A Nucleotide Switch in the Escherichia coli DnaA Protein Initiates Chromosomal Replication

The ATP-bound DnaA protein opens duplex DNA at the Escherichia coli origin of replication, leading to a series of initiation reactions in vitro. When loaded on DNA, the DNA polymerase III sliding clamp stimulates hydrolysis of DnaA-bound ATP in the presence of the IdaB/Hda protein, thereby yielding ADP-DnaA, which is inactive for initiation in vitro. This negative feedback regulation of DnaA activity is proposed to play a crucial role in the replication cycle. We here report that the mutant protein DnaA R334A is inert to hydrolysis of bound ATP, although its affinities for ATP and ADP remain unaffected. The ATP-bound DnaA R334A protein, but not the ADP form, initiates minichromosomal replication in vitro at a level similar to that seen for wild-type DnaA. When expressed at moderate levelsin vivo, DnaA R334A is predominantly in the ATP-bound form, unlike the wild-type and DnaA E204Q proteins, which in vitro hydrolyze ATP in a sliding clamp- and IdaB/Hda-dependent manner. Furthermore, DnaA R334A, but not the wild-type or the DnaA E204Q proteins, promotes overinitiation of chromosomal replication. These in vivo data support a crucial role for bound nucleotides in regulating the activity of DnaA during replication. Based on a homology modeling analysis, we suggest that the Arg-334 residue closely interacts with bound nucleotides.

Chromosomal duplication occurs only once during the cell cycle and is regulated mainly by ingenious controls that act during the initiation of replication (1)(2)(3). The Escherichia coli initiator protein, DnaA, binds to the chromosomal origin of replication (oriC) and promotes a series of reactions leading to the formation of a replication fork (4,5). This protein has high affinity for ATP and ADP, but only the ATP-bound form (ATP-DnaA) can initiate replication at oriC. ATP-DnaA (but not ADP-DnaA) causes local unwinding of the oriC DNA duplex, which creates a site of entry for the DnaB helicase. DnaB helicase loaded on the unwound site forms a complex with DnaG primase and expands the single-stranded region. DNA polymerase (pol) 1 III holoenzyme is then loaded on the primed DNA and begins synthesis (6,7). Several features of both DnaA and the oriC locus are important for the regulation of initiation.
Immediately after initiation, re-replication of oriC is restrained temporarily (8). E. coli DNA is modified at the adenine residue of the palindromic sequence GATC by DNA-adenine methyltransferase, but newly replicated DNA remains hemimethylated until acted upon by DNA-adenine methyltransferase (9). The hemimethylated oriC locus is bound by the protein SeqA, which likely inhibits the initiation of replication until full methylation is re-established (10 -12). This eclipse time is speculated to persist for ϳ10 min under conditions in which the cellular doubling time is 30 min (9,10).
Also after initiation, DnaA is likely inactivated by the hydrolysis of bound ATP to yield ADP-DnaA (3). In vitro, ATP hydrolysis is promoted by the pol III ␤ subunit that is loaded onto DNA (the so-called sliding clamp) and by the protein IdaB (13)(14)(15). During processive DNA synthesis driven by the pol III holoenzyme, the ring-shaped pol III ␤ subunit dimer encircles the post-replicated DNA duplex or the heteroduplex formed by an RNA primer on a DNA template (6,7). The requirement of the DNA-bound form of the sliding clamp for DnaA-ATP hydrolysis ensures the timely coupling between DnaA inactivation and the replication cycle (14,15). In previous studies, IdaB activity was detected in partially purified fractions (14) and more recently has been associated with a new protein, Hda (16). The inactivation of DnaA has been termed RIDA (for regulatory inactivation of DnaA) (14), which is characterized by a decrease in the level of ATP-DnaA following initiation (14,17). In a synchronized culture, it takes ϳ15-20 min for ATP-DnaA levels (representing 80 -90% of the total amount of ATPand ADP-bound DnaA) to decrease to the basal level (ϳ40% in the strain background examined). Thus, the sequestration of oriC by SeqA and the RIDA system presumably complement each other in defining the inter-initiation time of the cell cycle (3,18,19). In addition, DnaA is titrated by the datA locus, which is a site at 94.7 min that contains a DnaA box cluster, providing another means to repress DnaA activity and to prevent overinitiation (3,20,21).
Some dnaA mutants that lack a tight affinity for adenine nucleotides exhibit overinitiation of replication (3). Based on an assumption that the control of DnaA activity by bound nucleotides plays a crucial role in controlling initiation, an explanation for overinitiation is that the mutant DnaA does not undergo inactivation by RIDA and continues to promote initiation. If this is the case, a prediction that follows is that overinitiation could also be caused by a mutant DnaA that has high affinity for adenine nucleotides but is defective for the hydrolysis of bound ATP by RIDA. Such a mutant, however, has not yet been reported.
