The mechanisms driving bacterial chromosome segregation remain poorly characterized. While a number of factors influencing chromosome segregation have been described in recent years, none of them appeared to play an essential role in the process comparable to the eukaryotic centromere/spindle complex. The research community involved in bacterial chromosome was becoming familiar with the fact that bacteria have selected multiple redundant systems to ensure correct chromosome segregation. Over the past few years a new perspective came out that entropic forces generated by the confinement of the chromosome in the crowded nucleoid shell could be sufficient to segregate the chromosome. The segregating factors would only be required to create adequate conditions for entropy to do its job. In the article by Yazdi et al. (2012) in this issue of Molecular Microbiology, this model was challenged experimentally in live Escherichia coli cells. A Fis–GFP fusion was used to follow nucleoid choreography and analyse it from a polymer physics perspective. Their results suggest strongly that E. coli nucleoids behave as self-adherent polymers. Such a structuring and the specific segregation patterns observed do not support an entropic like segregation model. Are we back to the pre-entropic era?
Chromosome segregation in bacteria is one of the most mysterious events of the cell cycle. Bacterial chromosome segregation has only been studied in detail in four different bacteria, Escherichia coli, Caulobacter crescentus, Bacillus subtilis and Vibrio cholera (for recent reviews, Possoz et al., 2012; Reyes-Lamothe et al., 2012). These four organisms present related but clearly distinct strategies to segregate and organize their chromosomes. Several years of studies revealed a number of factors that influence chromosome segregation and organization. Among them the partition complexes of the ParABS family (Gerdes et al., 2010) and the capture–extrusion process by a central replication factory (Lemon and Grossman, 2000), appeared to have the strongest potential to drive or to be part of a segregation machinery. However, these factors are not conserved across all model bacteria. It is therefore unlikely that they form a bacterial equivalent of the kinetochore/spindle of the eukaryotic chromosome segregation machinery. Mutants strongly affecting chromosome segregation, which lead to the observation of a large number of anucleated cells in the population, have been found in the genes coding for the topoisomerases (Gyrase and TopoIV) and in the SMC proteins (MukBEF in E. coli). However it is likely that topoisomerases, because of their untangling properties, and SMC proteins, because of their condensing activities, contribute to segregation by producing a ‘segregation competent’ structuring to the chromosome rather than being part of bona fide segregation machinery.
We can argue that segregation of a large bacterial chromosome (3–7 Mb), about 3 × 109 Da could only be achieved by a protein complex of equivalent size. The ParABS complexes involved in segregation of large plasmids and present on the B. subtilis, V. cholera or C. crescentus chromosomes have adequate size (∼ 107–108 Da) (Ptacin et al., 2010). They are crucial for the segregation of the plasmids; however, their removal has limited effects on segregation of the chromosome (Ireton et al., 1994; Yamaichi et al., 2007). Such a system is lacking in E. coli. The discovery of the protein filaments forming a bacterial cytoskeleton (for recent review, (Ingerson-Mahar and Gitai, 2012) and evidence suggesting that MreB contributes to the segregation of the origin of replication in C. crescentus (Gitai et al., 2005) suggested that the cytoskeleton might play an important role in the segregation of the chromosome in rod/crescent shape bacteria. Such a gigantic structure has the potential to influence the whole chromosome; it could couple cell shape and chromosome structuring and segregation. However, mreB inactivation has no effects on E. coli chromosome segregation (Wang and Sherratt, 2010) and even in C. crescentus the phenotype of mreB mutant is only conditional (Shebelut et al., 2009).
The work of Jun and Mulder in 2006 provided a completely new way of seeing bacterial chromosome segregation (Jun and Mulder, 2006). They postulated and validated in silico that the physical properties of the bacterial chromosome provide the directionality for the action of segregation factors. Conformational entropy would be the major force leading to segregation. Therefore proteins identified as segregation factors may in fact function only to create the right physical conditions for entropy-driven segregation. This proposal relies strongly on the folding of the chromosome as a simple random-coil-like polymer compacted only by external crowding forces (Jun and Wright, 2010). This work encountered an immediate success in the bacterial chromosome community. A related segregation model was proposed by Bates and Kleckner (2005); it postulates that forces generated by the repulsion between cohesive sister chromatids are able to trigger chromosome segregation (Bates and Kleckner, 2005). For both models the leading elements for segregation are the intrinsic physical and mechanical properties of the chromosome.
