Regulation of the Activity of the Dual-Function DnaA Protein in Caulobacter crescentus

DnaA is a conserved essential bacterial protein that acts as the initiator of chromosomal replication as well as a master transcriptional regulator in Caulobacter crescentus. Thus, the intracellular levels of active DnaA need to be tightly regulated during the cell cycle. Our previous work suggested that DnaA may be regulated at the level of its activity by the replisome-associated protein HdaA. Here, we describe the construction of a mutant DnaA protein [DnaA(R357A)]. The R357 residue in the AAA+ domain of the C. crescentus DnaA protein is equivalent to the R334 residue of the E. coli DnaA protein, which is required for the Regulatory Inactivation of DnaA (RIDA). We found that the expression of the DnaA(R357A) mutant protein in C. crescentus, but not the expression of the wild-type DnaA protein at similar levels, causes a severe phenotype of over-initiation of chromosomal replication and that it blocks cell division. Thus, the mutant DnaA(R357A) protein is hyper-active to promote the initiation of DNA replication, compared to the wild-type DnaA protein. DnaA(R357A) could not replace DnaA in vivo, indicating that the switch in DnaA activity once chromosomal replication has started may be an essential process in C. crescentus. We propose that the inactivation of DnaA is the main mechanism ensuring that chromosomal replication starts only once per cell cycle. We further observed that the R357A substitution in DnaA does not promote the activity of DnaA as a direct transcriptional activator of four important genes, encoding HdaA, the GcrA master cell cycle regulator, the FtsZ cell division protein and the MipZ spatial regulator of cell division. Thus, the AAA+ domain of DnaA may play a role in temporally regulating the bifunctionality of DnaA by reallocating DnaA molecules from initiating DNA replication to transcribing genes within the unique DnaA regulon of C. crescentus.


Introduction
Faithful chromosomal replication requires regulatory networks that ensure the precise coordination of DNA replication with other cell cycle events. The prokaryotic DnaA protein or the eukaryotic origin replication complex (ORC) control the onset of DNA replication. DnaA and several ORC proteins are ATP-binding proteins that contain AAA+ (ATPase associated with diverse cellular activities) domains of very similar structures, suggesting mechanistic conservation throughout evolution [1,2]. Binding of ATP to these proteins is thought to induce conformational changes that activate these proteins to promote the initiation of DNA replication at the correct time of the cell cycle. Besides its function as an initiator of bacterial chromosome replication, DnaA is also a transcription factor that binds to many promoters to regulate their activities in diverse bacterial species [3,4]. It is however still unclear whether oscillations in the levels of DnaA-ATP influence the timing of gene transcription in most cases [5,6].
In the Escherichia coli prokaryotic model, DnaA proteins bind to DnaA boxes located in the origin of replication, but only the ATPbound form (DnaA-ATP) can initiate replication [1]. Soon after the initiation of DNA replication, the ATPase activity of the AAA+ domain of DnaA is stimulated by the essential Hda protein, resulting in the hydrolysis of the ATP bound to DnaA and thus in the regulatory inactivation of DnaA (RIDA) [1,7,8]. Hda supposedly needs to interact with the b-clamp of the DNA polymerase III loaded onto DNA to promote the inactivation of DnaA [9]. Although there exists several other mechanisms that regulate the timing of the initiation of DNA replication in E. coli [1], the RIDA system seems to be one of the most important [10,11]. Consistent with this model, it was shown that ATPase defective mutants of DnaA, and hda-deficient cells, display overinitiation phenotypes in E. coli [7,12,13]. The use of replisomeassociated proteins that act as negative regulators of initiation appears quite conserved in bacteria, as indicated by the discoveries of the YabA protein in Bacillus subtilis [14,15] or the HdaA protein in Caulobacter crescentus [16].
