|Home | About | Journals | Submit | Contact Us | Français|
It has been postulated that bacterial DNA replication occurs via a factory mechanism in which unreplicated DNA is spooled into a centrally located replisome and newly synthesized DNA is discharged towards opposite cell poles. Although there is considerable support for this view, it does not fit with many key observations. I review new findings, and provide alternative interpretations for old findings, which challenge this model. As a whole, current data suggest that the replisome, at least in slowly growing Escherichia coli cells, tracks along a stationary chromosome. These replisomes are not stationary, tethered or restricted in their movement, but rather travel throughout the nucleoid. One possibility is that the replisome navigates along a chromosome made up of looped domains as has been previously envisioned.
In bacteria, DNA replication is initiated at a unique site, ori, creating two replication forks. Forks proceed in opposite directions from ori (bidirectionally), eventually converging in a broad terminus region, ter. During rapid growth, division intervals are shorter than the time required to complete a single round of replication, so cells enter so-called multi-forked replication mode, in which extra initiations at nascent replication origins lead to branched chromosomes with overall higher net replication rates. Replisome dynamics during multi-forked replication may be fundamentally different than in slowly growing cells (see Hyper-structures and eukaryotic replication clusters), and for the purposes of this discussion, we will focus on observations made during slow growth. Replication at each fork is carried out by a single multi-protein complex known as the replisome, responsible for unwinding, priming and copying the leading and lagging strands simultaneously (see O’Donnell, 2006 for review). There are two general models that describe how replisomes move from ori to ter. In the ‘tracking’ model, sister replisomes (left- and right-oriented replisomes resulting from a single initiation event) migrate away from each other, tracking along the DNA template (Fig. 1A). In the ‘spooling’ model, it is the DNA that moves, spooling through a stationary replisome (Fig. 1B). A variation of the spooling replisome model is the so-called factory model, in which left and right replisomes are physically coupled together throughout DNA replication. This model, originally proposed by Dingman (1974), was based on the premise that the replication forks are tethered to the cell membrane (Jacob et al., 1966). In principle, tethering the replisomes to the membrane could provide an anchor from which the replisome could generate sufficient torque to pull the chromosome inward, and push replicated duplexes outward. This model further predicted that without the need for physical separation of the forks, the replisomes are oriented head to head, so that one replisome copies the entire top strand, and the other copies the entire bottom strand (Fig. 2B). This is fundamentally different from the traditional ‘tail-to-tail’ configuration (Fig. 2A), in which one replisome copies the left half of the chromosome and one replisome copies the right half of the chromosome. In theory, a spooling replisome could utilize either of these mechanisms if the replication machinery was anchored in place. However, a tracking replisome can only function in the tail-to-tail configuration, allowing replisome separation without unravelling the unreplicated duplex.
The spooling model received little attention until reports of factory-like replication foci (cells with only one spot) in several species of bacteria (examples in Bacillus subtilis, Caulobacter crescentus and Escherichia coli; Lemon and Grossman, 1998; Brendler et al., 2000; Jensen et al., 2001). This model was well accepted in the eukaryotic field, which had for some time realized that replication foci occur in discrete zones within the nucleus (Hozak et al., 1993; Newport and Yan, 1996). These replication clusters were termed ‘factories’, not in the strict sense of the original Dingman model (coupled replisomes), but rather in the sense that multiple replication forks loosely cluster together inside the cell. For the purposes of this review, eukaryotic replisome structures will be referred to as replication clusters, and the term ‘factory’ will be reserved for sister replisomes that are physically coupled. The spooling model was also received well in the chromosome dynamics field, as coupled replisomes provided a nifty mechanism for separation of sister duplexes, known as the replisome extrusion hypothesis. Since these original findings, several lines of evidence have called into question the validity of the spooling model: (i) there does not appear to be any cross-talk between sister replisomes after one replisome has been artificially blocked (Breier et al., 2005), (ii) replisomes move dynamically within the cell, and are often visible as two distinct foci (A. Wright, pers. comm.; Migocki et al., 2004; Bates and Kleckner, 2005; Berkmen and Grossman, 2006; Reyes-Lamothe et al., 2008) and (iii) newly replicated sequences do not immediately separate from each other (Bates and Kleckner, 2005; Nielsen et al., 2007), as predicted by the replisome extrusion model.
