DnaA Protein DNA-binding Domain Binds to Hda Protein to Promote Inter-AAA+ Domain Interaction Involved in Regulatory Inactivation of DnaA*

Chromosomal replication is initiated from the replication origin oriC in Escherichia coli by the active ATP-bound form of DnaA protein. The regulatory inactivation of DnaA (RIDA) system, a complex of the ADP-bound Hda and the DNA-loaded replicase clamp, represses extra initiations by facilitating DnaA-bound ATP hydrolysis, yielding the inactive ADP-bound form of DnaA. However, the mechanisms involved in promoting the DnaA-Hda interaction have not been determined except for the involvement of an interaction between the AAA+ domains of the two. This study revealed that DnaA Leu-422 and Pro-423 residues within DnaA domain IV, including a typical DNA-binding HTH motif, are specifically required for RIDA-dependent ATP hydrolysis in vitro and that these residues support efficient interaction with the DNA-loaded clamp·Hda complex and with Hda in vitro. Consistently, substitutions of these residues caused accumulation of ATP-bound DnaA in vivo and oriC-dependent inhibition of cell growth. Leu-422 plays a more important role in these activities than Pro-423. By contrast, neither of these residues is crucial for DNA replication from oriC, although they are highly conserved in DnaA orthologues. Structural analysis of a DnaA·Hda complex model suggested that these residues make contact with residues in the vicinity of the Hda AAA+ sensor I that participates in formation of a nucleotide-interacting surface. Together, the results show that functional DnaA-Hda interactions require a second interaction site within DnaA domain IV in addition to the AAA+ domain and suggest that these interactions are crucial for the formation of RIDA complexes that are active for DnaA-ATP hydrolysis.

ation protein, is required for formation of prereplicative complexes at origin sites to load the MCM2-7 replicative helicase (7). Cdt1 protein is ubiquitylated for targeted degradation by complexes of the DNA-loaded clamp (proliferating cell nuclear antigen) and the E3 ubiquitin ligase cullin 4⅐DNA damagebinding protein 1 complex, thereby repressing reinitiation of replication (30 -32).
E. coli Hda harbors a conserved clamp-binding motif (QL(S/ D)LF) at its N terminus (33,34) that is required for clamp binding and RIDA (34,35). In addition to this motif, Hda contains an AAAϩ domain with the Walker-type nucleotide-binding motif and several conserved amino acid sequence motifs (18,36). Hda AAAϩ does not bind ATP but specifically and stably binds to ADP to yield the activated monomeric form of Hda (19).
DnaA consists of four functional domains (1). The N-terminal domain I interacts with several proteins, including DiaA and DnaB (37)(38)(39). Domain II is a flexible linker (40,41). Domain III is an AAAϩ domain that shares amino acid sequence similarity with Hda (18,36). The C-terminal domain IV is a DNA-binding region that contains a helix-turn-helix motif (42)(43)(44). This domain specifically binds to the 9-mer DnaA boxes that are present at oriC, datA locus, and many other sites on the chromosome.
A model of the RIDA intermediate complex in which DnaA domain I interacts with the Hda⅐clamp complex, DnaA domain IV interacts with the DNA flanking the clamp, and DnaA domain III interacts with the Hda AAAϩ domain was proposed (20,22,34,45). Several amino acid residues within the Hda AAAϩ Box VI and Box VII motifs, such as Hda Arg-153 (Arg finger), Phe-118 (H-finger), and Asn-122 (E-finger), play a role in DnaA-ATP hydrolysis and in the DnaA-Hda interaction (see Fig. 1, B-D). In addition, DnaA Arg-334 within the AAAϩ sensor II motif is important for DnaA-intrinsic ATPase activity, the Hda interaction, and DnaA-ATP hydrolysis in RIDA (22,45). In a model of the DnaA⅐Hda complex, DnaA Arg-334 resides near the Hda AAAϩ domain and the nucleotide-interacting pocket of the DnaA AAAϩ domain (see Fig. 1, B-D) (22,45). However, interactions between Hda and DnaA in the RIDA system have not been identified except for those between the AAAϩ domains of the two (45). As only a limited region containing the Hda Arg finger plays a crucial role in specific interactions with DnaA domain III (45), a region in another DnaA domain might also be required for sustaining strict specificity in binding to Hda.
A structural model for a DnaA⅐Hda⅐DNA complex was constructed, and potential interactions between DnaA domain IV and Hda were identified. DnaA Leu-422 and Pro-423 residues within DnaA domain IV were required for RIDA activity and interaction with Hda in vitro, which is consistent with the results of in vivo analyses. By contrast, these amino acid residues did not play a crucial role in the process of DNA replication initiation. These findings suggest that cross-talk between DnaA domain IV and the Hda AAAϩ domain is required for the formation of an active RIDA complex.
Purification of Mutant DnaA Proteins-Mutant DnaA proteins were purified from KA450 cells bearing pL422A, pL422G, or pP423A as described previously for wild-type and other mutant DnaA proteins (39).