The C. crescentus bacterium undergoes a unique developmental cycle, dividing asymmetrically at each cell cycle, producing two different daughter cells: the stalked cell in S phase, and the swarmer cell in G1 phase until it differentiates into a stalked cell [17,18]. The G1-to-S transition is controlled by the CtrA protein, an important regulator of the initiation of chromosome replication in alphaproteobacteria. This response regulator binds to several sites in the C. crescentus origin of replication (Cori) to inhibit the initiation of DNA replication, probably by competing with DnaA binding [19]. It also binds to many promoters to regulate their activities [20,21]. Levels of the active phosphorylated form of CtrA are tightly controlled by the regulated synthesis, phosphorylation and degradation of the CtrA protein, so that it only accumulates in swarmer and pre-divisional cells [17,18]. A gradient of active CtrA establishes the asymmetry of the initiation of DNA replication between the two future daughter cells before cell division [22]. Active CtrA is however undetectable for about one third of the cell cycle during the stalked cell stage [23]. During the time when no CtrA is detectable, other factors restrict the initiation of chromosomal replication to only once per cell cycle. We previously identified HdaA as one of these factors, showing that the HdaA protein is necessary to ensure that chromosomal replication does not restart a second time during the same cell cycle [16], indicating that HdaA may stimulate the inactivation of DnaA once DNA replication has initiated, by a mechanism similar to the RIDA system in E. coli. To further address this possibility, we describe here the creation of a mutant DnaA protein in C. crescentus, with an R357A substitution in its AAA+ domain. We found that the expression of this protein caused severe over-initiation of chromosomal replication and blocked cell division in C. crescentus, unlike the expression of the wild-type DnaA, indicating that DnaA(R357A) is hyper-active as an initiator of DNA replication. We also showed that DnaA(R357A) cannot substitute for DnaA, indicating that the inactivation of DnaA by its AAA+ domain may be an essential process in C. crescentus, to restrict the initiation of DNA replication to only once per cell cycle. We also showed that, surprisingly, the R357A substitution in DnaA does not promote the activity of DnaA as a direct transcriptional activator of four C. crescentus genes, encoding HdaA, the GcrA master cell cycle regulator, and the FtsZ cell division protein and the MipZ spatial regulator of cell division, that are all required for normal cell cycle progression.

Results
Mutagenesis of a conserved arginine residue in the AAA+ domain of the C. crescentus DnaA protein Assuming that ATP-DnaA is the active form that initiates chromosomal replication in a bacterial species, DnaA mutants that can bind ATP but that are defective in hydrolysis of the bound ATP may cause over-initiation because the ATP-bound form of DnaA accumulates. In E. coli, such a mutant DnaA protein was constructed by a R334A amino-acid substitution [12]. The highly conserved R334 residue lays in the Box VIII sensor 2 motif of the AAA+ domain of DnaA (Fig.1). In vitro and in vivo experiments demonstrated that this mutant protein can not be inactivated by the RIDA mechanism in E. coli, and thus that its expression at moderate levels results in over-initiation of chromosomal replication, unlike the wild-type DnaA protein [12].
Our previous discovery of the HdaA protein [16], which is homologous to the E. coli Hda protein, suggested that RIDA may also occur in C. crescentus, and that this mechanism may restrict the initiation of chromosomal replication to only once per cell cycle in this particular bacterial species. To examine this possibility, we generated a mutant DnaA protein, named DnaA(R357A). The R357A substitution in the AAA+ motif of the C. crescentus DnaA protein is equivalent to the previously characterized R334A mutation in the E. coli DnaA protein (Fig.1). When expressed from the xylose-inducible xylX promoter on a medium copy number plasmid for four hours in rich medium containing xylose, we observed by immunoblot analysis that the levels of DnaA or DnaA(R357A) were approximately four-fold higher than the natural levels of DnaA expressed from its native locus in the wildtype strain (Fig.2). Hence, DnaA(R357A) was expressed at moderate levels that were comparable to the levels of the wildtype DnaA protein expressed from the same vector. We used these conditions to investigate and compare the effects of the expression of DnaA or DnaA(R357A) on cell morphology and chromosome content, and on the expression of four important genes directly activated by DnaA.
Moderate expression of DnaA(R357A), but not DnaA, blocks cell division When DnaA was expressed upon xylose induction for four hours or longer time periods from a medium-copy number plasmid in the JC366 strain, we observed by phase contrast microscopy that cells had a quite normal morphology (Fig.3). This observation shows that a four-fold over-expression of DnaA is not preventing cell division. We however still noticed that JC366 cells tended to divide at lower cell mass, giving slightly smaller daughter cells than JC919 cells containing the empty vector.