Distinguishing between these two models is often very difficult, and it is likely that the truth lies somewhere between them. For example, one could imagine that tracking replisomes might frequently become impeded from moving freely through the cellular/nucleoid milieu, and the DNA begins to thread through the replisome to avoid a replication stoppage. After the impediment passes, the replisome can continue its course tracking the chromosome. Frequent inconsistencies between reported data complicate the problem, and some of this may be due to the blunt nature of light microscopy as a tool for studying in vivo enzymology. Also however, there is increasing evidence that different bacterial species have evolved specialized mechanisms to organize their replisomes.
The view that bacteria might utilize a factory-like mechanism for DNA replication arose from groundbreaking work by Lemon and Grossman (1998), in which replisomes were imaged for the first time in living cells. Labelling the catalytic subunit of DNA polymerase with green fluorescent protein (GFP) in slowly growing B. subtilis cells, they discovered that a large proportion of cells contained a single focus (33–56% depending on growth rate). This finding did not fit with current logic, as DNA replication is bidirectional, and all cells should either contain two replisomes or none at all. The preponderance of cells with a single visible replisome focus was therefore inferred to mean that both clockwise and counterclockwise replication forks were colocalized in a single macromolecular complex, or factory, for much of the replication period. A smaller portion of cells containing two individual foci (19–32%) were longer on average than those cells containing a single focus, and were therefore concluded to represent the separation of replisomes from the factory as the forks approached the terminus. It was originally proposed that template DNA was spooled into an anchored replication factory, with newly replicated DNA being ejected in opposite directions (Lemon and Grossman, 2000). This model was later modified to account for the fact that replisomes are in fact not anchored (Berkmen and Grossman, 2006; discussed below). The factory model was bolstered by the finding that a completely different bacterium, C. crescentus, also exhibits coupled replisomes (Jensen et al., 2001). In this case, almost all cells exhibited either one or no replication foci (designated the clamp loader), suggesting that the forks did not separate during DNA replication. It is possible that the particular conditions used favoured paired replisomes, or that C. crescentus replisomes are arranged differently.
In E. coli, SeqA protein has often been used as a marker for replication forks. SeqA is an abundant protein that binds hemimethylated GATC sequences along the chromosome (Brendler and Austin, 1999), and forms distinct foci by immunofluoresence (Hiraga et al., 1998) and by SeqA–GFP fusions (Koppes et al., 1999; Onogi et al., 1999; Brendler et al., 2000; Molina and Skarstad, 2004). As hemimethylated GATC sequences occur only on newly replicated DNA (remethylation by Dam occurs ~2–5 min after passage of the fork; Slater et al., 1995), it is assumed that SeqA foci represent a tract of protein that decorate DNA exiting the forks (Brendler et al., 2000). Reports vary, but SeqA foci most often occur as two well-defined foci near midcell (e.g. Brendler et al., 2000). This arrangement is compatible with the replisome extrusion model, but could also result from a SeqA-mediated organization of replicated DNA from two independent and mobile replisomes. Clearly, SeqA foci represent several hundreds of molecules, which, given their strong and cooperative DNA binding characteristics (Brendler and Austin, 1999), might work as DNA ‘wranglers’ collecting the newly synthesized DNA from replisomes tracking DNA template throughout the nucleoid. Dual labelling experiments (D. Bates, unpublished) indicate that < 25% of SeqA foci colocalize with a DnaX focus (DNA polymerase clamp loader) under slow growth conditions, further indicating that SeqA does not mark the forks (this is not necessarily true under fast growth; see Hyperstructures and eukaryotic replication clusters below).