Recently, we found that the arginine 334 residue of DnaA (Arg-334) is required for ATP hydrolysis, because substitution of arginine by histidine (the R334H mutation) inhibits RIDA and the intrinsic ATPase activity of DnaA (15). However, this mutant protein is unstable in its ability to promote initiation (15,22,23). In this study, we show that the R334A substitution also inhibits RIDA and the intrinsic ATPase activity of DnaA and that the expression of the DnaA R334A protein causes the overinitiation of chromosomal replication. These findings emphasize the importance of the Arg-334 residue in ATP hydrolysis and the significance of bound nucleotides in the control of DnaA activity by RIDA, and hence in the regulation of replication in E. coli. Arginine 334 is a conserved basic residue found in the Box VIII motif of members of the AAA ϩ protein family, which includes chaperone-like ATPases (24). Based on the motifs conserved among AAA ϩ proteins and the structure of one such protein (25), the archaeal Pyrobaculum aerophilum Cdc6 protein, we constructed a model of the tertiary structure of the DnaA ATP-binding domain (domain III) by the homology-modeling method. This model provides an explanation for the possible role of the Arg-334 residue in ATP hydrolysis.
Previously, the intrinsic ATPase activity of another DnaA mutant protein, DnaA E204Q, was reported to be only onethird that of the wild-type DnaA (26,27). In this paper, we also investigated the in vitro and in vivo RIDA behaviors of this mutant protein and, in addition, re-examined several of the reported features in replicational initiation control in vivo.
Tryptone medium contained 10 g/liter tryptone (Difco) and 5 g/liter NaCl (33). Supplemented TG medium was as described (17) except that 0.2% glycerol was used instead of glucose. For the selection of plasmids, tetracycline (15 g/ml) and/or ampicillin (100 g/ml) were included in these media unless otherwise noted. Other media and plasmids have been described (34).
Site-directed Mutagenesis of the dnaA Gene-Site-specific mutagenesis was performed using the method described (29). Briefly, using M13mp19 DNA that carries the dnaA-coding region flanked by the BamHI and HindIII sites at the 5Ј and 3Ј ends, respectively (22), uracil-containing single-stranded DNA was prepared with BW313 (ung-1) cells. For introduction of the R334A mutation, the resulting DNA was hybridized with a primer, 5Ј-GATCTAACGTAGCTGAGCTG-GAAG-3Ј, the complementary DNA strand was synthesized in vitro, and the resulting double-stranded DNA was introduced into JM109 cells. The mutation was confirmed by sequencing. DNA fragments isolated by digestion with BamHI and HindIII were used for the plasmid constructions described above.
Purification of the DnaA R334A Protein-All procedures were essentially the same as those for purification of the DnaA R334H protein, as previously described by Takata et al. (23). Briefly, KA450 cells carrying pHS284-2 were grown in 15 liter of LB medium at 37°C, expression of the mutant dnaA gene was induced by adding 1% D-arabinose to log phase cells, and incubation was continued for 1.5 h. Ammonium sulfate at a final concentration of 0.22 g/ml was added to cleared lysates, and the resulting precipitate was collected by centrifugation, dissolved in buffer C (35), and dialyzed against the same buffer. Insoluble materials were collected by centrifugation, washed with 0.6 M ammonium sulfate, and dissolved with buffer C containing 0.6 M ammonium sulfate and 4 M guanidine HCl. A monomer fraction (fraction IV) of DnaA R334A protein was obtained by gel filtration chromatography using Superose 12 (Amersham Biosciences) equilibrated with buffer D (35).
In Vitro RIDA Systems-RIDA-active fractions II and III were prepared from WM433 as described (14). Briefly, fraction II was prepared by ammonium sulfate precipitation (0.24 g/ml) of fraction I proteins, and the activity was further concentrated as fraction III by the separation of proteins by DE52 column chromatography. These fractions were incubated with [␣-32 P]ATP-bound DnaA protein and the oriC plasmid M13E10 in the same buffer as that used for in vitro replication with WM433 extracts. DnaA protein with bound nucleotides was isolated by immunoprecipitation, and radiolabeled nucleotides were separated by polyethylenimine (PEI) cellulose thin layer chromatography and quantified as described (14).
P1 Nuclease Assay-This assay was done as described (23,36). Briefly, buffer (50 l) containing 60 mM Hepes-KOH (pH 7.6), 8 mM magnesium acetate, 30% (v/v) glycerol, 400 ng of the pBS oriC minichromosome (3.7 kb), 0.32 mg/ml bovine serum albumin, and 5 mM ATP was preincubated at 38 or 28°C for 1 min. ATP-bound DnaA protein was then added and incubation was continued for 3 min at 38°C or 6 min at 28°C, followed by incubation with P1 nuclease (10 units; Yamasa Co.) for 25 s at the same temperature. After the reaction was stopped by addition of 0.3% SDS, DNA was extracted with phenol/chloroform, precipitated with ethanol, digested in buffer with ScaI, and analyzed by agarose (1%) gel electrophoresis. DNA fragments of 2.1 and 1.6 kb are generated by this restriction enzyme when open complexes are formed, and thus the oriC site is digested with P1 nuclease (23). The extent of open complex is indicated by the percentage of these fragments. Gels were stained with GelStar (BioWhittaker, Walkersville, MD), and DNA band intensities were quantified by densitometric scanning of photographic negatives (Polaroid, Cambridge, MA).