Next, we analyzed the phenotypes associated with the overexpression of the mutant DnaA(R357A) protein. Phase contrast microscopy four hours after the addition of the xylose inducer showed that JC367 cells expressing DnaA(R357A) were significantly longer than cells over-expressing DnaA and than cells containing the empty control vector (Fig.3). When cells were grown in xylose-containing medium for longer time periods, many cells became very filamentous (Fig.3), indicating a severe reduction in the efficiency of cell division upon expression of DnaA(R357A).  Expression of DnaA(R357A) leads to over-initiation of chromosomal replication We previously showed that cells slightly over-expressing DnaA from an integrated second copy of the dnaA gene under the control of the xylX promoter do not over-initiate DNA replication [16]. We confirmed this result when DnaA was expressed at higher levels from the same promoter on our medium-copy number plasmid (Fig.2). We performed flow-cytometry experiments using a fluorescent dye that labels the DNA, to observe the number of chromosomes in single cells after a rifampicin treatment. We found that only (8.7+/-2.6)% of the cells contained more than two chromosomes, which is just three-fold higher than the control strain containing the empty vector [(2.9+/-0.4)%] (Fig.4A&B). Cells not treated with rifampicin prior to the flow cytometry analysis also had a profile quite comparable to that observed for the control strain (Fig.S1). Measurements of the ratio of Cori to ter (the replication terminus region) by quantitative-polymerase chain reaction (Q-PCR) from genomic DNA extracted from a population of these cells also confirmed that these cells do not over-initiate chromosomal replication (Fig.4C). Collectively, our data demonstrates that an over-expression of DnaA from a medium-copy number plasmid is not significantly affecting the precise once-per-cell-cycle control of the initiation of chromosome replication in C. crescentus. We nevertheless noticed that cells overexpressing DnaA more often accumulated two complete copies of the chromosome after rifampicin treatment, compared to the control cells (Fig.4A&B). This observation may indicate that chromosomal replication starts a little earlier in cells overexpressing DnaA than in control cells.
To determine whether the over-expression of DnaA(R357A) affects the initiation of chromosomal replication, we also performed flow cytometry experiments with JC367 cells. In accordance with our predictions, we observed that (69.4+/ -2.9)% of the cells expressing DnaA(R357A) for four hours contained more than two chromosomes after rifampicin treatment (Fig. 4A&B). In fact, many cells even contained numerous incomplete chromosomes per cells after rifampicin treatment. When we performed flow cytometry experiments without a rifampicin treatment, we observed a profile drastically shifted towards a higher amount of DNA content per cell, compared to what we observed with the control strain or with the strain that over-expresses the wild-type DnaA protein at similar levels ( Fig.S1). To confirm that these cells over-initiated DNA replication, we also measured the Cori/ter ratio by Q-PCR from genomic DNA extracted from these cells. Consistent with our single cell flow cytometry results, we observed that the Cori/ter ratio in cells expressing DnaA(R357A) is about two fold higher than that of cells expressing DnaA as similar levels (Fig.4C). All together, these results indicate that DnaA(R357A) can initiate new rounds of chromosomal replication, but in an uncontrolled manner, suggesting that DnaA(R357A) is hyper-active for its function as an initiator of chromosomal replication. It is very likely that DnaA(R357A) remains bound to ATP, and thus active for replication initiation, at all times of the cell cycle of C. crescentus, as it is the case for the DnaA(R334A) protein in E coli.

DnaA(R357A) cannot replace DnaA(WT) in vivo
We previously demonstrated that the HdaA protein is essential for normal cell cycle progression in C. crescentus and that a DhdaA strain can not be constructed in the absence of an extra copy of the hdaA gene expressed in trans [16]. These results suggested that the switch of DnaA activity promoted by HdaA after chromosome replication has initiated could be essential for the viability of C. crescentus. If this prediction is correct, it should be very difficult, or impossible, to construct a strain carrying the dnaA(R357A) allele as the sole copy of dnaA on the C. crescentus chromosome. We attempted to construct such a strain by two different and complementary approaches.