Pioneering work in the lab of Andrew Wright showed that E. coli replisomes, visualized by DnaX–GFP and SSB–GFP (single-stranded binding protein) fusions, are highly mobile structures, frequently switching between single foci and two well-separated foci (A. Wright, unpublished). Notably, SSB foci, likely marking the Okazaki fragments on the lagging strand, colocalized with DnaX foci, ruling out the possibility that polymerase subunit foci represent ‘reserve’ proteins instead of active forks. Based on the independent movement of the two foci, and the random nature of splitting and merging (seconds to minutes), Wright reasoned that the observed patterns were due to movement of two completely independent replisomes tracking on a rosette-shaped chromosome (Fig. 1A). To explain not only why foci were moving separately but also why they sometimes came back together, it was reasoned that ‘merging’ represented the tracking of replisomes towards centre on a looped domain structure. This so-called train on a track model, in which left and right replisomes track independently over a relatively immobile chromosome, is a great departure from the factory model.
Does the rate and position of these migrating replication foci reflect actual chromosome structure? Evidence suggests that it does. Although the true structure of the E. coli chromosome is unknown, it is estimated to be organized into ~50–400 topological domains (Higgins et al., 2005). From electron microscopy studies, it appears that the nucleoid might take the form of large loops radiating from a central core, i.e. a rosette structure (Robinow and Kellenberger, 1994). If one of these loops were about 0.5 µm in length (along the long axis of the cell), and assuming a DNA synthesis rate of about 1 kb s−1 fork−1, one would expect to see replisome movements in the range of 0.3–3 µm min−1. This is in line with average measurements of 0.4 µm min−1 for SSB protein (Reyes-Lamothe et al., 2008). Interestingly, if the tracking model turns out to be correct, then the dynamics of replisome movement may tell us something about actual chromosome structure. In subsequent studies, it was found that replisome foci also ‘split and merge’ in B. subtilis (Migocki et al., 2004; Berkmen and Grossman, 2006), suggesting a conserved mechanism. Focus movement here, however, was inferred to be an occasional separation of two loosely coupled replisomes, not the movement of forks along the template. Resolution of these opposing models requires analysis of chromosome dynamics (see Chromosome dynamics tend to favour a tracking replisome, below), but in either case, sister replisomes in E. coli and B. subtilis clearly do not form an obligate macromolecular complex or factory.
Further support for the tracking model came from the analysis of DnaX–GFP foci in synchronized E. coli cells (Bates and Kleckner, 2005). This study revealed that replisomes progress through defined stages during the cell cycle, from colocalized to highly separated. By precisely measuring the timing of DNA replication, it was shown that replisomes were initially colocalized in a single focus near the midcell for the first third of S-phase, and then separated into two highly dynamic foci until replication termination. Split replisomes occurred randomly over the entire nucleoid, and occasionally overlapped to form one focus. Colocalization of DnaX foci early in the replication period could be interpreted to mean that E. coli replisomes start out as a physically coupled, spooling factory, then transition to independent tracking replisomes. If the coupled helicases are oriented head to head (Fig. 2B; Fang et al., 1999), such a transition would require the complete disassembly of both replisomes and an exchange of template strands. A more likely conclusion is that both early and late replisomes utilize a tracking mechanism, with early colocalization resulting from either: (i) left and right forks replicating a tightly packed origin region (Espeli et al., 2008) or (ii) delayed initiation/elongation of the left fork, as was demonstrated by microarray analysis (Breier et al., 2005). Intrinsic attraction of the replisomes is possible, however, and cannot be discriminated from purely coincidental overlap of replisomes. In a separate study (Reyes-Lamo the et al., 2008), a similar but slightly shorter period of early colocalization of replisomes was found in E. coli, with dynamic separation and occasional overlap of replication foci occurring throughout most of S-phase. A transition from midcell replication to dispersed replication part way through S-phase was also found by electron microscopy imaging of 3H pulse-labelled cells (Koppes et al., 1999).
