INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial chromosome replication starts at a fixed site (oriC) and at a specific cell mass (the initiation mass) in the cell division cycle. In early studies, the mass per oriC at initiation was proposed to be constant, irrespective of the growth rate (8). However, a later report suggested that the initiation mass increased by about 50% monotonically with decreasing growth rate (from 2.5 to 0.3 doublings per hour) (45). Nonetheless, in a given culture, initiation occurred within a very narrow range of cell mass, indicating the tight coupling of replication initiation to cell mass (4). Two major regulatory mechanisms are assumed to ensure the fidelity of replication initiation. One is positive in that it triggers initiation at the appropriate initiation mass, while the other is negative in that it suppresses extra initiations until the next round of initiation. The DnaA protein concentration (14) and change in DnaA activity (26) have been proposed to embody such positive factors in Escherichia coli. In contrast, sequestration of newly replicated origins into the cell membrane by SeqA (29) and inactivation of DnaA by DnaN (20) may represent the negative mechanisms present in E. coli. In addition to sequestration, SeqA also serves as a negative modulator of replication initiation because the initiation mass is decreased in the null mutant (4). Thus, another idea was proposed to determine the timing of replication initiation, in which replication initiation is controlled by changes in the balance between positive and negative factors (29). However, the molecular mechanism regulating the balance remains to be uncovered. For Bacillus subtilis, there have been, as yet, no reports about such positive and negative mechanisms. No homologous genes to seqA have been found on the genome (25), suggesting that a mechanism similar to that of SeqA may not be operating in B. subtilis.

The DnaA protein is required for initiation of replication in both E. coli and B. subtilis (31, 33) and is conserved in many bacteria (47). Overproduction of DnaA leads to earlier initiation of replication in the cell division cycle in E. coli, indicating that the DnaA protein level was the primary factor that determines the timing of initiation (27). Later, it was reported that its overproduction also interfered severely with replication elongation, although the reason was not clear (3). In support of the fact that the DnaA level is a primary positive factor, initiation of replication from oriC occurred synchronously at a threshold level of DnaA, albeit under special conditions (11). To explain how the DnaA level controls the timing of initiation, the initiator titration model was proposed (10). According to this model, newly synthesized DnaA molecules first bind to high-affinity DnaA boxes dispersed on the chromosome, and, subsequently, the remaining free DnaA molecules bind to a lower-affinity DnaA box within oriC that triggers initiation. In fact, there are several regions with high-affinity DnaA boxes on the E. coli chromosome (39). Furthermore, when one such region (datA) (21) was deleted from the chromosome, overinitiation and asynchronous initiation were observed (22). Thus, in E. coli, the DnaA protein level appears to play a pivotal role in determining the timing of replication initiation.

In the present studies, a relationship between the timing of initiation and cellular DnaA protein levels was detected, suggesting that such levels could also act as a positive regulator of initiation in B. subtilis. In addition, overproduction of DnaA alone induced the SOS response (46) in this organism. This induction was probably due to depletion of the DnaN protein ( subunit of DNA polymerase III) caused by repression of the native dnaA-dnaN operon at high DnaA levels. Indeed, when DnaN was supplied from another locus on the chromosome, cell elongation caused by the SOS response was not observed at high DnaA levels. The supply of DnaN enabled us to examine the effects of high DnaA levels on replication initiation by flow cytometry.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and plasmids. The B. subtilis strains used in this study are listed in Table 1. Construction of strains NIS2000, NIS2020, and NIS2022 is described below. Transformation of B. subtilis cells was carried out as described previously (24). E. coli strain C600 was used throughout as a cloning host.