First, we tried to construct a [DdnaA::V PxylX::dnaA(R357A)] strain by generalized transduction using bacteriophage WCR30. Two phage lysates were prepared from strain GM2471 containing the DdnaA::V mutation and strain JC125 containing a DCC1613::V . Cells were grown to exponential phase in PYE medium plus 0.2% glucose before 0.3% xylose was added for four hours. Values were normalized to a control strain, which corresponds to the wild-type strain (NA1000) grown in PYE medium and treated with rifampicin for three hours. Results are the average of data from four independent experiments and standard deviations were always lower than 0.5%. doi:10.1371/journal.pone.0026028.g004 mutation. The JC125 strain has no apparent phenotype and grows like the wild-type strain: it was used as a control for transduction efficiency. The V cassette confers resistance to spectinomycin and streptomycin. Both phage lysates were used for transduction assays using the NA1000 wild-type strain and the JC323 and JC324 strains containing a copy of the dnaA or the dnaA(R357A) gene, respectively, under the control of the xylX promoter at the xylX locus. Upon selection on spectinomycin and streptomycin containing plates, we isolated numerous DCC1613::V transductants in the three different genetic backgrounds, regardless of the presence of the xylose inducer (Table 1), demonstrating that transduction efficiency is not significantly affected by the expression of dnaA or dnaA(R357A). We also isolated numerous DdnaA::V mutant colonies from our transduction assay into JC323, but only on xylose-containing plates ( Table 1). This result demonstrated that the wild-type allele of dnaA expressed from the xylX promoter is sufficient to complement the unviable DdnaA::V mutation. In contrast, no DdnaA::V mutant colonies were isolated from the transduction assays into the wild-type strain ( Table 1), confirming that dnaA is essential for viability in C. crescentus [24]. Similarly, we were not able to isolate DdnaA::V mutant colonies from the transduction assay into the JC324 strain expressing DnaA(R357A) in the presence of the xylose inducer on PYE plates (Table 1). This result suggested that DnaA(R357A) can not replace DnaA in vivo.
Second, we tried to exchange the native dnaA allele for the dnaA(R357A) allele by double recombination using the suicide plasmid pNPTS138-DnaA(R357A) carrying the dnaA(R357A) allele. We systematically recovered the dnaA allele on the chromosome after re-excision of the plasmid (data not shown), indicating again that the dnaA(R357A) allele is probably not viable as the sole copy of dnaA on the C. crescentus chromosome.
Given that the DnaA(R357A) protein is probably functional to initiate chromosomal replication as indicated by our previous findings (Fig.4), these last genetic evidences suggest that the switch in DnaA activity mediated by its AAA+ domain is an essential process in C. crescentus and may also explain why the HdaA protein is essential for normal cell cycle progression in C. crescentus.
DnaA(R357A) is not more active than DnaA to promote the transcription of four important C. crescentus genes DnaA is not only the initiator of chromosomal replication in nearly all bacteria, but it also acts as an important transcriptional regulator by directly binding to promoters in multiple bacterial species. In B. subtilis, for example, it was estimated that DnaA directly regulates the transcription of about 50 different genes [4], while DnaA directly regulates the transcription of minimum 13 genes in C. crescentus [3,16,25]. One of the still outstanding questions regarding DnaA activity as a transcription factor is whether the nucleotide bound to DnaA often influences its binding and activity at DnaA-regulated promoters in diverse bacterial species [2].
We investigated whether the DnaA(R357A) protein may also be hyper-active to promote transcription from four well-characterized DnaA-activated promoters in C. crescentus. These are: (i) the hdaA promoter, controlling the expression of the HdaA repressor of the initiation of chromosomal replication [16]; (ii) the gcrA promoter, controlling the expression of the GcrA master regulator of the C. crescentus cell cycle [3,25,26]; (iii) the ftsZ promoter, controlling the expression of the FtsZ cell division protein [3,27,28]; (iv) the mipZ promoter, controlling the expression of the MipZ spatial regulator of cell division [3,29]. We chose these four promoters because they have a quite different structure with respect to the position and the number of DnaA boxes that they contain. Indeed, the gcrA and mipZ promoters contain only one DnaA box, the ftsZ promoter contains two DnaA boxes, and the hdaA promoter contains up to six DnaA boxes [16]. In addition, the four genes that we selected are essential or required for normal cell cycle progression in C. crescentus.
We first compared by immunoblot analysis the amounts of GcrA and HdaA that accumulated in cell extracts from the wildtype strain and from the strain that over-expresses DnaA. We found that GcrA and HdaA accumulated at higher levels in DnaA over-expressing cells than in wild-type cells, although the effect was more pronounced for HdaA than for GcrA (Fig.2). HdaA accumulated at similar levels in cells expressing DnaA(R357A) compared to cells expressing DnaA (Fig.2). In contrast, GcrA accumulated at lower levels in cells constitutively expressing DnaA(R357A) compared to cells expressing DnaA (Fig.2). Altogether, these observations suggested that DnaA(R357A) is not more active than DnaA to promote the transcription of at least certain genes that belong to the direct DnaA regulon.
