|Home | About | Journals | Submit | Contact Us | Français|
Initiation of chromosome replication is a key event in the life cycle of any organism. Little is known, however, about the regulatory mechanisms of this vital process. Conventionally, the initiation mechanism of chromosome replication in microorganisms has been studied using plasmids in which an origin of chromosome replication has been cloned, rather than using the chromosome itself. The reason for this is that even bacterial chromosomes are so large that biochemical and genetic manipulations become difficult and cumbersome. Recently, the combination of flow cytometry and genetic methods, where modifications of the replication origin are systematically introduced onto the chromosome, has made possible detailed studies of the molecular events involved in the control of replication initiation in Escherichia coli. The results indicate that requirements for initiation at the chromosomal origin, oriC, are drastically different from those for initiation at cloned oriC.
An oriC-containing fragment was first isolated from the chromosome as an autonomously replicating sequence (ARS) and the fragment was subsequently subcloned into a plasmid carrying the Col E1-type replication origin. Since the ColE1 origin, but not oriC, requires the polA gene product (DNA polymerase I) for initiation, the ARS activity of an oriC fragment cloned in such a hybrid plasmid (oriC plasmid) can be determined by assaying the replication ability of the plasmid in polA mutant cells. This hybrid plasmid made it possible to analyze the effects of many mutations, including lethal ones, introduced into oriC and proved its usefulness in defining the minimal oriC fragment required for ARS activity (245 bp, Figure 1). Until recently, all mutational analyses of oriC have been carried out with plasmids, allowing identification of a number of cis-acting regulatory elements within and near oriC. This led to the widely accepted concept that regulation of initiation at oriC is complex and involves a number of different elements in the oriC region (reviewed in Messer and Weigel, 1996). Cloning of oriC was essential for Kornberg and collaborators to develop a reconstituted replication system containing only purified proteins (see Kornberg and Baker, 1992). With this system, the functions of many cis- and trans-acting components were characterized in detail. Therefore, oriC plasmid replication has served as a leading model not only for bacterial chromosomes but also for other replicons including eukaryotic viruses and chromosomes.
The first and critical step of initiation at oriC is DNA strand separation, which is performed by the initiator protein, DnaA (see Kornberg and Baker, 1992; Skarstad and Boye, 1994; Messer and Weigel, 1996). After binding to its five cognate 9 bp sequences (DnaA boxes, Figure 1) within oriC DnaA forms a nucleoprotein complex around which oriC DNA is wrapped. Under conditions in which DNA is negatively supercoiled and ATP is bound to DnaA, the DNA duplex becomes unwound at the three AT-rich 13mer sequences (Figure 1). Next, the DnaB helicase is loaded onto the unwound region by the aid of DnaA and DnaC, forming a prepriming complex. The helicase can unwind the DNA duplex in both directions and priming of new strands and bidirectional chain elongation follow. It should be emphasized that these are the minimal factors and reactions required to make initiation occur in vitro. Several other factors have been shown to affect initiation efficiency, such as topoisomerases, histone-like proteins and transcription (see below).
oriC plasmids share many important features of replication initiation with the chromosome (see Messer and Weigel, 1996); both of them replicate bidirectionally, require transcription and de novo protein synthesis, and they respond similarly to many host mutations. When the cells contain multiple copies of oriC all of them fire in synchrony, whether they are located on the chromosome or on a plasmid (Skarstad et al., 1986; Helmstetter and Leonard, 1987). Based on these and other observations, oriC plasmids have long been thought to be faithful models for the chromosome and results obtained with the plasmids have been taken to reflect what is occurring at a chromosomal oriC. However, on some important points the results are contradictory and controversial. For example, the ARS activity of cloned oriC changes depending on various factors such as the type of cloning vector, the cloning sites, the size of the cloned oriC fragment, the orientation of oriC with respect to the vector, and the host strain. More importantly, recent direct measurements of the activity of chromosomal oriC are at variance with analogous plasmid experiments, compelling a reassessment of the validity of using cloned ARS elements as models of chromosomal origins.