Determination of the Relative Proportions of oriC and terC Sequences by Southern Blot Analysis-Cells were exponentially grown in supplemented TG medium and harvested by centrifugation of 1.5-ml aliquots. Total DNA was prepared by the method described by Atlung and Hansen (37). Briefly, cell pellets stored at Ϫ20°C were thawed, resuspended in 270 l of 10 mM Tris-HCl (pH 8.0), 10 mM NaCl, 10 mM EDTA (pH 8.0), 1% Triton X-100, and incubated for 2 h at 37°C with 150 g of lysozyme and 150 g of proteinase K. The lysate was heated at 65°C for 2 h, and nucleic acids were precipitated with ethanol and resuspended in 100 l of TE buffer. DNA (2 g) was digested with EcoRI and PstI, and the fragments were separated by electrophoresis (50 V) on a 20-cm 0.7% agarose gel. The gel was soaked in 0.5 g/ml ethidium bromide for 40 min, photographed, soaked in 0.37% HCl for 10 min, washed twice in deionized water, and soaked for 40 min in 0.5 M NaOH, 1.5 M NaCl solution and for 60 min in 0.5 M Tris-HCl (pH 7.2), 3 M NaCl solution. DNA was then transferred to a GeneScreen Plus filter (Du-Pont, Wilmington, DE) by the capillary method (34). After transfer, the gel was stained with ethidium bromide and photographed again to verify a complete transfer of the DNA. The filter was hybridized with [␣-32 P]dCTP-labeled probes for the oriC and terC regions. After washing, the filter was exposed on an imaging plate (Fuji Film, Japan) and the intensity of the bands was quantified by using BAS2500 (Fuji Film). Each gel also included DNA samples prepared from KA413 (dnaA46) cells incubated for 2 h at 42°C, which allows sufficient time for replication of the entire chromosome, to normalize the hybridization signals. For preparation of the oriC or terC probes, M13mpRE85 was digested with EcoRI and PstI and pKA58 with EcoRI and HindIII. Restriction fragments (0.5 kb from M13mpRE85 and 1.0 kb from pKA58) were isolated from 0.7% agarose with the QIAEX II gel extraction kit (Qiagen, Valencia, CA) and resuspended in TE buffer. Approximately 25 ng of oriC and terC fragments were labeled with the Megaprime DNA labeling system (Amersham Biosciences). Total counts of 10 8 cpm were obtained for both probes.
Measurement of DNA Synthesis-Cells were cultured in tryptone medium with 3 Ci of [ 3 H]thymine (Moravek Biochemicals, Brea, CA) at a concentration of 25 g/ml (38). Acid-insoluble materials were prepared with chilled 5% trichloroacetic acid, and radioactivity was measured with a scintillation counter.
Determination of Cellular Levels of the Adenine Nucleotide Forms of DnaA Protein-Cells were labeled by growth in supplemented TG medium containing 0.4 mCi/ml [ 32 P]orthophosphate as described (17). Briefly, cleared lysates (750 l) from aliquots (2 ml) of the culture were prepared and mixed with anti-DnaA antiserum (5 l) that had been pre-incubated in 100 l of a cleared lysate (30 mg/ml protein) of KP7364 (⌬dnaA::spec) for 30 min at 0°C. Protein A-Sepharose (60 l; 50% slurry) was added, suspended for 30 min at 2°C, and washed repeatedly in chilled buffers. After removal of the final washing solution, immunoprecipitates were extracted in a solution (20 l) of 1 M HCOOH and 5 mM each of ATP, ADP, and AMP. Radiolabeled nucleotides were separated by PEI-cellulose thin layer chromatography and quantified as described (14).
Homology Modeling-The homology study was performed using the homology module of the INSIGHT II molecular modeling package (MSI/ Biosym, San Diego, CA). Because the AAA ϩ modules of the Saccharomyces cerevisiae Cdc6 protein are similar to that of DnaA (24), the atomic coordinates for the P. aerophilum Cdc6 homolog, whose structure has been solved by x-ray crystallography (25), were obtained from the Brookhaven Protein Data Bank. For structurally conserved regions (SCRs), which are located in ␣-helices and ␤-strands common to DnaA domain III and P. aerophilum Cdc6 domains I and II, the corresponding coordinates in the P. aerophilum Cdc6 protein were used as a structural template. Loop structures connecting SCRs were selected from ten loop coordinates in the homology module. The initial DnaA domain III structure was refined in four steps. In step 1, the regions connecting SCRs and loops in the initial DnaA domain III structure were energy-minimized using the consistent-valence force field implemented in DIS-COVER, with an 8-Å cutoff for nonbonded interactions. During energy minimization, template forcing was used to restrain the main chain at the SCRs in the initial DnaA domain III structure by using a harmonic potential with a force constant, K ϭ 100 kcal/mol Å 2 . In step 2, the loop regions in the initial DnaA domain III structure were similarly energyminimized. For step 3, the molecular dynamics began with the energyminimized structure at 300K, 10,000 steps. In step 4, the resulting model was further energy-minimized by using 100 steps of the steepest descent and 1000 steps of the conjugate gradient.
The coordinates of ADP bound to DnaA domain III were initially determined using those for ADP in the Cdc6-ADP complex as a structural template (25). The molecular dynamics of the initial ADP-DnaA domain III complex structure was computed using the consistent-valence force field implemented in DISCOVER at 300K, 10,000 steps. During simulations, template forcing was used to restrain the main chain at the SCRs in the initial ADP-DnaA domain III structure. Thereafter, the resulting model was further energy-minimized by using 100 steps of the steepest descent and 1000 steps of the conjugate gradient. The domain III structural images shown here were generated using the MOLSCRIPT program (39).