To confirm that the activity of DnaA as a transcriptional regulator of our four selected genes is not increased by the R357A mutation, we introduced a low copy number plasmid carrying a transcriptional fusion between the hdaA promoter, the gcrA promoter, the ftsZ promoter or the mipZ promoter and the lacZ gene [16,25,26], into the strains that over-express DnaA or DnaA(R357A). We measured b-galactosidase activities as an indication of promoter activities (Fig.5). Four hours after xylose addition into the medium, we observed that b-galactosidase activities from each promoter that we tested were higher in cells over-expressing DnaA than in wild-type cells. Consistent with previous data [16,25], these results confirm that DnaA activates the transcription of all four genes. Next, we compared the effect of the over-expression of DnaA or DnaA(R357A) on the activities of these four promoters. Surprisingly, we observed that the activation of the transcription from these four promoters was never higher when DnaA(R357A) rather than DnaA was over-expressed (Fig.5).
These interesting results indicate that DnaA(R357A) is not more efficient than DnaA to activate the transcription of these four important genes. Thus, the R357A mutation in DnaA uncouples the ability of DnaA to initiate DNA replication from its activity as a transcription factor regulating minimum four genes, only leading to increased activity in the initiation of DNA replication. In addition, we observed that DnaA(R357A) is even less efficient than DnaA to activate the transcription of gcrA, ftsZ and mipZ (Fig.5B&C&D), suggesting that the switch in the activity of DnaA that takes place at the time when DNA replication is initiated, may promote the expression of these genes.

Discussion
In this study, we showed that the DnaA(R357A) mutant protein in C. crescentus retains its ability to promote the initiation of chromosomal replication in vivo, and is even hyper-active as an initiator compared to the wild-type DnaA protein, as indicated by the severe over-replication phenotype of cells that over-express DnaA(R357A) (Fig.4). In addition, we showed that the DnaA(R357A) protein cannot replace DnaA ( Table 1), suggesting that the inactivation of the initiator DnaA is an essential process in C. crescentus. In contrast, we observed that the activity of DnaA(R357A) as a transcription factor that stimulates the transcription of four genes is not higher than that of DnaA (Fig.5), indicating that the AAA+ domain of DnaA may not inactivate DnaA as a transcriptional regulator of these genes in C. crescentus. Below, we discuss the role of the AAA+ domain of DnaA in the regulation of both activities of DnaA in the control of the C. crescentus cell cycle.

Model for a RIDA system in C. crescentus
The R357A substitution in the AAA+ motif of the C. crescentus DnaA protein is equivalent to the previously characterized R334A mutation in the E. coli DnaA protein that inhibits RIDA and the intrinsic ATPase activity of DnaA in vivo and in vitro [12,30]. It is thus likely that the R357 residue in the AAA+ domain of the C. crescentus DnaA protein participates in the hydrolysis of an ATP bound to DnaA, to inactivate DnaA immediately following the initiation of chromosome replication (Fig.6). Consistent with this model, the C. crescentus DnaA(R357A) protein would be bound to ATP at all times of the cell cycle, as it is the case for the E. coli DnaA(R334A) protein. Then, DnaA-ATP would initiate chromosome replication whenever and wherever active CtrA is absent, leading to C. crescentus cells that have undergone additional chromosome replication initiations like we observed (Fig.4). The C. crescentus HdaA protein may be a functional homolog of the E. coli Hda protein, stimulating the ATPase activity of the AAA+ domain of DnaA when bound to the replisome. In agreement with this model, we observed that the phenotype of HdaA-depleted cells [16] resembles that of DnaA(R357A) over-expressing cells (Fig. 3  and 4), and that the HdaA protein dynamically co-localizes with the replisome [16]. Our results suggest that the inactivation of the initiator DnaA by the hydrolysis of the ATP bound to DnaA by its AAA+ domain is an essential process in C. crescentus, as we were not able to replace the wild-type dnaA allele by the mutant dnaA(R357A) allele on the C. crescentus chromosome (Table 1 and data not shown). This could then explain why the HdaA protein is also essential for normal cell cycle progression in C. crescentus [16]. When DnaA(R357A) was expressed together with DnaA in strains such as JC367 or JC324, DnaA was probably competing with DnaA(R357A) when binding the Cori prior to the initiation of chromosomal replication, thereby maintaining cells alive by the inactivation of the wild-type subset of the multiple DnaA molecules bound to the Cori after the initiation of chromosomal replication.
