role in migration of the chromosome in addition to their role in the initiation of chromosome

54

Daisuke Shiomi & Hironori Niki

replication. However, a plasmid carrying oriC exhibits a different localization pattern from that of the chromosome (17); moreover, the Ori domain of chromosome of cells lacking oriC migrates to cell poles (18). This observation indicates that a sequence other than oriC is involved in the migration of chromosomes to the cell poles. In B. subtilis, a 100-kb region proximal to the replication origin has an activity important to movement of the chromosomes to the cell poles (19). However, it is unknown whether a specific sequence in this region is required for the migration or specific sequences are scattered in the region.

8. Chromosome segregation A gross outline of the mechanisms of chromosome segregation in E. coli is understood. As shown in Figure 6, during the process of chromosome migration, the replication machinery localizes to midcell and chromosome replication begins. Replicated chromosomes migrate to the cell poles to be easily segregated into daughter cells. Replicated chromosome regions containing oriC are actively segregated to cell poles. A specific sequence proximal to oriC is required for the migration. Newly replicated chromosomes at the cell poles are folded by topoisomerase and MukB condensin. Chromosomes replicated at midcell are lead to and folded at cell poles. Finally, the replication termination region is replicated, and these replicated regions are segregated to daughter cells by resolution of the catenane and the chromosome dimer while cytokinesis occurs.

9. Plasmid partitioning The maintenance of extrachromosomal DNA, particularly plasmid DNA, differs from that of chromosomal DNA. Although plasmids are not generally essential for bacterial growth, host cells can acquire novel phenotypes when they retain a plasmid. Most plasmids are circular, and their sizes can vary greatly, from a few kilobases to several hundred kilobases. Furthermore, their copy numbers can vary from a few copies/cell to several hundred copies/cell, and the regulation of plasmid replication is diverse. Some plasmids can be transferred between bacteria. Because of this phenomenon, the number of bacteria carrying plasmids can be increased. Because the R plasmid, which is a multidrug resistant plasmid, can be transferred from one host to another and, therefore, confer drug resistance to the new host cells, new problems are raised in the medical field. Serious problems can also arise when plasmids carry virulence factors rather than drug resistance. Some of bacteriophages, viruses of bacteria, can exist as plasmids. P1 phage in E. coli exists as a plasmid during lysogeny.

Nucleoid and chromosome

55

The ability of mobile plasmids to be transferred is due to tra genes and mobile plasmids are larger than other plasmids. Plasmids tend to be larger than average if they carry virulence factors or drug resistance factors. Generally, as the size of a plasmid increases, the copy number in the cell decreases. This relationship is especially remarkable for R plasmid and F plasmid that determine bacterial sex. Plasmids can encode a mechanism that regulates plasmid replication to ensure that plasmid copy numbers is the same as host chromosome copy number. Even for plasmids that are more than 100 kb, their structure is maintained, and they are inherited stably. For this reason, F plasmid serve as a cloning vector of BAC (Bacterial Artificial Chromosome) library with 100 to 200 kb genomic DNA. The probability of a cell retaining a plasmid becomes problematic when plasmid copy number is the same as chromosome copy number. If plasmid copy number is high enough and plasmids are distributed randomly in a cell, plasmids can be replicated and partitioned reliably in to daughter cells. Therefore, the probability of cells losing high-copy plasmids would be very low. However, if the plasmid copy number is two, it is necessary to transmit plasmids to daughter cells precisely. In fact, the F, P1, and R plasmids each have a specific partition mechanism for plasmid transmission. Here we will describe partition mechanisms for transmission of the F plasmid and the P1 plasmid. Each of these plasmids require three loci for plasmid partitioning; F plasmid requires sopA, sopB, and sopC (20), and P1 plasmid requires parA, parB and parS (21). The sopA/sopB genes and the parA/parB genes encode proteins. Because sopC and parS genes are cis-acting, they are considered centromere-like sites (Figure 7). The sopA and sopB genes form an operon, and sopC is located downstream of that operon; the parABC loci are organized in a similar fashion. Additionally, the protein coding genes have some homology in the amino acid sequences. Both SopA and ParA have a sequence motif indicative of ATPase activity, and there is homology throughout these proteins. SopB and ParB are also homologous proteins. In contrast, there is sequence homology between the cis-acting sites, sopC and pars; however, both sites contain inverted repeat sequences. Thus, although the DNA sequences to be recognized are different from each other, the fundamental functions of sopAB and parAB genes in plasmid partitioning are the same. Figure 7 is a schematic diagram depicting the functions of sopA, sopB and sopC genes in plasmid partitioning. SopB binds to the sopC region and the SopB-sopC complex localizes to the midcell (22). Because SopB protein alone cannot localize to midcell, it is plausible that SopB protein is activated by binding to sopC and then can interact with a host factor that localizes to