RESULTS
Initiation Activity of the DnaA R334A Protein-Assuming that ATP-DnaA is the active form that initiates chromosomal replication in vivo, DnaA mutants that are defective in hydrolysis of the bound ATP may cause overinitiation because the ATPbound form of DnaA accumulates. We previously found that the DnaA R334H protein is defective in ATP hydrolysis in vitro, but its ability to promote initiation was unstable at 30°C and defective at 20°C in vivo and in vitro (15,22,23). In this study, we substituted the DnaA arginine 334 with alanine (see "Materials and Methods"). Using methods similar to those used for the preparation of wild-type DnaA and the DnaA R334H protein (23), we overproduced the DnaA R334A protein in a dnaA-null host strain and purified it to near homogeneity (Fig. 1A).
DnaA R334A protein can initiate minichromosomal replication in vitro and can bind adenine nucleotides ( Figs. 1 and 2). When assessed at 30°C in a minichromosomal replicationcompetent crude extract (28), the initiation activity of DnaA R334A was found to be comparable with that of the wild-type protein (Fig. 1B). Moreover, ADP-bound DnaA R334A is unable to initiate replication, similar to what is observed for the wildtype protein (Fig. 1C), which indicates that bound nucleotides also control the activity of the mutant protein. The respective binding affinities for ATP and ADP are comparable for the mutant and the wild-type DnaA proteins (Fig. 2); the dissociation constants (K d ) of DnaA R334A for ATP and ADP are 19 and 18 nM, respectively, and those of wild-type DnaA are 14 and 26 nM, respectively, consistent with previous data (36). Similar results regarding nucleotide affinities and initiation activity were obtained for DnaA R334H (15,23).
Another DnaA mutant, the DnaAcos protein, overinitiates chromosomal replication at 30°C (2,38). When this protein is incubated in replication-competent crude extracts at 30°C, minichromosomal replication continues for over 45 min, whereas replication initiated by the wild-type DnaA ceases within 20 min (40). This is explained by observations that DnaAcos, but not the wild-type DnaA, is resistant to DnaAspecific inactivation by components of this extract, and thus the initiation activity of DnaAcos is more stable than that of wildtype DnaA (13,40). This DnaA-inactivating system is now termed RIDA (14,16,17). When the activity of the DnaA R334A protein was assessed under similar conditions, the results obtained were basically similar to those seen for the DnaAcos protein (Fig. 1, D and E).
The initiation activity of the DnaA R334H protein is coldsensitive, in that open complex formation at oriC, which can be detected by the P1 nuclease assay, is significantly inhibited at 28°C but not at 38°C (the optimal assay temperature), when compared with what is observed for wild-type DnaA (23). The ability of the DnaA R334A protein to promote open complex formation, as assessed by the P1 nuclease assay, is not significantly inhibited at 38°C nor even at 28°C, similar to what was shown for wild-type DnaA at each temperature (Fig. 1, F  and G).
The DnaA R334A Protein Is Insensitive to RIDA-We next asked whether DnaA R334A-bound ATP is hydrolyzed in a RIDA-dependent manner in vitro (Fig. 3) to determine whether the DnaA R334A protein is less sensitive to RIDA, as was shown for DnaA R334H (15). In these experiments, we used partially purified protein fractions (fractions II and III) (14,15) containing the DNA polymerase III holoenzyme and the IdaB/ Hda protein, which are required for RIDA. The wild-type DnaA-bound ATP in these fractions is efficiently hydrolyzed to yield ADP-form molecules, whereas DnaA R334A-bound ATP is not (Fig. 3, A and B). Although a slight residual activity (30 -40% hydrolysis) was seen for DnaA R334A, the distinction between DnaA R334A and wild-type activity is evident.
Similar experiments were performed with DnaA E204Q. Although the intrinsic ATPase activity of this protein was reported to be approximately one-third that of the wild-type protein (26), its sensitivity to RIDA had not been examined. As shown in Fig. 3C, the hydrolysis of DnaA E204Q-bound ATP is efficient in the RIDA reaction system, similar to what was seen for the wild-type protein.
Intrinsic ATPase Activity of the DnaA R334A Protein-To shed further light on the mechanism of the RIDA insensitivity of DnaA R334A, we assessed its intrinsic ATPase activity (36). After incubation in reaction buffer, DnaA protein was isolated by immunoprecipitation, and the recovered nucleotides were separated by PEI-cellulose thin layer chromatography, as was done for the RIDA assay (above). The results indicate that the intrinsic ATPase activity of DnaA R334A is minimal at 38°C (Fig. 4). Similar results were obtained when the protein was incubated at 30°C (data not shown). As DnaA R334A is competent for the initiation of replication and exhibits affinity for adenine nucleotides, and as its activity is controlled by bound nucleotides, these results indicate that the R334A mutation specifically affects hydrolysis of the bound ATP.
In Vivo DnaA-ATP Hydrolysis-We next assessed the in vivo relative proportions of ATP-and ADP-bound mutant DnaA proteins. Cells were grown in a synthetic medium containing [ 32 P]orthophosphate, DnaA protein was isolated by immunoprecipitation, and the bound nucleotides were analyzed by thin layer chromatography (14). In normally growing cells, the ATPbound form of DnaA represents ϳ10 -20% of the total ATP-and ADP-DnaA molecules (14,17).