Perhaps the greatest challenge to the factory model was the discovery that blocking one replisome does not affect progression of the other replisome. In an elegant study, Cozzarelli and colleagues (Breier et al., 2005) blocked one fork by transient expression of Tus protein, which bound an ectopic ter site inserted on one arm of the chromosome. Examination of fork progression by microarray analysis showed that blocking one fork did not impede the rate of its sister replisome in any way. Similarly, multiple copies of the lac operator sequence (lacO array), which impose a replication roadblock when bound with Lac repressor protein, did not block replication of the chromosome by the other fork (Possoz et al., 2006).
Along these lines, one can imagine dire consequences of having two tightly coupled sister replisomes. For example, many strains (e.g. W3110) carry large inversions due to recombination at the ribosomal RNA genes, with one replichore significantly longer than the other. If replications around the chromosome were tightly linked, finishing replication of the longer arm would seem difficult at best. Similarly, strains engineered with unbalanced chromosomes by the Louarn and Kuempel labs to identify elements in the terminus region (Kuempel et al., 1989 and references therein) are perfectly viable. Such strains should not have been able to complete DNA replication under a coupled replisome model, because the left and right replisomes were forced to function independently for part of replication. Furthermore, if selection existed for coupled replisomes, such inversions should immediately revert – and they do not.
In B. subtilis and E. coli, replisome foci, when present as a single focus, are located near midcell (e.g. Lemon and Grossman, 1998; Bates and Kleckner, 2005; Reyes-Lamothe et al., 2008). Several lines of evidence suggest that this is a consequence of replication initiation occurring at midcell. (i) In E. coli, oriC is located at midcell at the time of replication initiation, and within 0.3 µm of a solitary DnaX focus (Bates and Kleckner, 2005). (ii) Replisome foci localize with origins that have been mislocalized by induction of a sporulation-specific oriC localization protein in B. subtilis (Berkmen and Grossman, 2006), or by growth in an E. coli mukB strain exhibiting mislocalized origins (Reyes-Lamothe et al., 2008). (iii) The replisome also initially locates to the origin in C. crescentus, which is situated near the stalked cell pole at the start of the cell cycle (Jensen et al., 2001). This theme of replication focus localization being dictated by chromosome position is repeated in eukaryotes (below), and is indicative of the underlying mobility of replication proteins and relative immobility of the chromosome.
Split replisome foci on the other hand occupy a diffuse region over the nucleoid, and are usually, but not always, found in opposite cell halves (Bates and Kleckner, 2005; Reyes-Lamothe et al., 2008). Foci can be separated up to the entire cell length, with an average inter-focus distance of ~0.6 µm in B. subtilis (Berkmen and Grossman, 2006) and ~1.1 µm in E. coli (Bates and Kleckner, 2005), each with large standard deviations (over half the mean). It is unclear whether this difference between E. coli and B. subtilis is significant or is simply due to cell size variation between these organisms. In synchronized cultures, the inter-replisome distance increased linearly during S-phase until just prior to termination, when it decreased (Bates and Kleckner, 2005). A similar pattern was observed in time-lapse experiments (Reyes-Lamothe et al., 2008). Interestingly, the latter authors have pointed out that this observed pattern of replisome migration matches an orientation of the chromosome predicted by their model for chromosome positioning in which ori and ter are located near midcell, with left and right chromosome arms extending into opposite directions along the long axis of the cell (Wang et al., 2006). C. crescentus also exhibits migrating replisome foci but, in this case, a solitary focus migrates from the stalked pole to midcell (Jensen et al., 2001). The authors suggest that this migration might be caused by passive displacement of the ‘factory’ by newly replicated DNA, but it could also be that the replisomes are tracking the chromosome, given that the parental DNA is situated in the cell with the origin near the stalked pole and the terminus near the opposite pole (Viollier et al., 2004).