ATP and ADP Binding Assays-The ATP and ADP binding activities of DnaA proteins were determined by a filter retention assay as described previously (39).
Reconstituted RIDA System-The reconstituted staged RIDA assay was performed essentially as described previously (19). First, the DNA-loaded clamps were isolated using a gel filtration spin column as described previously (34). Next, [␣-32 P]ATP-DnaA (0.25 pmol) was incubated at 30°C for 20 min in the presence of 30 M ADP, 10 ng of the isolated DNAloaded clamps, and the indicated amounts of the C-terminal hexahistidine-fused Hda (Hda-cHis) in RIDA buffer (12.5 l) containing 20 mM Tris-HCl (pH 7.5), 8 mM dithiothreitol, 8 mM magnesium acetate, 0.01% Brij-58, 10% glycerol, 0.1 mg/ml bovine serum albumin, and 120 mM potassium glutamate. Nucleotides bound to DnaA were recovered on a nitrocellulose filter, separated by thin-layer chromatography, and quantified by a BAS2500 imaging analyzer (Fujifilm).
Intrinsic ATPase Activity-The DNA-dependent intrinsic ATPase activity of DnaA proteins was assessed as described previously (34). Briefly, [␣-32 P]ATP-DnaA (0.5 pmol) was incubated at 30°C for the indicated time in RIDA buffer (25 l) containing 15 ng of X174 replicative form II DNA (4.3 fmol as a circle). Nucleotides bound to DnaA were monitored by thinlayer chromatography as described for the RIDA reaction.
DNA Binding Activity-The DNA binding activity of DnaA proteins was determined by surface plasmon resonance (SPR) analysis as described previously (46,47). Binding of DnaA to DnaA box was performed as described previously (46). For analysis of DnaA binding to a nonspecific DNA, biotinylated dsDNA bearing a sequence that has no specific affinity for DnaA (48) was prepared by PCR using the primers 5Ј-bio-TCC-TTGAACTATATCGGGCAG-3Ј (where bio is biotin) and its non-biotinylated complementary strand. This DNA (ϳ500 resonance units) was immobilized on the sensor chip SA (Biacore) according to the manufacturer's instructions. DnaA protein was incubated on ice for 15 min in buffer containing 25 mM Hepes-KOH (pH 7.5), 150 mM potassium acetate, 1 mM magnesium acetate, 0.005% Surfactant P20 (Biacore), 0.1 mM ATP, 0.025 g/ml poly(dA-dT)⅐poly(dA-dT), and 0.025 g/ml poly(dI-dC)⅐poly(dI-dC). A sample was injected into the Biacore X flow cells filled with the same buffer at 22°C at a flow rate 40 l/min.
DnaA-Hda Interaction Using Pulldown Assay-The pulldown assay was performed as described previously (45). Hda-cHis (5 pmol) was incubated on ice for 15 min in 12.5 l of NP200-20 buffer containing 5 mM Tris-HCl (pH 7.5), 10% glycerol, 200 mM potassium glutamate, 0.01% Brij-58, 8 mM 2-mercaptoethanol, 8 mM magnesium acetate, and 20 mM imidazole in the presence of 0.1 mM ADP and Co 2ϩ -conjugated magnetic beads (40 g; Invitrogen) equilibrated with the same buffer. The DNA-loaded clamps (1 pmol as the clamp) and the indicated amounts of ATP-DnaA were included, and the mixture was incubated on ice for 10 min. The beads and bound proteins were collected and washed with NP200-20 buffer (25 l). Proteins were eluted in 3 l of NP200-1000 buffer (the same as NP200-20 except for 1 M imidazole) and analyzed by SDS-12% PAGE and silver staining. Band intensities were measured with ImageJ software.
DnaA-Hda Interaction Using SPR Analysis-The Hda binding activity of DnaA proteins was determined by SPR analysis using an NTA sensor chip (Biacore). Buffers used for this analysis were as follows: (i) regeneration buffer (10 mM Hepes-KOH (pH 8.3), 150 mM NaCl, 350 mM EDTA, and 0.005% Surfactant P20; (ii) wash buffer (10 mM Tris-HCl (pH 7.6), 300 mM NaCl, 1 mM EDTA, and 0.05% SDS; (iii) nickel buffer (10 mM Hepes-KOH (pH 7.6), 150 mM NaCl, 50 M EDTA, 0.01% Brij-58, and 5 mM NiCl 2 ; and (iv) eluent buffer (10 mM Hepes-KOH (pH 7.6), 150 mM potassium acetate, 5 mM magnesium acetate, 50 M EDTA, 0.01% Brij-58, 8 mM 2-mercaptoethanol, and 0.1 mM ADP). All reactions were performed at 20°C. The sensor chip contained two flow cells, and one flow cell was used as a reference cell with the N-terminal hexahistidine-fused green fluorescent protein (His-GFP). After washing with regeneration buffer and wash buffer, the NTA surface was saturated with Ni 2ϩ by loading nickel buffer into the two flow cells at a flow rate of 20 l/min for 30 s. Hda-cHis and His-GFP proteins were each injected into one of the flow cells in eluent buffer at 2 l/min until 2500 resonance units was obtained. DnaA protein was diluted in eluent buffer without ADP and with 0.1 mM ATP. Using the KInject command, samples containing DnaA protein were injected at 20 l/min. The association time was 210 s, and the dissociation time was 420 s.