As constitutive expression of the dnaA29 (R334A) gene carried on pBR322 inhibits the growth of normal host cells (see below), controlled expression of this and other dnaA-alleles was achieved by placing them downstream of the lac promotor on a pBR322-derivative in lacI q cells. When dnaA29 (R334A) expression is induced by the addition of 1 mM IPTG, the proportion of ATP-bound DnaA proteins increases from ϳ10% to 65% (Figs. 5A and 6). Upon induction, the overall amount of ATPbound molecules significantly increases (Fig. 6, A-C); a slight increase in the level of ADP-bound molecules is also observed, which may be attributed to a slight residual sensitivity to RIDA After incubation for 20 min at 30°C, nucleotides incorporated into acid-insoluble material were quantified. C, Wild-type DnaA and DnaA R334A were incubated for 15 min at 0°C in buffer containing 1 M ATP or ADP. Then, the indicated amounts of those proteins were added to the same reaction as above for quantifying replication activities. WT-ATP and R334A-ATP, wild-type DnaA (E) and DnaA R334A (Ⅺ) proteins preincubated with ATP; WT-ADP and R334A-ADP, wild-type DnaA(q) and DnaA R334A (f) proteins preincubated with ADP. D, Wild-type DnaA (E) and DnaA R334A (q) (1 pmol each) were incubated at 30°C for the indicated time in reactions (25 l) containing a WM433 replication-competent crude extract, and minichromosomal DNA replication was quantified as above. E, The ATP-forms (0.75 pmol) of wild-type DnaA (E) and DnaA R334A (q) were incubated at 30°C for the indicated time in an in vitro RIDA system that contains buffer (5 l) with 2 mM ATP, the M13E10 minichromosome (200 ng) and WM433-derived crude extract (fraction II; 200 g) (see "Materials and methods"). Residual activities for minichromosomal replication were assessed as described above. Relative activities are indicated as a percentage (267 and 170 pmol for wild-type DnaA and DnaA R334A, respectively, were 100% replication). F and G, Activity for open complex formation was assessed by the P1 nuclease digestion assay. The indicated amounts of the ATP-forms of wild-type DnaA (E) and DnaA R334A (q) were incubated at 38°C for 3 min (F) or at 28°C for 6 min (G) in reactions (50 l) containing 5 mM ATP, the pBS oriC minichromosome (400 ng) and 34 ng HU protein. After incubation with P1 nuclease, products specifically digested at the oriC site were quantified by agarose gel electrophoresis and densitometric scanning of photographs of GelStar-stained gels (see "Materials and Methods").  (Fig. 3). In contrast, induction of the wild-type protein is not associated with an increase in the level of ATP-DnaA (Fig. 5A), consistent with our previous data (14). Instead, the level of ADP-bound molecules dramatically increases, suggesting that DnaA-bound ATP is rapidly hydrolyzed (Fig. 6, D-F). The data shown in Fig. 6 (B and E) also suggest that the total levels of nucleotide-bound DnaA are comparable between the two strains before and after induction, especially from 0 to 130 min. Thus, we used the same conditions to assess the frequency of initiation, as described below.
When DnaA E204Q protein is similarly expressed, ATP-form molecules do not accumulate to significant levels (Fig. 5A). To eliminate background caused by the presence of the host wild-type DnaA, dnaA (E204Q) was expressed from a pBR322 derivative in a dnaA-null rnhA double mutant. The absence of the rnhA function allows chromosomal replication from alternative origins to take place (41). In the dnaA-null mutant, a significant increase in the level of ATP-bound DnaA E204Q protein was not seen (Fig. 5B). These results are consistent with the sensitivity of the DnaA E204Q protein to RIDA (Fig. 3C).
Overinitiation of Chromosomal Replication by the DnaA R334A Protein-Whereas ATP-bound DnaA R334A is active for initiation in vitro (Fig. 1), ATP bound to this protein is only poorly hydrolyzed in vitro (Fig. 3) and in vivo (Figs. 5 and 6). Assuming that ATP-DnaA is the active form for initiation in vivo and that the initiation activity of DnaA R334A is stable in vivo, expression of this protein may cause the overinitiation of chromosomal replication. To test this assumption, we quantified the relative proportions of oriC and terC sequences in cells that bear DnaA-expressing plasmids (Table I). If overinitiation occurs, the oriC/terC ratio will increase after the induction of DnaA by IPTG. As excessive levels of the wild-type DnaA protein are reported to promote overinitiation (37, 42), we also tested a control strain that induces the wild-type DnaA. Before and after the inducer IPTG was added to cultures of exponentially growing cells that bear each expression plasmid, cells were harvested for Southern analysis with probes for oriC and terC.
Upon induction of DnaA R334A, the oriC/terC ratio increases significantly, whereas induction of the wild-type DnaA or DnaA E204Q proteins results in only a slight increase or no change (Table I). After induction of DnaA R334A, the oriC/terC ratio increases gradually over 120 min, to a maximal level of ϳ5.5 times the basal level (Fig. 7). In contrast, induction of the wild-type DnaA increases the ratio by at most 2.4-fold (Fig. 7), which is consistent with previous observations (37, 42). There is no significant difference in the overall level of total DnaA before induction between the wild-type and DnaA R334A-expressing strains, as assessed by immunoblotting (38); after  (36). Nucleotide-bound DnaA proteins were recovered by immunoprecipitation and the nucleotides were analyzed by PEI-cellulose thin layer chromatography. The relative intensities of ATP and ADP were quantified with a BAS2500 image analyzer (Fuji Film). The ratio of ADP-DnaA to total ATP-and ADP-bound DnaA is shown as a percentage.