56

Daisuke Shiomi & Hironori Niki

Figure 7. Gene compositions and functions of plasmid partitioning by Par system. Gene composition of the partitioning operons of the F and P1 plasmids are shown. These gene products have similar functions. ParA has two functions; it regulates transcription of itself and binds to the ParB-parC complex. These two functions are regulated by nucleotide binding states of ParA. ParB binds to parC, which is a centromere-like sequence, and ParB tethers the plasmid to a specific site in a cell.

the cytoplasmic membrane at the midcell. As a result, F plasmid can localize to midcell through the SopB protein. F plasmid is replicated at midcell. The newly replicated plasmids migrate to the 1/4 and 3/4 sites along the long axis of a cell via SopA (23). Then, during cytokinesis, replicated plasmids are partitioned into daughter cells. After cell division, the 1/4 and 3/4 sites become the midcells of the respective newborn cells, and the F plasmids can localize to the midcell in each newborn cell. Interestingly, F plasmid lacking the sopABC loci localizes to the cell poles because the nucleoid localizes in the cytoplasm; consequently, the F plasmid is occluded from the cell center. It is thought that partitioning of P1 plasmid by parABS loci is similar to that of F plasmid partitioning. It has been suggested that the ATPase activities of SopA and ParA are required for migration of the respective plasmids to the 1/4 and 3/4 sites, although the mechanisms behind this migration are not known. Genes involved in partitioning of R1 plasmid are different from those involved in partitioning of the F and P1 plasmids. The ParA ATPase protein of the R1 plasmid belongs to the actin superfamily. The crystal structure of ParA of the R1 plasmid has revealed that this protein is structurally

Nucleoid and chromosome

57

homologous to actin (24). Furthermore, immunofluorescence microscopy has revealed that, in E. coli, ParA forms a filamentous structure, which resembles cytoskeleton (25); this observation indicates that the R1 plasmid is moved by a cytoskeletal protein.

10. Origin of genes involved in plasmid partitioning Originally, genes homologous to sopA and sopB have been identified in extrachromosomal entities, such as plasmids and phages. More recently, as the genomes of many bacteria have been sequenced genes homologous to sopA and sopB have been found as chromosomal loci. To date, it has been found that sopA and sopB genes form operons in the chromosomes of over 70 bacteria. However, bacteria lacking chromosomal operons encoding sopAB include E. coli and closely related bacteria (26). Because SopA is homologous to ParA and SopB is homologus to ParB, the protein families that include SopA and SopB are called the ParA family and ParB family, respectively. Phylogenic trees of the ParA and ParB families have revealed interesting characteristics (Figure 8). Each family is further divided into two groups; one in which amino acids sequences are conserved and the other in which amino acids sequences are diverse. Conserved amino acids are found in each group. The first groups of parA and parB homologs are mostly found among chromosomal loci, while the latter groups are found mostly among plasmid and phage loci. Thus, the ParA and ParB families are each divided into a chromosomal group and an extrachromosomal group. Vibrio species have two chromosomes. Vibrio cholera, which has one large and one small chromosome has sets of ParA and ParB on each chromosome. The ParA and ParB genes on a large chromosome belong to a chromosomal group while those on a small chromosome belong to an extrachromosomal group. This observation indicates that the small chromosome was originally a plasmid and that then became a chromosome by acquiring parts of a host chromosome. This conjecture fits with a characteristics of a region of replication origin. Because the composition and the order of genes proximal to the replication origin on the small chromosome are similar to those on the chromosomal region conserved among various bacteria species, it is difficult to hypothesize that the region of the replication origin was transferred to the bacterial chromosome from plasmids and phages. Instead, the parAB genes that had existed in ancient bacteria have been maintained in present bacteria and part of the chromosome containing parAB genes would have been transfered to an extrachromosomal entitiy. Because an extrachromosomal entity can be transferred among bacteria, parAB genes have diversified during these transitions.