NIS2000. Pspac-lacI and cat fragments were amplified from pDH88 DNA (13) by PCR with two sets of primers (Table 2), 1 and 2 and 3 and 4, respectively, and were purified with Centricon 100 (Amicon, Beverly, Mass.). Since a 37-nucleotide sequence of the 5' end of primer 3 is complementary to the sequence of primer 2, these two PCR products can be combined by PCR as described previously (15). Indeed, by utilizing the overlapping sequence, the two fragments (each 5 to 10 ng) were combined, and the resulting fragment (Pspac-lacI-cat) was further amplified by PCR (25 cycles) with two outer primers, 1 and 4. Next, two fragments of about 7 kb near the purA locus, from yycA to purA and from trnY to yyxA, were amplified from the chromosomal DNA (50 ng) by long PCR (25 cycles of 15 s at 95°C [denaturation] and 6 min at 65°C [annealing and extension]) with rTth DNA polymerase XL (PE Applied Biosystems) and the following primer sets: primers 5 and 6 for yycA~purA and primers 7 and 8 for trnY~yyxA. Three PCR products (yycA~purA, Pspac-lacI-cat, and trnY~yyxA) were purified with Centricon 100. To combine these fragments by PCR, 22- and 28-nucleotide sequences of primers 6 (7th to 28th bases from the 5' end) and 7 (1st to 28th bases), which are complementary to those of primers 1 (1st to 22nd bases) and 4 (1st to 28th bases), respectively, were used as shown in Table 2. As a result, a long DNA fragment, yycA~purA-Pspac-lacI-cat-trnY~yyxA, was amplified by long PCR (20 cycles of 15 s at 95°C and 12 min at 65°C, followed by a further 12 cycles, increasing the extension time by 15 s per cycle) with the outer two primers 5 and 8 and was used to transform CRK6000 cells after being purified with Centricon 100. 

NIS2020. A promoter region of the rpmH gene (nucleotides 4214716 to 4214638) was amplified from the chromosome by PCR with primers 9 and 10 (Table 2). The lac operator sequence is included in primer 10 (24th to 42nd bases from the 5' end). Because a 32-nucleotide sequence of primer 9 (1st to 32nd bases) is complementary to that of primer 6 at the 3' end, this PrpmH-lacO fragment was combined with a long DNA from yycA to purA by long PCR with primers 5 and 10, resulting in yycA~purA-PrpmH-lacO. In addition, the lacI-cat-trnY~yyxA fragment was amplified from the NIS2000 chromosome by long PCR with primers 11 and 8. Because two primers, 10 and 11, contain artificial BamHI sites (Table 2), the two PCR fragments were ligated after digestion with BamHI and used to transform CRK6000. Again, yycA~PrpmH-lacO and lacI-cat~yyxA fragments were amplified from chromosomal DNA of the transformant (NIS2010) by long PCR with the primer sets 5 and 12 for the former and 13 and 8 for the latter. The coding region of dnaA with the SD sequence (nucleotides 315 to 1754) was also amplified from the CRK6000 chromosome by PCR with primers 14 and 15 (Table 2). By utilizing overlapping sequences between primers 12 and 14 (23 bases) and between primers 13 and 15 (43 bases) (Table 2), three PCR fragments (yycA~PrpmH-lacO, dnaA, and lacI-cat~yyxA) were combined and amplified by long PCR with the outer two primers 5 and 8 as described above. The resulting fragment, yycA~purA-PrpmH-lacO-dnaA-lacI-cat-trnY~yyxA, was used as a donor DNA for transformation of CRK6000 after being purified with Centricon 100. The nucleotide sequences of PrpmH-lacO-dnaA regions in five chloramphenicol-resistant transformants were determined, and one of them (NIS2020) contained no mutations in the region.