FIG. 5. In vivo nucleotide-bound forms of DnaA. Cells were grown in supplemented TG medium containing [ 32 P]orthophosphate, and nucleotide-bound DnaA proteins were recovered by immunoprecipitation. The results of PEI-cellulose thin layer chromatography to identify DnaA-bound nucleotides are shown. The origin and positions to which ATP and ADP migrate are indicated by arrows. Anti-DnaA antiserum (A) and preimmune serum (P) were used. The ratio of ATP-DnaA to total ATP-and ADP-bound DnaA is shown as a percentage. A, pHSL99 (lacI q )-bearing KH5402-1 (wild-type dnaA) was used as the host strain for the plasmids pSN306 (wild-type dnaA) (WT), pSN305 (dnaA29 (R334A)) (R334A), pSN307 (dnaA400 (E204Q)) (E204Q) and the vector pSN300 (none). Cells were grown at 37°C until the optical density (A 660 ) of the culture reached 0.2, and were further incubated for 90 min at the same temperature in the presence (ϩ) or absence (-) of 1 mM IPTG. For cells bearing pSN307 (dnaA400(E204Q)) incubated at 42°C, results similar to those shown above were obtained (data not shown). B, KA451 (dnaA::Tn10 rnhA::cat) was used as the host for the following pBR322 derivatives bearing dnaA alleles: pMZ002-1 (E204Q), pMZ002-2 (WT), and pMZ002 (none). Cells were grown at 37°C until the optical density (A 660 ) of the culture reached 0.3. NA, not available.
induction, the DnaA content increases to 3-3.5 times the uninduced level in both these strains (data not shown). These results indicate that expression of DnaA R334A causes overinitiation of chromosomal replication in vivo.
Chromosomal Replication-There are two types of overini-tiation with respect to the extent of replication (3,43). In one type, when additional replisomes are present the whole genome is replicated, resulting in chromosomal overreplication. The overinitiation that occurs in the dnaAcos mutant, the GroEoverexpressing dnaA46 mutant, and the DnaA46-overexpressing dnaA46 mutant results in this "inertia" mode of replication.
In the other type, only oriC and nearby sequences are replicated by additional replisomes. The overinitiation that occurs in wild-type DnaA-overexpressing cells and DnaA A184V protein-overexpressing cells results in this "attenuation" mode of replication.
To distinguish which type of replication takes place in DnaA R334A-bearing cells, we measured overall DNA synthesis by monitoring the incorporation of [ 3 H]thymine in thyA-defective host cells (Fig. 8). In all cases in which the wild-type DnaA, DnaA R334A, and DnaA E204Q proteins are induced, significant overreplication of the whole chromosome is not observed. Similar results have been reported for cells in which the wildtype DnaA is overexpressed (37, 42). Thus, additionally formed replisomes in DnaA R334A cells are likely restricted to the vicinity of oriC. In contrast to what has been previously reported (26,27), we never observed overreplication promoted by DnaA E204Q; in those previous experiments, JM109 host cells with an intact thyA gene were used to measure the incorporation of [ 3 H]thymine (see "Discussion").
Inhibition of Cell Growth by the DnaA R334A Protein-Overinitiation inhibits the growth of some dnaA mutants (3). In the dnaAcos mutant, cell division is inhibited in an sfiAindependent manner, resulting in cellular filamentation and the inhibition of cell proliferation (44). An alternative oriCindependent replication system operates in the absence of RNase HI, and cells lacking oriC can grow even in the presence of overinitiating dnaA mutant alleles such as dnaAcos (3,16,38). We found that the introduction of the dnaA29 (R334A) allele on a pBR322 derivative (pHS299-1) inhibits colony formation when the host cells are employing the normal replication system, but does not inhibit colony formation when the oriC-independent system is operating (Table II). These results suggest that the DnaA R334A protein specifically affects initiation at oriC and are consistent with the data suggesting that DnaA R334A promotes overinitiation.
The dnaA400 (E204Q) allele (borne on the pBR322 derivative pMZ002-1) was previously reported to cause a similar inhibition, although the host strain used in this study was JM109, which has a genetic background entirely different from that of KH5402-1, the parental strain of KA450 (⌬oriC dnaA(Am) rnhA(Am)), which was used as the positive control in that report (26). When KH5402-1 (wild-type dnaA) is used as FIG. 6. Increase of the ATP form upon induction of DnaA R334A. KH5402-1 (pHSL99) cells bearing pSN305 (dnaA29 (R334A)) (A-C) or pSN306 (wild-type dnaA) (D-F) were grown in supplemented TG medium containing [ 32 P]orthophosphate at 37°C until the optical density (A 660 ) of the culture reached 0.2. 1 mM IPTG was then added, and the incubation was continued for the indicated time (see "Materials and Methods"). Portions (2 ml) of the cultures were withdrawn, nucleotide-bound DnaA protein was recovered by immunoprecipitation, a 1.2 l portion of each solution was spotted on a PEI-cellulose sheet, and the nucleotides were analyzed as described in the legend to Fig. 5. As a reference for quantifying the labeled nucleotides, various concentrations of [ 32 P]orthophosphate solution were spotted on the same sheet before electric imaging. A and D, thin layer chromatography of DnaA immunoprecipitates. The origin and positions to which ATP and ADP migrate are indicated by arrows. Anti-DnaA antiserum (␣-DnaA) and preimmune serum (pre) were used. B and E, quantification of nucleotides analyzed by thin layer chromatography. C and D, the ratio of ATP-DnaA to total ATP-and ADP-DnaA is shown as a percentage. a The host strain was KH5402-1 bearing pHSL99 (lacI q ). Growth of cells and IPTG induction were as described for Fig. 5.