The smallest element having ARS activity, as determined by deletion analysis of oriC plasmids, the minimal oriC (Figure 1), was considered the smallest origin sequence required for chromosome replication. However, the minimal oriC is unable to support plasmid replication when cloned into certain plasmids (Asai et al., 1990). It turned out that an additional AT-rich sequence (Figure 1), immediately to the left of the minimal oriC, is necessary for the ARS activity in these plasmids and that the sequence can be functionally replaced by a promoter directed away from oriC. These results imply that the boundaries of the minimal oriC are conditional rather than absolute, and suggest that the minimal origin sequence necessary for chromosome replication may be different from the smallest ARS.
The importance of the different DnaA boxes at cloned oriC has been demonstrated by mutational analyses. In particular, the significance of DnaA box R4 (Figure 1) was emphasized in experiments where the distance between R4 and the rest of the minimal oriC was varied. The ARS activity of oriC was abolished unless the distance was changed by one helical turn (Woelker and Messer, 1993), suggesting that a specific phasing of R4 with respect to other element(s) within oriC is critical for formation of the initiation complex at cloned oriC.
As the first attempt to address directly the significance of the different oriC elements for chromosome initiation, Bates et al. (1995) recently evaluated the requirement for the DnaA box R4. In contrast to the above results, a chromosomal oriC was shown to be initiated in the absence of R4. Cells carrying an R4-deletion mutation on the chromosome are viable and chromosome replication still depends on DnaA and presence of the mutant oriC, giving the first demonstration that the sequence requirement for the minimal oriC differs dramatically between chromosomal and cloned oriC sites. Flow cytometric analyses show that initiation at oriC lacking R4 is inefficient and that initiation synchrony is lost, suggesting that the initiation complex is formed more effectively on the chromosome in the presence of R4.
The more stringent requirements of cloned oriC for R4 could be explained by competition for DnaA protein between wild type and mutant oriC sites. A strain replicating its chromosome from an integrated R1 origin, inserted into and thereby inactivating oriC, allows limited replication of an oriC plasmid carrying an R4 deletion, but only when DnaA protein is overproduced (Langer et al., 1996). In this strain, other oriC mutations on plasmids are also suppressed, at least partially, and the suppression is more pronounced by overproduction of DnaA, suggesting that the mutant origins require higher concentrations of DnaA for initiation than does the chromosomal wild type oriC. Thus, plasmids replicating from such mutant oriC sites may be only partially able to maintain a replication efficiency compatible with stable plasmid maintenance. A prediction of this model is that when an R4-deleted oriC is placed on the chromosome, it should support chromosome replication, because there is no competition for DnaA. This has indeed been observed (Bates et al., 1995).
An alternative explanation for the discrepancy in the requirement for box R4 is that the minimal origin sequence necessary for initiation, in any given case, is determined in part by the superhelical structure of oriC DNA. In support of this, initiation at a chromosomal oriC carrying a deletion of R4 requires transcription near oriC (Bates et al., 1997) and transcription is known to alter the local superhelicity. The following evidence suggests that chromosomal and cloned oriC sites likely exist in different topological states. (i) The two sites show differences in the requirements for histone-like proteins, which change the structure of oriC DNA (see below). (ii) Certain oriC plasmids are unable to transform mutant cells with decreased negative DNA supercoiling (Leonard et al., 1985). The latter observation indicates that cloned oriC is more sensitive to DNA relaxation than chromosomal oriC and suggests that R4 may become more important when template supercoiling is lowered. This idea is consistent with the in vitro observation that deletion of R4 severely inhibits initiation in a reconstituted replication system where the number of free supercoils has been reduced by high amounts of the histone-like protein HU (Bramhill and Kornberg, 1988).
The ability of DnaA to induce duplex opening at the 13mers in vitro is modified by several factors, such as temperature, the density of free supercoils (those not constrained by DNA binding proteins), and the histone-like proteins HU, IHF, and Fis (Baker and Kornberg, 1988; see below). Binding of these proteins induces DNA bending and helps form higher order DNA structures in many different physiological processes (Schmid, 1990). At low concentrations, HU or IHF assists DnaA in duplex opening by facilitating formation of the oriC-DnaA complex (Skarstad et al, 1990). DNA bending may also promote initiation directly by reducing the energy necessary for strand separation. Cells deficient in both HU and IHF are viable (Kano et al., 1991), demonstrating that neither of them is essential for initiation from chromosomal oriC, although flow cytometric analyses indicate that they are necessary for synchronous initiation (Boye et al., 1992; Jaffé et al., 1997). On the other hand, oriC plasmids are unable to replicate in cells lacking both HU and IHF (Kano et al., 1991), showing that the requirements of transacting factors as well are different between chromosomal and cloned oriC sites.