NIS2022. cat located downstream of lacI in NIS2021 was replaced by neo with pCm::Nm (from BGSC) (44) by transformation, and then the neomycin-resistant purA region containing an additional dnaA-dnaN was transferred to NIS6311 (17) by transformation. Among neomycin-resistant transformants, chloramphenicol-sensitive ones were selected, in which oriN and cat integrated at spoIIIJ of NIS6311 (Table 1) were replaced with the native spoIIIJ of NIS2021 by cotransformation. Finally, the existence of the ochre mutations in dnaA (34) was confirmed by sequencing.

pSM5100. Both the dnaA and dnaN genes (nucleotides 358 to 1759 and 1913 to 3077, respectively, at http://bacillus.genome.ad.jp/) were amplified by PCR with primer sets by using artificial SpeI-BglII and BglII-BamHI sites, respectively. After digestion of these fragments with the restriction enzymes, they were ligated at the BglII sites and cloned between the XbaI and BglII sites of a pSM5000 derivative that contained a fragment (from EcoRI to BamHI) of pDH88 (13) at the EcoRI site of pSM5000 (34). Both XbaI and BglII sites are located in a multicloning site in the pDH88 portion. Subsequently, approximately the 5' half of dnaA and a part of the pDH88 portion were removed by digestion with SalI followed by self-circularization. Finally, the other part of the pDH88 portion (from ClaI to ScaI) located downstream of dnaN was replaced with a BamHI fragment conferring tetracycline resistance in pBEST307 (18) after blunting both DNA fragments. Because bla was destroyed by ScaI digestion, E. coli clones were selected by tetracycline resistance (20 µg/ml).

Growth conditions. B. subtilis cells were grown in antibiotic medium 3 (Difco Laboratories, Detroit, Mich.) supplemented with adenine and guanosine (requirements for growth; 20 µg/ml) at 30°C throughout this study. When necessary, various concentrations of isopropyl--D-thiogalactopyranoside (IPTG) and drugs were added (chloramphenicol, erythromycin, and tetracycline at 10, 1, and 5 µg/ml, respectively). Because addition of drugs decreased the growth rate slightly, cells were grown without drugs for flow cytometry.

Immunoblotting. Preparation of cell lysates from exponentially growing cells and separation of proteins were carried out as described previously (12). Proteins were blotted on a Hybond-P polyvinylidene difluoride membrane (Amersham Pharmacia Biotech, Piscataway, N.J.), and the membrane was incubated with anti-DnaA rabbit polyclonal antibody (33). After the membrane was treated with a second antibody (goat anti-rabbit immunoglobulin G-horseradish peroxidase conjugate), signals were detected by the ECL-Plus enhanced chemiluminescence system (Amersham).

Examination of the SOS response. The SOS response of cells was examined by monitoring expression from a damage-inducible promoter (dinC) that was fused with a reporter gene, bgaB, encoding Bacillus stearothermophilus -galactosidase at the amyE locus on the chromosome. To measure the activity of the -galactosidase expressed, samples were prepared from cells as described previously (48). The -galactosidase assay was carried out with the fluorogenic substrate MUG (4-methylumbelliferyl--D-galactoside) and a Fluoroskan II fluorometer (Labsystems, Helsinki, Finland). One unit of specific activity was defined as 1 pmol of MUG hydrolyzed/ml of culture sample/min, normalized for culture cell density at 600 nm.

Northern hybridization. Total RNA was extracted from B. subtilis cells basically as described previously (16). Five micrograms of total RNA in each lane was separated in a denaturing agarose (1%) gel containing formaldehyde (2%), followed by blotting on a nylon membrane, Hybond-N⁺ (Amersham). The presence of nearly equal amounts of RNA in each lane was confirmed by staining with methylene blue. Preparation of RNA probes and the following Northern hybridization method were carried out as described previously (1). Briefly, RNA probes were prepared as follows. First, coding regions of the dnaA and dnaN genes with lengths of 257 bp (nucleotides 829 to 1085 at http://bacillus.genome.ad.jp/) and 266 bp (nucleotides 1958 to 2223), respectively, were amplified by PCR. The promoter sequence for T7 RNA polymerase was added to reverse primers. RNA probes labeled with digoxigenin (DIG) were obtained by in vitro transcription by using these PCR fragments as a template and the DIG RNA labeling kit (SP6/T7) (Roche Molecular Biochemicals, Mannheim, Germany). Hybridization and detection of signals were carried out with the DIG luminescent detection kit (Roche) according to the supplier's manual.