b The ratio indicated was calculated with respect to a standard oriC/ terC ratio (defined as 1) obtained using dnaA46 mutant cells that completed replication of the entire genome during a 2-h incubation at 42°C.

FIG. 7.
Overinitiation by induction of DnaA R334A. KH5402-1 (pHSL99) cells bearing pSN305 (R334A; q) or pSN306 (WT; E) were grown at 37°C in supplemented TG medium containing thymine (25 g/ml) until the optical density (A 660 ) reached 0.1, 1 mM IPTG was added, and the incubation was continued. At the indicated time after induction, portions (1.5 ml) were withdrawn for Southern analysis using 32 P-labeled oriC and terC probes (see "Materials and Methods"). The relative intensities of the oriC and terC regions were quantified with a BAS2500 image analyzer (Fuji Film). The oriC/terC ratio at the time of induction (time 0) was defined as 1.
a host, the dnaA400 (E204Q) allele does not inhibit cell growth at 37°C (Table II) or 42°C (data not shown). Immunoblotting analysis indicated that the relative amounts of DnaA in KA450 cells bearing pHS299-1 (dnaA29 (R334A)) and pMZ002-1 (dnaA400 (E204Q)) are ϳ0.84 and 2.4, respectively, if the level of wild-type dnaA in KA450 cells bearing pMZ002-2 (wild-type dnaA) is defined as 1 (data not shown). Thus, the inhibition conferred by pHS299-1 likely is not the result of overexpression of the mutant DnaA protein.
Homology Modeling of DnaA Domain III-To determine how the Arg-334 residue functions in ATP hydrolysis, we constructed a model of DnaA domain III, the ATP-binding domain (5,45). This domain contains AAA ϩ motifs, including the Walker-type NTP binding motifs and AAA ϩ -specific motifs (24). We found that the primary amino acid sequence and the predicted secondary structure of DnaA domain III have significant homology with a region from the archaeal P. aerophilum Cdc6 protein (domains I and II) that includes AAA ϩ motifs (25) (Fig.  9A). The structure of the ADP-bound form of this protein has been solved by x-ray crystal diffraction analysis (25). Specifi-cally using the AAA ϩ motif homology, we carried out a homology-modeling analysis of DnaA domain III, as described under "Materials and Methods" (Fig. 9B). In the predicted structure, the ␣-helix that includes the DnaA Box VIII motif is located near the bound ADP. The Arg-334 side chain is close to the ADP ␤-phosphate (Ͻ4 Å), which suggests that this residue has an important role in mediating interactions with ATP. In contrast, when the Arg-334 residue is replaced with His or Ala, the side chains of these residues are further (Ͼ5.5 Å) from the phosphate bond (Fig. 10). Similarly, the Glu-204 residue is located at a distance that precludes a close interaction with ATP (Fig. 9B). DISCUSSION We found that the DnaA R334A protein is defective in the hydrolysis of bound ATP in vitro and in vivo but that it retains the ability to promote initiation in a bound nucleotide-dependent manner. As a moderate level of this protein, but not of the wild-type protein, causes overinitiation in vivo, we speculate that the ATP-bound form, but not the ADP-DnaA form, of DnaA is active for initiation in vivo. These results suggest that the nucleotide bound to DnaA plays a crucial role as a molecular switch for initiating chromosomal replication. Furthermore, our data support the notion that RIDA plays an indispensable role in the in vivo regulation of this initiation switch. Recently, a newly identified protein, Hda, was found to be required for RIDA in vitro and in vivo (16). This protein exhibits IdaB activity, which was proposed to be involved in the RIDA reaction. Inactivation of the hda gene results in an increase in cellular ATP-DnaA content, up to ϳ70% of total ATP-and ADP-DnaA molecules, and in the overinitiation of chromosomal replication. Our present data are highly consistent with these results.