The intracellular levels of DnaA fluctuate during the C. crescentus cell cycle, being the most abundant at the time when chromosomal replication is initiated during the swarmer-to-stalked cell transition [25,31]. Both the transcription of the dnaA gene and the proteolysis of the DnaA protein are temporally regulated, to ensure that DnaA accumulates the most at the right time of the cell cycle [31,32,33,34]. Our data suggest that the proteolysis of DnaA in stalked cells is not sufficient to inhibit the initiation of chromosomal replication after the first round of replication has started, and that the activity of DnaA also needs to be downregulated at that moment of the cell cycle for normal cell cycle progression. Consistent with this proposal, we found that the constitutive expression of a DnaA(R357A) mutant protein resulted in severe over-initiations of chromosomal replication (Fig.4), while the constitutive expression of DnaA did not. Jonas et al. recently reported that the over-production of an M2-tagged DnaA protein resulted in over-initiation of chromosomal replication in C. crescentus [35], unlike what we observed when we over-produced the native DnaA protein. This difference may be attributable to the artificial addition of the M2 epitope to the DnaA protein by Jonas et al., or to differences in growth conditions that may promote DnaA accumulation at higher levels.
CtrA and the RIDA system have complementary functions in the control of the initiation of chromosomal replication in C. crescentus The CtrA master regulator is conserved in alpha-proteobacteria but is not found in entero-bacteria like E. coli. In C. crescentus, it acts as an inhibitor of the initiation of chromosomal replication that is present and active in swarmer cells and in the swarmer compartment of pre-divisional cells (Fig.6B). Thus, CtrA essentially inhibits replication by binding to the origin of nonreplicating chromosomes [19,22,36]. Accordingly, we observed that (11.3+/-1.6)% of the cells expressing DnaA(R357A) still contained only one chromosome by flow cytometry analysis after rifampicin treatment (Fig.3). This observation suggests that CtrA is still able to inhibit the initiation of DNA replication in the presence of the hyper-active DnaA(R357A) mutant protein in swarmer cells. HdaA protein co-localizes with the replisome assembled at the replication forks once DNA replication is ongoing [16]. In E. coli, Hda needs to interact with the active replisome to stimulate the AAA+ domain of DnaA and thereby promote the inactivation of DnaA [9]. By analogy, we propose that the C. crescentus RIDA system inactivates DnaA as soon as chromosomal replication has started, and maybe until the end of the replication process (Fig.6B). This coincides with the period of the cell cycle when active CtrA is poorly abundant or absent, suggesting that RIDA and the activity of CtrA have complementary roles in controlling the timing of the initiation of chromosomal replication. We propose that active CtrA determines when the initiation of chromosomal replication can start, while the RIDA system ensures that chromosomal replication starts only once per cell cycle in C. crescentus. Is the RIDA system involved in the transcriptional control of DnaA-regulated genes in C. crescentus?
Besides its function as the initiator of chromosomal replication, DnaA is also a master transcriptional regulator in C. crescentus [3]. DnaA activates the transcription of about forty different genes at the swarmer-to-stalked cell transition, by directly binding to minimum thirteen different promoters. Genes whose expression is directly regulated by DnaA encode proteins involved in multiple processes essential for cell cycle progression. Examples are the FtsZ protein required for cell division, the NrdB protein involved in nucleotide synthesis, and the cell polarity factor PodJ.
To explore the potential role of the AAA+ domain of DnaA in the regulation of the activity of DnaA as a transcription factor, we compared the effect of the over-expression of DnaA or DnaA(R357A) on the activities of several promoters driving the expression of proteins involved in the RIDA system, in cell cycle regulation and in cell division in C. crescentus. Interestingly, the selected genes are not regulated by DnaA in E. coli. We found that DnaA(R357A) is not more active than DnaA to activate the transcription of each selected genes (Fig.5), suggesting that DnaA-ATP is not more active than DnaA-ADP when acting as a transcription factor regulating the expression of these genes (Fig.6). Thus, the DnaA mutant we generated decouples the activity of DnaA as a transcription factor regulating several genes from its activity as an initiator of DNA replication.