Split replisome foci tend to locate between the 1/4 and 3/4 positions; ~80–90% in B. subtilis cells (Migocki et al., 2004; Berkmen and Grossman, 2006), and 83% in E. coli (Bates and Kleckner, 2005). It has been suggested that this is evidence against a tracking replisome model, which should exhibit replisomes equally dispersed throughout the cell. This conclusion may not be right however, because nucleoid volumes are only ~25% of cellular volumes (Zimmerman, 2006). Second, 3D DAPI studies indicate that the great majority of the total nucleoid volume (78%) is located between the cell quarter positions (D. Bates, unpublished). In agreement with this estimation, an analysis of subcellular location of genetic loci spaced ~230 kb apart by FISH showed that ~85% of sites on an unreplicated chromosome were located between the cell quarter positions (Niki et al., 2000). Therefore, one would expect that 78–85% of replication foci should occur between the cell quarters under a tracking replisome model, reasonably close to the observed 80–90% values observed for inter-replisome distances.
The fundamental characteristic that defines spooling and tracking replisomes is not so much replisome movement as chromosome movement, and specifically movement of the unreplicated DNA. In the tracking model, unreplicated DNA need not move towards the replisome. In the spooling model however, unreplicated DNA leaves its ‘resting’ location and moves towards the replisome, usually located in the midcell region. Presuming that the force for spooling is provided directly by the polymerase/helicase motor proteins, template DNA would probably experience huge rates of movement as the replisome pulled the DNA from its stable resting state (chromatin). This was directly tested in B. subtilis by examining the locations of the replisome and of a chromosomal site (marked by a lacO array) in living cells (Lemon and Grossman, 2000). A gradual inward movement of the lacO sequence prior to segregation into two foci was indeed observed. The movement of DNA was not dramatic, going from its resting location between the cell quarter and midcell towards midcell in 30 min, but it is still supportive of the spooling model. Notably, it is not clear whether the replisome also moved during this period, or when the exact time of replication was, making inferences on pre-replication movement difficult.
An analogous study in E. coli showed very different results (Reyes-Lamothe et al., 2008). Here, almost no DNA movement was seen prior to segregation of the locus into two foci, suggesting a tracking replisome is at work. The replisome, on the other hand, often exhibited movement towards the lacO array several minutes before the array split into two foci. Furthermore, replisome foci exhibited significantly higher rates of mobility (diffusion constant) than did chromosomal loci prior to segregation. The presence of an exception perhaps best proves this rule: solitary ter foci clearly move inward towards midcell near the end of DNA replication (Bates and Kleckner, 2005; Nielsen et al., 2006; Reyes-Lamothe et al., 2008). This has been inferred to represent either replication of the last bit of chromosome (pulled towards a septum-interacting replisome), or movement of two catenated ters towards the FtsK/Topo IV decatenation machinery (Espeli et al., 2003). In either case, this is the best in vivo example of DNA being pulled towards a stationary protein assembly. Given the fact that ter movement is almost universally accepted as ‘atypical’, one is left to conclude that other chromosome regions are not pulled towards a stationary complex.
One attractive feature of spooling replisomes is their proposed ability to facilitate sister chromosome segregation by the so-called replisome extrusion mechanism (Lemon and Grossman, 2001; Sawitzke and Austin, 2001). In this model, as parental DNA is spooled into the replisome, newly replicated DNA is ejected towards opposite poles. However, it has recently been shown that sister chromosomes in E. coli incur a period of colocalization at several sites along the chromosome lasting 5–20 min (Bates and Kleckner, 2005; Nielsen et al., 2007). Thus, newly replicated sequences are linked together before separating, with some regions of the chromosome exhibiting a longer period of linkage than others. This phenomenon has often been referred to as ‘cohesion’ because of obvious analogies with chromosome cohesion in eukaryotes. From this finding, it is apparent that DNA ‘extruded’ from the replisome is not immediately brought to opposite poles, as was prior convention, but instead is kept in a central cellular compartment, until subsequent separation 5–20 min later (Bates and Kleckner, 2005). It is difficult to reconcile the replisome extrusion hypothesis, in which replicated DNA is thought to segregate immediately in opposite directions, with the observed colocalization of just-copied sister chromosomes.