Other findings highlighting the nucleotide-directed control of DnaA and of other initiation proteins in eukaryotic systems have recently emerged. One report demonstrates that ATP-DnaA, but not ADP-DnaA, binds to a specific sequence called the ATP-DnaA box (46), which is a 6-mer motif that lies adjacent to the 9-mer DnaA box. This sequence is present in the dnaA promoter, and ATP-DnaA represses the transcription of the dnaA gene more tightly than does ADP-DnaA. An interaction between the ATP-DnaA and ATP-DnaA boxes found at oriC is suggested to be required to open duplex DNA (47). In S. cerevisiae, the origin recognition complex, a protein assembly, must be bound by ATP to bind yeast origins of replication (autonomous replicating sequences). Further, autonomous replicating sequence singled-stranded DNA stimulates ATP hydrolysis of the origin recognition complex subunit Orc1, which may play an important role in the inhibition of untimely initiations (2,48,49). The nucleotide-dependent control of replication-initiating proteins may be ubiquitous in prokaryotic and eukaryotic replicons. As ATP-dependent initiation is a feature of the DnaA R334A and DnaA R334H proteins (Fig. 1) (15), Arg-334 is unlikely to be the residue responsible for the allosteric change that triggers the molecular switch. The mecha-  The extents of identity and similarity (including identity) between the two amino acid sequences are 18 and 31%, respectively. The identical and similar amino acids are indicated with yellow and green boxes, respectively. Amino acid positions are indicated and Arg-334 is shown by a bold font. The defined AAA ϩ motifs indicated are boxed. ␣-helices and ␤-strands in P. aerophilum Cdc6 that were used for structural modeling are indicated by red and blue bars, respectively. B, Ribbon drawing of the structure of DnaA domain III. A structural model of the ADP-bound DnaA domain III was constructed by a homology modeling method (see "Materials and Methods"). ADP and the side chains of the Glu-204, Arg-334 and Leu-367 residues are shown as a ball-and-stick model. The Leu-367 residue defines the C terminus of this domain, and the N terminus is shown as N-term.
nisms regarding this process remain to be elucidated.
The initiation of replication is regulated by multiple pathways during the E. coli cell cycle (3,18,19). Upon induction of the DnaA R334A protein, the ATP-DnaA proportion of total nucleotide-bound DnaA molecules increases rapidly (Fig. 6), whereas the increase in the oriC/terC ratio is slower than the increase in the level of ATP-DnaA (Fig. 7). This discrepancy may be caused by RIDA-independent regulatory systems that operate under these conditions. The SeqA protein binds the hemimethylated oriC DNA that is transiently present after replication, which temporarily blocks re-initiation (10). ϳ300 DnaA protein molecules can be bound by the datA site, which is an ϳ1-kb chromosomal segment that contains four DnaA boxes (20). These systems may temporarily inhibit re-initiation at oriC, even when excessive levels of ATP-DnaA are present.
DnaA is a member of the AAA ϩ protein superfamily (24). Proteins in this family share common modules containing AAA ϩ -specific motifs, in addition to the ATP (GTP)-binding Walker-type A, B, and C motifs. Analyses of crystal structures of several AAA ϩ proteins (E. coli HslU, Thermus thermophilus RuvB, and P. aerophilum Cdc6) show that in each protein a basic amino acid residue in the AAA ϩ Box VIII motif is located close to the ␤-(or ␥-)phosphate of the bound nucleotide (24, 25, 50 -52). These studies thus suggest a role for this basic residue in mediating interactions with ATP. An arginine residue corresponding to Arg-334 of the DnaA Box VIII motif is widely conserved in DnaA homologs from many bacterial species (5) and also in members of the AAA ϩ superfamily (24). The results of a homology-based modeling analysis are consistent with the idea that the Arg-334 residue is located near the ATP binding site and that it plays a critical role in the hydrolysis of ATP (Fig. 9). In the structural model, the distance between the side chain of the Arg-334 residue and the ␤-phosphate is calculated to be 3.6 -3.9 Å, which means that one water molecule can be stabilized between the arginine side chain and the phosphate (Fig. 9B). This structure could easily produce a hydroxyl ion that could stimulate the hydrolysis of bound ATP. In contrast, the structural models of the DnaA R334A and R334H proteins suggest that this water molecule is not stabilized because the distance between the phosphate moiety and the substituted residue is at least 5.5 Å (Fig. 10). Thus, this modeling data can explain the importance of the Arg-334 residue in stabilizing a water molecule and in promoting ATP hydrolysis.
We note that the interaction of ATP-DnaA with the sliding clamp and Hda during RIDA may alter the mechanism of DnaA-ATP hydrolysis. Like DnaA, Hda is a member of the AAA ϩ family (16). As is known for other proteins in this family, the association of DnaA with Hda may form a more efficient catalytic center for ATP hydrolysis by supplying an additional catalytic residue (24,52). Even so, the Arg-334 residue would be also required for ATP hydrolysis in this complex.
Several observations indicate that overinitiation inhibits the complete replication of the genome (3, 37, 42, 43). Whether DnaA protein itself directly affects replication fork progression remains to be elucidated. Because the hydrolysis of ATP bound to DnaA R334A is inhibited, a complex formed between the ATP-bound DnaA R334A protein and the ␤-subunit sliding clamp might be stabilized, thereby inhibiting replisome movements. Alternatively, only when overinitiation occurs, other unknown factors might interact with DnaA and components of the replisome, resulting in the inhibition of replisome movement.
Based on the results of [ 3 H]thymine incorporation experiments, DnaA E204Q was reported to cause chromosomal overreplication (26,27). However, these experiments used as the host strain JM109, which has the wild-type thyA gene. In wild-type thyA cells, the incorporation of [ 3 H]thymine does not reflect chromosomal replication. Although we repeatedly attempted to determine whether DnaA E204Q indeed can promote overreplication in mutant thyA backgrounds under various conditions, including those previously used, and with the same expression plasmids, we never obtained similar results. Even when JM109 was used, the previous data were never reproduced. In the structural model (Fig. 10), the distance between Glu-204 and the phosphate moiety of bound ADP is at least 9.0 Å, which suggests that close interaction is unlikely.