58

Daisuke Shiomi & Hironori Niki

Figure 8. Phylogenic tree of ParB protein.

It has been suggested that the ParA and ParB proteins encoded in the B. subtilis and C. crescentus chromosomes play roles in chromosome segregation similar to the roles the homologous proteins play in plasmid partitioning. However, it has been shown that these chromosomally encoded proteins do not play a role in chromosome segregation. Instead, they have other functions. ParA and ParB may play a role in the segregation of relatively small chromosome so that they may have functioned in ancient bacteria, which possibly had small chromosomes. However, as size of bacterial genome increased over 1 Mb, ParA and ParB could not play a role in segregation and instead other segregation systems emerged. In fact, E. coli and its bacteria relatives lack parA and parB genes.

11. Summary Recent advances in cell biology have revealed mechanisms of chromosome segregation. In addition, cytoskeletal proteins are widely conserved in bacteria, and these proteins play roles in cell division and

Nucleoid and chromosome

59

maintenance of cell shape. These findings indicate that the mechanism of chromosome segregation in bacteria to in eukaryotic cells have some similarities. Studies on plasmid partitioning support this idea. Nevertheless, the fact that replication and chromosome segregation take place simultaneously in bacteria is different from the situation in eukaryotes. In order to accomplish replication and segregation simultaneously, bacterial cells have to fold newly replicated chromosome. For this purpose, SMC proteins are required. In summary, chromosome segregation mechanisms in bacteria and eukaryotes may not be as different as was previously thought. It is expected that chromosome segregation mechanisms common to bacterial and eukaryotes will be uncovered in the near future.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26.

A. Murray, and T. Hunt: The Cell Cycle an introduction, Oxford University Press:NY, p.183-200 (1993) T. Horiuchi, M. Hidaka, Cell 54, 515-523 (1988). K. P. Lemon, A. D. Grossman, Science 282, 1516-1519 (1998). F. X. Barre et al., Proc. Natl. Acad. Sci. U S A 98, 8189-8195 (2001). O. Espeli, C. Levine, H. Hassing, K. J. Marians, Mol. Cell 11, 189-201 (2003). D. E. Pettijohn, R. Hechet, Cold Spring Harbor Symp Quant Biol 38, 31-41 (1973). T. Hirano, R. Kobayashi, M. Hirano, Cell 89, 511-521 (1997). M. Hirano, T. Hirano, EMBO J 21, 5733-5744 (2002). S. Gruber, C. H. Haering, K. Nasmyth, Cell 112, 765-777 (2003). H. Niki, A. Jaffé, R. Imamura, T. Ogura, S. Hiraga, EMBO J. 10, 183-193 (1991). J. A. Sawitzke, S. Austin, Proc. Natl. Acad. Sci. U S A 97, 1671-1676 (2000). S. Moriya et al., Mol. Microbiol. 29, 179-187 (1998). H. Niki, S. Hiraga, Genes Dev. 12, 1036-1045 (1998). C. D. Webb et al., Cell 88, 667-674 (1997). R. B. Jensen, L. Shapiro, Proc. Natl. Acad. Sci. U S A 96, 10661-10666 (1999). H. Niki, Y. Yamaichi, S. Hiraga, Genes Dev. 14, 212-223 (2000). H. Niki, S. Hiraga, Mol. Microbiol. 34, 498-503 (1999). G. S. Gordon, R. P. Shivers, A. Wright, Mol. Microbiol. 44, 501-507 (2002). L. J. Wu, J. Errington, EMBO J 21, 4001-4011 (2002). T. Ogura, S. Hiraga, Cell 32, 351-360 (1983). A. L. Abeles, S. A. Friedman, S. J. Austin, J. Mol. Biol. 185, 261-272 (1985). H. Mori et al., J. Biolo.Chemi. 264, 15535-15541 (1989). H. Niki, S. Hiraga, Cell 90, 951-957 (1997). F. van den Ent, J. Moller-Jensen, L. A. Amos, K. Gerdes, J. Lowe, EMBO J 21, 6935-6943 (2002). J. Moller-Jensen, R. B. Jensen, J. Lowe, K. Gerdes, EMBO J 21, 3119-3127 (2002). Y. Yamaichi, H. Niki, Proc. Natl. Acad. Sci. U S A 97, 14656-14661 (2000).