During rapid growth, bacterial cells undergo multi-forked replication to increase the cellular rate of DNA synthesis. This has the effect of increasing the number of replication origins per DNA molecule, and is analogous to chromosome replication in eukaryotes, which utilize multiple origins to replicate much larger chromosomes. There is some evidence to suggest that this analogy might extend to how these organisms organize the multiple replisomes; in E. coli, rapidly growing cells typically contain six to twelve active replication forks, as predicted from the number of replication origins per cell measured by flow cytometry, but only contain two to six replication foci, as imaged by BrdU pulse labelling (Molina and Skarstad, 2004). Replication foci are heterogeneous in size and are often irregularly shaped, with some larger foci expected to contain six replication forks. Interestingly, SeqA foci are quite well colocalized with the BrdU foci at this growth rate, suggesting that multiple replication-related proteins aggregate into a single molecular complex, or ‘hyperstructure’ (Molina and Skarstad, 2004). Hyperstructures have been proposed to represent a physical association of related proteins as a means to regulate diverse but related processes including origin firing, replication fork progression, nucleotide synthesis and even gene expression (Mathews, 1993; Norris et al., 2000). Supporting this idea, it was recently shown that several enzymes involved in DNA repair localize to sites of replication in rapidly growing cells (Renzette et al., 2005; Meile et al., 2006; Darmon et al., 2007). The existence of replication hyperstructures certainly satisfies most definitions of a factory, but more information is needed before final conclusions can be drawn about whether hyperstructure replisomes are spooling or tracking.
Eukaryotic cells also contain large replication complexes. In mammalian cells, replication foci or ‘clusters’ (large, irregular-shaped foci ~0.8 µm across) probably represent ~10–20 active forks, replicating a single ~1 Mb chromosome ‘territory’ comprised of ~10 rosettelike loops (Berezney et al., 1995; Jackson and Pombo, 1998). Evidence suggests that all origins (5–10) within a single chromosome territory are synchronously initiated (Newport and Yan, 1996), but the ensuing replisomes are not physically coupled because clusters can be resolved into individual foci (Leonhardt et al., 2000; below). Pulse labelling with nucleotide analogues has shown that neighbouring chromosome territories are subsequently initiated for replication as a fork from the previous territory travels to it (Sadoni et al., 2004). Such a domino-like mechanism suggests that the replication machinery progresses along the chromosome from territory to territory, instead of chromosomes unfolding and moving towards the replication machinery.
The most convincing evidence for a spooling replisome is in budding yeast, in which two sites 30 kb on either side of a replication origin were shown to move together before replication and then separate after a cohesion period (Kitamura et al., 2006). While this kind of movement could be coincidental, or result from replication-induced changes in chromatin structure, it is clearly what would be predicted from a spooling replisome mechanism. Dynamics of replicating DNA in Drosophila (Calvi and Spradling, 2001) and in human cells (Sadoni et al., 2004) have not revealed such spooling-like movements, suggesting that higher eukaryotes might utilize a tracking mechanism.
Apparent differences in replisome behaviour among bacterial species may reflect fundamental differences in the way they replicate their genomes. Perhaps this should not be surprising given that E. coli, B. subtilis and C. crescentus are each separated by over two billion years of evolution (Battistuzzi et al., 2004), more than twice the separation of yeast and humans. B. subtilis replisomes have several structural differences, including a unique catalytic polymerase, the alpha subunit of Pol III holoenzyme, responsible for lagging strand synthesis (Dervyn et al., 2001). Furthermore, origin localization is dissimilar: E. coli origins are generally located a fixed distance from the cell poles, implying a tethering mechanism (Bates and Kleckner, 2005), whereas B. subtilis origins maintain a approximately constant inter-origin distance as the cell and nucleoids grow. Finally, of these three organisms, only E. coli contains SeqA or a GATC methylation system, suggesting that newly replicated DNA might be handled differently in E. coli.