Fis binds strongly to oriC between DnaA boxes R2 and R3 (Figure 1), where it induces DNA bending (Gille et al., 1991). The data on the function of Fis in initiating chromosomal and plasmid copies of oriC are also contradictory. Several lines of evidence suggest that oriC plasmid replication requires Fis in vivo. (i) oriC plasmids transform fis mutants with significantly reduced frequencies at 37 and 42°C (Gille et al., 1991; Filutowicz et al., 1992). (ii) oriC plasmids are poorly maintained in fis mutants even at low temperatures (Filutowicz et al., 1992), suggesting that Fis is a component of the replication machinery at all temperatures. (iii) Base changes in the Fis binding site inactivate cloned oriC (Roth et al., 1994). In contrast, Fis is dispensable for initiation of chromosome replication. fis mutants grow reasonably well under standard laboratory conditions (Filutowicz et al., 1992) and flow cytometric analyses show that the DNA/mass ratio is only slightly reduced (Bates et al., 1997). In fis null mutants, DNA synthesis stops upon a shift from 32 to 44°C (Filutowicz et al., 1992), suggesting that, as is the case for oriC plasmid replication, chromosome replication requires Fis at high temperatures. To our knowledge, Fis is essential for chromosome replication under normal growth conditions only in a mutant strain that lacks the DnaA box R4 on the chromosome (Bates et al., 1995). However, Fis is required to maintain initiation synchrony (Boye et al., 1992). This regulatory role of Fis may depend on its ability to prevent duplex opening at the 13mers, an ability recently demonstrated in vitro (Hiasa and Marians, 1994; Wold et al., 1996).
The idea that one or more transcriptional events independent of any protein synthesis are required for initiation of chromosome replication arose from early experiments demonstrating that rifampicin, an inhibitor of RNA polymerase, prevents initiations at a time when protein synthesis is no longer required (see Messer and Weigel, 1996). On the chromosome, oriC is located between two transcriptional units, the gidAB operon and the mioC gene (Figure 1). Based on the findings that oriC plasmids require both of these transcriptions for efficient replication (see below), a great deal of attention has been directed towards elucidating the roles of gid and mioC transcription in initiation. Given that the protein products of these genes are dispensable for initiation, two possible roles have been postulated for these transcriptions. First, the RNA synthesized may serve as primers for DNA synthesis. Second, the RNA transcript may activate an otherwise inert oriC, for which several different mechanisms have been proposed. It should be noted that these two possibilities are not mutually exclusive.
In reconstituted replication systems, transcription by RNA polymerase is not essential for initiation and RNA primer synthesis can be efficiently carried out by DnaG primase alone. Transcription is necessary to activate initiation only under unfavourable conditions for origin unwinding, including reduced levels of supercoiling, low temperature, or high levels of HU protein (Baker and Kornberg, 1988). Transcriptional activation has been shown to occur in vitro through formation of an R-loop. The position and orientation of promoters with respect to oriC is rather irrelevant for the activation. Thus, both gid and mioC transcriptions could activate oriC via the same mechanism. A GC-rich clamp between an R-loop and the 13mers inhibits transcriptional activation, suggesting that helix instability generated by an R-loop is propagated to the 13mers, thereby stimulating strand opening (see Skarstad et al., 1990; and references therein).
Unlike in vitro systems, transcription is essential in vivo for initiation of oriC plasmid replication. However, a specific transcriptional event required for initiation has not yet been identified; both gid and mioC transcriptions are dispensable for oriC plasmid replication. Even though these transcriptions are not essential, they greatly increase the copy number, and thereby stability, of oriC plasmids. This phenomenon has been observed with most oriC plasmids so far analyzed, including an oriC plasmid that contains relatively long (more than 2 kb) chromosomal sequences at both sides of oriC (Bates et al., 1997). The precise mechanism(s) of the activation of initiation by these transcriptions is yet to be elucidated.