Flow cytometry. Chloramphenicol was added to cells grown exponentially at a concentration of 200 µg/ml to inhibit new rounds of initiation. Incubation was continued for a further 5 h to complete ongoing replication. A previous paper reported that this drug inhibited replication initiation and cell division simultaneously in B. subtilis cells (40). In fact, the cell length at initiation estimated by flow cytometry corresponded approximately to that obtained previously by an extensive microscopic analysis (41) (data not shown). For flow cytometry, the cells were fixed with ethanol and treated further as described previously (27), and the number of replication origins per cell was measured with a Bryte HS flow cytometer (Bio-Rad Laboratories, Hercules, Calif.).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Overproduction of the DnaA protein induces the SOS response. A copy of the dnaA gene was placed under the control of a fusion between the rpmH promoter (gene coding for ribosomal protein L34) and the lac operator. It was then integrated into the purA locus on the chromosome with lacI and cat genes (Fig. 1A). Upon addition of IPTG, the DnaA level increased, as shown by immunoblotting (Fig. 1B). However, cells could not form colonies in the presence of 50 µM IPTG, although they grew in liquid medium for a short period (which permitted us to measure DnaA levels under this condition). At 100 µM or greater concentrations of IPTG, cells could not grow at all in the liquid medium.

View larger version (24K): [in this window] [in a new window]   FIG. 1.   SOS induction caused by overproduction of the DnaA protein. (A) Schematic presentation of the relevant structure of a DnaA-overproducing strain (NIS2020). Open arrows indicate the location and direction of primers used at the final stage of its construction. From left to right are shown primers 5, 12, 14, 15, 13, and 8, respectively (Table 2). Briefly, yycA~PrpmH-lacO and lacI-cat~yyxA DNA fragments were amplified from the NIS2010 chromosome by long PCR with two sets of primers: primers 5 and 12 and 13 and 8, respectively. The dnaA gene was amplified by PCR with primers 14 and 15. After the three PCR products were purified with Centricon 100, they were combined and amplified by long PCR with two outer primers, 5 and 8, utilizing overlapping sequences between primers 12 and 14 and 15 and

Transcription of the dnaA-dnaN operon is regulated by DnaA. Because the promoter of the dnaA-dnaN operon is surrounded by DnaA boxes (32), we suspected that overproduction of DnaA might repress the expression of dnaA-dnaN and cause depletion of DnaN ( subunit of DNA polymerase III), which could also explain why replication elongation was inhibited by overproduction of DnaA. To test this possibility, transcription of the operon in the dnaA-null (NIS6311) and DnaA-overproducing (NIS2020) strains was examined by Northern hybridization. In the null mutant, initiation of chromosome replication was dependent on a plasmid (mini-pLS32) integrated into the chromosome (12). Moreover, because two ochre mutations were introduced at the 5th and 10th codons of dnaA, the length of the dnaA transcript would not be affected by these mutations.

View larger version (35K): [in this window] [in a new window]   FIG. 2.   Analysis of transcripts from the dnaA-dnaN operon in the dnaA-null mutant and the DnaA-overproducing strains by Northern hybridization. (A) Transcripts detected by a dnaA probe. (B) Transcripts detected by a dnaN probe. Lanes: 1, wild-type strain (CRK6000); 2, dnaA-null mutant (NIS6311) (17); 3 to 6, DnaA-overproducing strain (NIS2020) cultivated with no IPTG or with 5, 10, and 20 µM IPTG, respectively. Arrows indicate positions of size markers. Asterisks indicate the major transcripts detected. Faint bands (1.6 kb) detected in all lanes in panel A are probably generated by cross-hybridization with 16S rRNA. (C) Schematic presentation of structures of the native dnaA-dnaN operon (a) and the additional dnaA region in the DnaA-overproducing strain (b) and length of transcripts predicted in these regions. Bent arrows indicate the positions of promoters and the direction of transcription. Thick bars represent probable regions of transcription. Probes used to detect transcripts are shown as gray boxes.