The E. coli chromosome is 3 000 000 kDa in mass without protein. Even single domains represent a tremendous mass to be manipulated by the replisome. A single anchored replisome must simultaneously pull the unreplicated DNA inward, and push the replicated DNA away from it. Even if we assume that the polymerase/helicase motors could generate enough force to pull the DNA template large distances through the cell towards a midcell factory, it is difficult to imagine how two mobile replisomes, now a universally accepted fact in E. coli and B. subtilis, could create the necessary torque without being anchored to some underlying structure. Tracking polymerases likely require no less energy than spooling polymerases, but have the advantage of utilizing the chromosome itself to generate torque. Furthermore, the combined forces imposed on the replisome by incoming and outgoing DNA will propel the replisome forward towards the unreplicated DNA and away from the replicated DNA. One advantage of the Dingman spooling model was that left and right fork forces are opposing, resulting in a zero-balance at the factory.
Arguably, many of the results concerning replisome dynamics – that is, the number, position and mobility of replication foci in cells – can usually be interpreted to support either camp. Several key issues must be addressed: (i) the timing of DNA replication is often assumed rather than measured. DNA replication does not equate to the time a replisome focus happens to coincide with a genetic locus, or even with the time that a genetic locus splits into two visible foci (due to chromosome cohesion). Therefore, inferences of unreplicated DNA spooling into a replisome, or newly replicated DNA extruding from a replisome cannot be made without an independent measure of DNA synthesis. (ii) Our knowledge of chromosome structure is insufficient. Without knowing the layout of the underlying DNA template, how can we infer whether replisome focus movements are diffusion through the cytoplasm, or tracking along the chromosome? This and many of the questions surrounding replisome dynamics will require further studies on small and large-scale chromosome structure in bacteria. Much emphasis has been placed on the dynamics of segregating DNA, with great benefits, but very little attention has been paid to the orientation and movement of parental DNA. (iii) The low resolution of fluorescence microscopy does not reveal small-scale protein movements. This problem is exacerbated in higher eukaryotes, where 1 Mb of DNA, the size of a typical chromosome territory, is about 600 nm across (Zink et al., 1999), only twice the resolving power of a light microscope. The absence of directed movement of replication foci on this scale (e.g. Hozak et al., 1993) is therefore not surprising, especially given that these ‘foci’ (actually clusters of foci as shown by deconvolution; Leonhardt et al., 2000) are really quite large.
Despite the limitations of fluorescence microscopy, it is likely that microscopy will eventually answer these questions, even in bacteria. Bacteria are < 1/10 the size of eukaryotic cells, but they provide a uniquely ‘stretched’ chromosome template over which we can measure replisome movement. A hypothetical bacterial DNA domain, some 50 kb in length (4.6 Mb chromosome divided into 100 domains), would occupy some 0.5 µm, the radius of a typical nucleoid. This works out to a predicted fork speed of 0.6 µm min−1, a very measurable number. For comparison, vertebrate replisomes travelling over an equivalent amount of DNA (1/20 of a typical chromosome territory) only move an estimated 40 nm due to tighter packing of eukaryotic chromosomes. Furthermore, new fluorescence imaging techniques such as fluorescence imaging with one nanometer accuracy (FIONA) utilizing total internal fluorescence and a single-molecule detector (Xie et al., 2008) might eventually allow us to resolve overlapping replication foci into discernible positions.
I would like to thank Nancy Kleckner, Aude Bourniquel, Mohan Joshi, Suckjoon Jun and Peter Kuempel for discussions and comments, and Andrew Wright for sharing unpublished data.