Experiments that can directly test the impact of entropy compared with that of putative segregation machinery are difficult to set up. At this time, several recent results analysing the structure of purified nucleoids (Romantsov et al., 2007; Pelletier et al., 2012), modelling and simulating the crowding of the bacterial cytoplasm and nucleoid (McGuffee and Elcock, 2010; Pelletier et al., 2012) suggest that entropic forces might be sufficient to compress the nucleoid and support for a model in which the chromosome behaves as a loaded entropic spring. Moreover the existence of a long cohesion step in the cell cycle has been confirmed and part of its determinants characterized (Joshi et al., 2011; Lesterlin et al., 2012), therefore the conditions required for an abrupt segregation triggered by cohesion loss and subsequent self-avoiding repulsion of the sisters are encountered in the cell. Although analysis of individual loci labelled with fluorescent tags consistently reports that strong intranucleoid (Danilova et al., 2007; Wiggins et al., 2010) and nucleoid-cell architecture interactions (Espeli et al., 2012) influence nucleoid shape and segregation, entropy driven segregation became a popular model. This model has been challenged in vivo in the article by Yazdi and collaborators in this issue of Molecular Microbiology.
Fast-growing E. coli cells present self-adherent polymer-like chromosomes
In this issue of Molecular Microbiology, Yazdi, Guet, Johnson and Marko present a thorough analysis of the chromosome folding in live E. coli cells. They examine whether the E. coli chromosome behaves as a confined but otherwise unconstrained random coil polymer or alternatively is structured with the characteristics of a self-adherent nucleoprotein complex. According to polymer physics concepts, Yazdi et al. postulated that an unconstrained random coil model implies several important physical characteristics. First, the chromosome should have a diffuse homogeneous appearance; second, segregating nucleoids have to be immediately adjacent; third, no complex geometrical structures should persist for minutes (Fig. 1). By opposition a self-adherent structure could present a define topography with movement governed by specific timing rules and various dynamics range.
Yazdi et al. observed and measured nucleoid dynamics from a physical perspective. We could also say with a ‘Fisical’ perspective since their nucleoid shape reporter is a fusion between the Fis, nucleoid-associated protein and GFP. Fis displays hundreds of binding sites dispersed on the chromosome. The average distance between occupied Fis binding sites, in exponentially growing cells, is 20 kb (Grainger et al., 2006). Recent super-resolution experiments suggest that Fis binding does not involve clustering of the binding sites (Wang et al., 2011), as observed for H-NS. All these characteristics make Fis a good reporter of chromosome geometry. This work might have a strong impact on the research field because the authors took advantage of a custom made microchamber that permits calibrated growth conditions and allows monitoring of chromosome dynamics with various microscopy techniques on a number of cells, making quantitative analysis of bulk chromosome segregation possible for live cells. Yazdi and colleagues described chromosome structuring with characteristics that are growth medium-dependent, presumably dependent on the type of cell cycle. Rapidly growing cells, in LB medium, present nucleoids with complex topography consisting of a number (2–4) of domains. Fluorescence correlation analysis and FRAP (fluorescence recovery after photobleaching) performed for the first time on the bacterial nucleoid, revealed that the domain structuring evolves slowly (with a minute long time-scale). They observed an abrupt separation of the sister nucleoids (few hundreds of nm in less than 5 min). Segregated nucleoids were frequently connected by a thin extended DNA filament crossing the mid-cell. Altogether these results strongly suggest that in rich medium the nucleoid is a compact, self-adhering object rather than a purely confinement-shaped random coil (Fig. 1). The corollary of this analysis is that a simple entropy driven mechanism is not the sole driver of chromosome segregation.
Global translation of the nucleoid
Yazdi and collaborators observed a global displacement of the nucleoid from a polar positioning to a symmetrical positioning around the mid-cell axis (Fig. 1). This event has been described previously in a snapshot analysis of synchronous cells by Bates and Kleckner (2005). The present work extends this observation to live cells in different medium, and provides its first kinetic characteristics. However, it is not yet easy to interpret this chromosome translation; Bates and colleagues proposed that the origin and terminus of replication are released from putative determinants previously linked them to fixed cellular positions. It is possible that the anchoring of the Ter macrodomain to the septal ring through the MatP–ZapB association is involved in the control of the nucleoid translation (Espeli et al., 2012).