The DnaA-overproducing strain recovers from its growth defects by trans-supply of DnaN from another locus on the chromosome. As expected, transcription of dnaN was largely reduced in the DnaA-overproducing strain in the presence of 20 µM IPTG (Fig. 2B, lane 6). This result suggests strongly that the growth defects observed in this strain in the presence of high concentrations of IPTG are due to a deficiency of DnaN. If this is the case, then the supply of DnaN from another locus on the chromosome could be expected to restore growth in the presence of high concentrations of IPTG. To check this possibility, a copy of the dnaN gene was added at the purA locus of the DnaA-overproducing strain (Fig. 3A).

View larger version (13K): [in this window] [in a new window]   FIG. 3.   4',6'-Diamidino-2-phenylindole (DAPI)-stained image of NIS2021 cells overproducing DnaA and DnaN. (A) Schematic presentation of relevant structure of NIS2021. Plasmid pSM5100, described in Materials and Methods, was integrated into the NIS2020 chromosome by single crossover with the 3'-half of dnaA as a homologous region for the recombination. The tetracycline-resistant NIS2021 clones were selected in the presence of 50 µM IPTG, which inhibited the cell growth of NIS2020. Proper integration of the plasmid into this purA region, but not into the native dnaA-dnaN operon, was confirmed by PCR. A broken line means that two regions are apart on the chromosome. (B) NIS2021 cells stained with DAPI. Cells grown exponentially in the presence of 20 µM IPTG are shown. The scale bar represents 10 µm.

The DnaA protein level influences the initiation mass. Since the problem of inhibition of cell growth by overproduction of DnaA was solved by adding a copy of dnaN at the purA locus, the effects of various DnaA levels on replication initiation were examined by using the DnaA-DnaN-overproducing strain NIS2021. First, a dnaA-null mutation was introduced into the native dnaA gene of the strain to confirm that the IPTG-inducible dnaA gene at purA is active for replication initiation (Fig. 4A). The NIS2022 cells grew even in the absence of IPTG, probably due to leakiness of the inducible rpmH-lacO promoter. In fact, immunoblotting detected DnaA, but it was reduced to about 20% of the wild-type level (Fig. 4B). The effects of the reduced level on replication initiation were analyzed by flow cytometry. Although B. subtilis cells generally tend to form chains, which can make flow cytometry difficult to interpret, our parent strain (CRK6000) does not form chains under the conditions used for this work. Therefore, flow cytometry was used to determine how many replication origins are present in individual cells (Fig. 4D). The results showed that such cells contained two, four, or eight replication origins under our cultivation conditions. However, most of the cells contained four origins, indicating that initiation occurred at a late stage of the cell division cycle (Fig. 4D). When NIS2022 cells, derivatives of CRK6000, were grown in the absence of IPTG, there were mainly three types of cells having two, three, or four origins detected by flow cytometry (Fig. 4C). Detection of an odd number of origins within the cell indicated that asynchronous initiation was occurring despite the fact that both the growth rate and distribution of the cell size were similar to those of the parent cells (Table 3). Due to the asynchrony of initiation, it was difficult to determine the initiation mass of these cells. Instead, the average mass/origin ratio (proportional to the initiation mass) was calculated from the average number of origins per cell determined by flow cytometry (Fig. 4C), and the average cell mass was also measured by flow cytometry (Fig. 4D and Table 3). As a result, it was increased by about 30% compared to that of wild-type cells (Table 3), suggesting that replication initiation is delayed in the cells containing a low level of DnaA (about 20% of the wild-type level). When 5 or 10 µM IPTG was added, the DnaA levels were almost comparable to wild-type levels (about 80 or 130%, respectively) (Fig. 4B), and DNA histograms obtained by flow cytometry showed patterns similar to those of the wild-type cells (Fig. 4C and D). Since asynchronous initiation was not apparently observed in these cases, the initiation mass (cell mass per replication origin at initiation) could be calculated as described in the legend to Fig. 4D. It was nearly identical in each case to that of the wild-type cells (Table 3). These results indicate not only that DnaA produced from the additional dnaA gene is active for replication initiation, but also that wild-type DnaA levels are required for initiation of replication at the proper cell mass.