We also observed that DnaA(R357A) was less efficient than DnaA in activating gcrA, ftsZ and gcrA transcription (Fig.5B&C&D). This effect is particularly striking for the gcrA and mipZ promoters, which are not significantly activated by the expression of DnaA(R357A), compared to the control strain containing an empty vector (Fig.5B). One possible interpretation is that these three promoters may be more efficiently activated by DnaA-ADP than by DnaA-ATP (Fig.6). Genes specifically activated by DnaA-ADP may then wait for the initiation of DNA replication to be activated by DnaA, thereby coordinating the initiation of DNA replication with the expression of specific genes. Consistently, the transcription of gcrA is at its maximum in stalked cells, right after the initiation of DNA replication [25,26]. Also, GcrA activates the transcription of genes whose products are involved in the elongation of DNA replication and in chromosome segregation [26]. Thus, it may be logical for these genetic modules to be expressed only after the initiation of DNA replication. We propose that the AAA+ domain of DnaA plays a role in temporally regulating the bifunctionality of DnaA by reallocating DnaA molecules from initiating DNA replication to transcribing genes within the DnaA regulon.
The extent to which the regulation of the activity of DnaA may affect the regulation of other genes directly controlled by DnaA remains unkown in C. crescentus and future detailed investigations studying the regulation of each gene will be required to determine in which cases the AAA+ domain of DnaA may regulate the activity of DnaA as a transcription factor. In other bacterial species, there are now several examples of DnaA-regulated genes whose expression was shown to depend on both the concentration and the nucleotide-bound state of DnaA [2,5,37,38], but the role of the nucleotide bound to DnaA in the regulation of the activity of DnaA as a transcription factor still remains poorly understood in most cases [2]. Thus, the full extend to which DnaA is utilized to regulate the timing of gene expression during the bacterial cell cycle is an interesting avenue for future research and C. crescentus is an ideal model to study such questions.

Materials and Methods
Bacterial strains and growth conditions C. crescentus strains were grown in PYE complex media [39] at 28uC. Plasmids and strains used are listed in Table 2 and  Table 3. Antibiotics used for C. crescentus liquid cultures include rifampicin (15 mg/mL), chloramphenicol (1 mg/mL), kanamycin (5 mg/mL) and oxytetracycline (1 mg/mL). Antibiotics used for E. coli liquid cultures include chloramphenicol (20 mg/mL), kanamycin (30 mg/mL) and oxytetracycline (12 mg/mL). Plasmids were mobilized from E. coli S17-1 [40] into C. crescentus by bacterial conjugation or introduced by electro-transformation. Bacteriophage WCR30 was used for general transduction into C. crescentus.
Construction of the pX-DnaA(R357A) plasmid. pDnaA (R357A) was digested by NdeI and BamHI to extract the mutant dnaA insert, which was cloned into NdeI-BamHI-digested pXGFP4C1 (with the gfp gene eliminated).
Construction of the pJSX-DnaA(R357A) plasmid. same as construction of pJSX-DnaA, except that pX-DnaA(R357A) was used as template DNA for the PCR amplification.

Promoter activity assays
The b-galactosidase activity of strains containing pLacZ290 derivatives was assayed from exponential phase cultures in PYE media, as previously described [44]. b-galactosidase activities represent the average of minimum three independent assays. Figure S1 A moderate overproduction of DnaA(R357A), but not DnaA, leads to severe over-initiation of chromosomal replication. Representative profiles obtained by flow cytometry analyses of cells from strains JC919 (containing the pJS14 empty vector), JC366 (containing pJSX-DnaA) and JC367

Supporting Information
[containing pJSX-DnaA(R357A)] grown to exponential phase in PYE medium plus 0.2% glucose (PYEG) before 0.3% xylose was added (PYEGX) to half of the culture for four hours. Cells were fixed and stained with Vybrant DyeCycle orange. The horizontal axis indicates the number N of complete chromosomes: 1N, 2N or more than 2N (+). The vertical axis indicates the number of cells. (DOC)