Helical shape of the nucleoid
An additional observation of Yazdi et al. is that the E. coli chromosome forms a coil structure inside the cell. This structure is similar to a helix but with no fixed chirality. The folding of the bacterial chromosome as a coiled structure has been already proposed in few recent studies. First, using fluorescent nucleotides, Berlatzky and collaborators described the geometry of the replicating B. subtilis chromosome: labelled nucleotides were incorporated during the replication allowing them to visualize the newly synthesized chromosome (Berlatzky et al., 2008). This chromosome took the form of a helical structure wrapping up around the old chromosome. Second, Umbarger and collaborators, using chromosome conformation capture carbon copy (5C), described the physical interaction network of the C. crescentus chromosome it revealed an ellipsoidal structure with chromosome arms wrapped sinusoidally through space (Umbarger et al., 2011). Thus, the non-replicating or replicating chromosomes of three different species have been described by three different methods as forming a coiled structure. Since these three species belong to different clades the question is raised of whether a helical structure is a common feature of most bacterial chromosomes. For the last 10 years many proteins have been described as forming helical structures in rod shape bacteria (Jones et al., 2001; Ben-Yehuda and Losick, 2002; Ebersbach and Gerdes, 2004; Gibbs et al., 2004; Boeneman et al., 2009; Russell and Keiler, 2009). Even if the helical pattern observed for the MreB filament family has been recently reinterpreted as series of moving patches perpendicular to the cell axis (Dominguez-Escobar et al., 2011; Garner et al., 2011), coiled structures became a common feature in many bacteria. As suggested in Yazdi's article the nucleoid geometry could be the result of protein structures previously described or the chromosome by itself could be the instigator of the chromosomal helical structure and maybe even of the other helical structures in bacterial cells.
Why does the chromosome segregate differently in LB and minimal medium?
Only the data obtained with fast-growing cells clearly show that the bacterial chromosome is best described as a self-adherent polymer whose segregation is an active process. At slow growth rate the edges of the nucleoid are less precisely defined, the existence of domains is not obvious and most importantly, the segregating nucleoids stay adjacent to one another until the cell itself divides. All these characteristics could fit with a random coil organization of the chromosome. It is important to note that, like in rich medium, the nucleoid dynamics are slow and the helical shape of the nucleoid can still be observed in poor media.
So, do bacteria possess two different chromosome organizations and two mechanisms to segregate their nucleoids, one adapted to slow growth, mostly based on entropy, and one adapted to faster growth? The principal structuring determinants of the chromosome are: plectonemic supercoiling, nucleoid-associated proteins, and the processes of replication and transcription. Supercoiling level (Balke and Gralla, 1987), type II topoisomerases (L. Le Chat and O. Espéli, unpubl. results) and NAPs concentrations are similar during steady-state growth in rich or poor media (Travers and Muskhelishvili, 2005). A major difference is observed for the clustering of RNA polymerases, presumably around rRNA operons (Cabrera et al., 2009), and the clustering of DNA polymerases (Fossum et al., 2007) which are only observed in rich medium. Could these clusters, by themselves provoke a specific folding of the entire chromosome? The hypothesis that entropy could be enough to drive segregation at a slow growth rate but would not be sufficient at fast growth rate is sustained by several observations. First choreography of individual chromosomal loci differs according to the growth rate (Fossum et al., 2007; Adachi et al., 2008). Second a number of mutants affecting proteins involved in DNA management show weak phenotypes at slow growth rates and dramatic defects in LB; see for example MatP and MukB minus cells (Niki et al., 1991; Mercier et al., 2008). Could these factors involved in dedicated chromosome segregation machinery that has evolved mostly to allow fast growing? What could be the selective pressure involved in maintaining such an accessory segregation machinery? The questions remain open. On a final point, many studies on bacterial chromosome segregation are realized with slow growing cells, mostly because fast-growing cells add more technical limitations and difficulties. However, the opposable results described in this work ask for a careful re-examination and some re-experimentation of bacterial chromosome behaviour in fast-growing conditions.
We thank D. Grainger, K. Marians, J. Marko and Christophe Possoz for helpful discussions. We thank C. Herbert for careful reading of the manuscript. L.L.C. benefits from ANR 2010 JCJC SISTERS fellowship to O.E.