View larger version (30K): [in this window] [in a new window]   FIG. 4.   Effects of overproduction of DnaA protein on initiation of chromosome replication. (A) Schematic presentation of relevant structure in a test strain (NIS2022). A thick vertical bar represents the two ochre mutations near the 5' end of the native dnaA gene (a thick gray arrow) in NIS2022. (B) Relative DnaA protein levels in NIS2022 cells grown in various concentrations of IPTG as determined by immunoblotting. W, wild type (CRK6000). (C) Flow cytometry of cells grown in the presence of various concentrations of IPTG. DNA histograms of cells after incubation for 5 h in the presence of chloramphenicol are shown. (D) Concept for estimating initiation mass. The wild-type cell culture used in this study is shown as an example. (Left) DNA histogram of cells after a 5-h incubation with the drug. (Right) Light scatter histogram of log-phase cells (at the time of addition of the drug). As determined (42), the genome equivalents in the DNA histogram represent the number of replication origins present at the time of addition of the drug to individual cells. In this case, replication had not initiated in cells having four origins (black area in this figure), but had initiated in cells having eight origins (white area). The cell mass at initiation (Mi) is estimated by dividing the light scatter distribution into two portions (not initiated and initiated) according to the ratio of the number of the two types of cells (black/white). The initiation mass is obtained by dividing this mass by the number of origins present at the time of initiation (in this case, Mi/4).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

In this report, we have shown that dnaA and dnaN constitute an operon in B. subtilis and that it is autoregulated by DnaA. Autoregulation of dnaA is also observed in other bacteria. In E. coli, transcription of chromosomal dnaA was repressed by overproduction of DnaA and derepressed by inactivation of the protein in temperature-sensitive dnaA mutants (2, 5, 23). This repression was due to binding of the ATP-bound form of DnaA (not ADP-DnaA) to DnaA boxes in the promoter region (43). Recently, in Streptomyces, deletion of DnaA boxes in the promoter region of dnaA increased its transcription, suggesting autoregulation (19). In B. subtilis, there are eight DnaA boxes, including ones differing in 1 base from the consensus sequence, in the promoter region of dnaA. Binding of DnaA to these boxes was demonstrated by DNase I footprinting (9). Two consensus DnaA boxes facing each other and located 40 to 50 bp downstream of the transcription start site showed the highest binding affinity in the footprinting analysis. Deletion of the two boxes also abrogated the strong incompatibility phenotype that this promoter region exerts when cloned on multicopy plasmids, probably due to titration of DnaA (32). These results suggest that the two consensus DnaA boxes play a major role in autorepression of B. subtilis dnaA.

Although dnaA is autoregulated in both E. coli and B. subtilis, an additional regulation mechanism seems to overlap the promoter region of B. subtilis dnaA. In E. coli, addition of extra DnaA boxes derepresses transcription of dnaA by titrating DnaA (23). However, in B. subtilis, similar treatment did not derepress the dnaA-dnaN operon (36), indicating the existence of a more restrictive control in expression of the B. subtilis dnaA operon. In this context, it should be noted that transcription of the operon was inhibited when replication initiation was blocked by using dnaB mutants at the nonpermissive temperature and that the inhibition was released after initiation resumed (38). These results may suggest that transcription of the operon occurs only after replication initiation and then ceases by autoregulation with newly synthesized DnaA. Once repressed, it stays repressed by an unknown mechanism or mechanisms even in the presence of extra DnaA boxes, as described above, until the next round of replication initiation. This model of dnaA expression also suggests that the DnaA protein required for the next round of replication initiation would be available soon after the present round of initiation is completed. Consistent with this assumption, initiation occurred once even in the absence of protein synthesis during resumption of replication in the temperature-sensitive dnaB27 mutant by temperature downshift (37), despite the fact that transcription from the dnaA promoter was completely blocked at the nonpermissive temperature (38). Because transcription of dnaA is repressed after replication initiation by sequestration in E. coli (6), expression patterns of dnaA in the cell division cycle may be very different between B. subtilis and E. coli.