Surprisingly, when on the chromosome, oriC seems to have no need for gid or mioC transcription for efficient initiation. This was first shown in experiments where several mioC mutations that had previously been shown to affect oriC plasmid replication, were introduced onto the chromosome and the effects of the mutations on chromosome replication were analyzed by flow cytometry (Løbner-Olesen and Boye, 1992). Under a variety of growth conditions, all parameters measured (cell mass, DNA/mass, number of origins per cell, and timing of initiation) were the same for wild type and mioC mutant cells. This work was recently extended by showing that deletion of both mioC and gid promoters has little, if any, effect on chromosome replication (Bates et al., 1997). Even in cells deficient in IHF and/or Fis, deletion of these promoters showed only subtle effects on the DNA/mass ratio. This is in stark contrast to the observation that mioC transcription is essential for oriC plasmid replication in cells lacking IHF (Kano et al., 1991). Only when the chromosomal oriC was severely impaired by deletion of the DnaA box R4, was transcription from either mioC or gid required to activate initiation. Again, a marked difference between the requirements of cloned and chromosomal oriC has been demonstrated.
The above experiments demonstrated that transcription of gid or mioC do not represent the rifampicin sensitive step in initiation. We suggest three reasonable explanations for the effect of rifampicin on initiation: First, it is possible that transcription is not actually required for chromosome replication but that rifampicin blocks initiation directly. Conceivably, rifampicin could arrest initiation by forming an inhibitory complex with RNA polymerase that is fixed within or near oriC. This possibility is suggested by the finding that rifampicin blocks DNA synthesis in a reconstituted system utilizing DnaG primase, but only in the co-presence of RNA polymerase holoenzyme (Ogawa et al., 1985). Second, it is possible that transcription originating within oriC is responsible for activating initiation. Unfortunately, these promoters are embedded in the complex sequence of oriC, making mutational analysis extremely difficult. On the basis of genetic data, DnaA protein has been suggested to modify a fraction of the RNA polymerases by direct interaction, which might activate oriC by initiating transcription within oriC (Hansen, 1995). Third, the rifampicin effect may not be specific to a single promoter, but a result of the global shut down of all transcription on the chromosome. Studies have shown that the sedimentation rates of E. coli nucleoids are greatly decreased upon treatment with rifampicin (Pettijohn and Hecht, 1973). This suggests that rifampicin treatment results in drastic changes in the overall topology of the chromosome, which might render oriC incapable of duplex melting and/or binding the required initiation proteins.
The evidence presented above indicates that many cis- and trans-acting components required for oriC plasmid replication are dispensable for chromosome replication. It is possible that the actual mechanism of initiation on the chromosome may be rather simple: As long as the chromosome contains sufficient negative supercoils, DnaA alone may be able to trigger origin unwinding without the aid of any auxiliary components such as histone-like proteins and specific transcriptional events. Some requirements for initiation at cloned oriC might be laboratory artifacts created by moving oriC into small plasmids which are apparently less capable, as compared to the chromosome, of adapting to changes in DNA supercoiling. The idea is consistent with the fact that most components that are required only for cloned oriC affect DNA topology. It is also possible, on the other hand, that these auxiliary components may become essential for chromosome replication under certain environmental conditions. In fact, regulation of chromosome replication is disturbed, even under ideal laboratory growth conditions, by mutations in some of the auxiliary components. It has been demonstrated that various environmental signals such as temperature, osmolarity, nutrients, pH, and availability of oxygen change the density of DNA supercoiling in vivo. The environmentally induced changes are of a similar magnitude to those induced by gyrase inhibitors or mutant topoisomerases, which are known to have significant effects on DNA replication and other cellular processes. Therefore, it is not unlikely that initiation of chromosome replication is achieved without any auxiliary components only under conditions that provide proper DNA supercoiling of the chromosome. It is now clear that chromosomal and cloned oriC sites have different requirements for initiation in the same cell. Thus, the fundamental question of how E. coli controls DNA synthesis, and in particular how this process is adapted to environmental changes, can only be addressed by analyzing replication initiation with chromosomal oriC. However, data accumulated with the oriC-plasmid model system will undoubtedly be helpful in pursuing this question. In addition, a comparison of plasmid-based and chromosomal origins should be considered in studies of other bacteria and of yeasts, where the definition of ARS activity is heavily based on plasmid and minichromosome model systems.
This work was supported by Grant GM22092 from the National Institute of Health and by Grant BIR-9218818 from the National Science Foundation to T.K. and by a grant from the Norwegian Cancer Society and the Norwegian Research Council to E.B.