In E. coli, several models have been proposed concerning regulation of replication initiation. The initiator titration model (10) proposes that the initiator DnaA is titrated by binding to many DnaA boxes dispersed on the chromosome, and after their saturation, free DnaA molecules start to bind to a lower-affinity box within oriC to trigger initiation. In other words, the balance between the DnaA level and the number of DnaA boxes within the cell is the critical factor for timing of replication initiation (14). In fact, overproduction of DnaA leads to excess and earlier initiation of replication (3, 27), and addition of extra DnaA boxes delays initiation (7). Furthermore, deletion of a chromosomal site (datA) that titrates DnaA strongly (21) results in overinitiation (22). These observations support the initiator titration model. The second model is that replication initiation is controlled by changes in the balance between positive and negative factors (29). DnaA and SeqA act as positive and negative regulators, respectively (4). The authors postulate that all factors required for initiation are present at the origin at an early time, but SeqA negatively affects the initiation process at this stage. The third one is based on changes in DnaA activity. Only the ATP-bound form of DnaA is active for initiation, the levels of which increase sharply before replication initiation in the cell division cycle (26, 30), suggesting that the change in nucleotide forms of DnaA plays an important role in timing of replication initiation. Despite these various models, it remains uncertain which of the mechanisms proposed is the primary determinant for timing of replication initiation in E. coli. In B. subtilis, we examined the effects of the DnaA level on replication initiation and observed that cell mass per origin varied with its level. Specifically, the mass per origin increased by 30% at a low DnaA level (about 20% of the wild-type level) and decreased by 10 to 15% at about a three- to fivefold increase in DnaA levels. When intracellular DnaA levels were nearly comparable to the wild-type level, initiation occurred at the proper cell mass. Thus, our results suggest that the DnaA level could act as a positive factor for determining the timing of replication initiation in B. subtilis, if it increases with progression of the replication cycle as in E. coli (14). As described in the section presented above, however, transcription of dnaA seems to occur only after initiation of replication, and it could be autoregulated by newly synthesized DnaA. These events suggest that the DnaA level may reach a constant level at an early stage of the replication cycle in B. subtilis. Thus, it is necessary to examine changes in the DnaA level in the cell division cycle in detail before concluding that the DnaA level acts as the determinant of replication initiation in B. subtilis. Nevertheless, the initiator titration model proposed in E. coli appears unlikely to occur in B. subtilis, because addition of extra DnaA boxes did not derepress transcription of the dnaA operon as described above.

When asynchronous initiation of replication was induced with high DnaA levels, cells with only one origin or more than eight origins were rarely detected, similar to what was observed with wild-type DnaA levels (Fig. 4C). This observation is in significant contrast to that of E. coli dam mutants, in which asynchronous initiation was occurring and a significant fraction (~15%) of cells contained one origin or a higher number of origins than were observed in wild-type cells (28). Presumably, this asynchrony of the mutants was due to repetitive initiations at the same origin and a loss of discrimination between new and old origins, because newly replicated origins were not sequestered (28). However, in B. subtilis (asynchronous initiation under high DnaA levels), discrimination between new and old origins appears to be maintained. Synchrony of initiation is tightly controlled, because cells with an odd number of origins are hardly detected in log-phase wild-type cells. However, when a synchronized culture prepared from the "baby cell machine" was analyzed, cells with three origins appeared even in wild-type cell cultures soon before the number of cells with four origins reached the peak (28). These results suggest that replication initiation occurs sequentially from one to the other origins present in the cell within a very short time period. Asynchronous initiation observed in cells containing low and high DnaA levels in this study may reflect the nature of sequential initiations, although the effects are magnified under these conditions.