In Vitro Minichromosome Replication Systems-The DnaAdependent replication assay using the minichromosome M13E10 replicative form I DNA and a protein extract prepared from WM433 was performed as described previously (22). The reconstituted replication assay using the minichromosome pBSoriC replicative form I DNA and purified DNA replication proteins was also performed as described previously (46). P1 Nuclease Assay for Open Complex Formation-The P1 nuclease assay was performed as described previously (39).
In Vivo Analysis of Nucleotide-bound DnaA Forms-This in vivo analysis was performed as described previously (22). Cells were grown at 37°C in a phosphate labeling medium containing 0.4 mCi/ml [ 32 P]orthophosphate until the A 660 of the culture reached 0.2 and then further incubated at the same temperature for 90 min in the presence of 1 mM isopropyl ␤-D-(Ϫ)thiogalactopyranoside. Cleared lysates (750 l) from aliquots (2 ml) of the culture were prepared and mixed with anti-DnaA antiserum (5 l) that had been preincubated on ice for 30 min in a cleared lysate (60 l) of KP7364. Protein A-Sepharose (60 l; 50% slurry) was added, suspended at 4°C for 30 min, and washed repeatedly in chilled buffers. After removal of the final washing solution, immunoprecipitates were extracted in a solution (20 l) containing 1 M HCOOH and 5 mM each ATP, ADP, and AMP. Radiolabeled nucleotides were separated using thin-layer chromatography and quantified using an imaging analyzer.

Construction of Structural Model for DnaA Domains III-
IV⅐Hda⅐dsDNA Complex-To identify a novel region required for the DnaA-Hda interaction, a homology model of the E. coli DnaA domains III-IV⅐Hda complex was constructed using the Aquifex aeolicus AMPPCP-DnaA oligomer structure as a starting point (Fig. 1, B and C) (49). The resulting model was consistent with the results of our previous analysis of the DnaA domain III⅐Hda complex structure that suggested specific interactions between Hda Arg-153 Arg finger and DnaA-bound ATP, between Hda Ser-152 and DnaA Arg-334 Arg finger, and between Hda Phe-118 and the DnaA Walker B motif (Fig. 1, B and D) (45). After consideration of the structure of E. coli DnaA domain IV complexed with DnaA box DNA (43), dsDNA was arranged to bind specifically with DnaA domain IV.
The resulting model showed that DnaA Leu-422 and Pro-423 residues within DnaA domain IV would reside on an interface with the Hda AAAϩ domain (Fig. 1, B-D). These residues are located in the L3 loop and ␣4 helix of the DnaA domain IV structure (43), and could interact with a downstream Hda sensor I motif, including Hda Arg-138, Gln-142, and Asn-144 residues.
DnaA L422A/G and P423A Are Insensitive to RIDA-To determine whether DnaA Leu-422 and Pro-423 play an important role in RIDA, we substituted those amino acid residues with alanine. DnaA Leu-422 was also substituted with glycine because glycine is generally considered to be more effective in causing functional change than alanine. Using methods similar to those for wild-type DnaA, DnaA L422A, L422G, and P423A mutant proteins were overproduced in a dnaA-null host strain and purified to Ͼ90% purity as judged by SDS-PAGE and Coomassie Brilliant Blue staining ( Fig. 2A). Mutant protein binding activities for ATP and ADP were comparable with those of wild-type DnaA (Fig. 2B).
To determine the sensitivity of DnaA mutant proteins to RIDA in vitro, a RIDA assay using a reconstituted staged system was used. The DNA-loaded clamps were isolated by gel filtration and incubated in the presence of [␣-32 P]ATP-DnaA and Hda. The results indicated that, unlike wild-type DnaA, DnaA L422A and L422G were not sensitive to RIDA (Fig. 2C). The sensitivity of DnaA P423A to RIDA was decreased (Fig. 2C). Thus, the DnaA Leu-422 residue was crucial for RIDA in vitro. However, the three mutant proteins had intrinsic DNA-dependent ATPase activity levels that were similar to that of wild-type DnaA (Fig. 2D). These results suggested that DnaA Leu-422 and Pro-423 residues specifically functioned in RIDA-specific DnaA-ATP hydrolysis.
To determine whether DnaA mutant proteins interact with the DNA-loaded clamp⅐Hda complex, a RIDA reaction competition assay was performed (45). A small amount of the DNAloaded clamps was incubated with [␣-32 P]ATP-DnaA, high amounts of Hda, and the indicated amounts of non-radiolabeled competitor wild-type or mutant ATP-DnaA. The competitive inhibition of the RIDA reaction caused by DnaA L422A and L422G was less than that caused by wild-type DnaA (Fig.  2E). DnaA P423A inhibited the reaction moderately, but this was still less than the inhibition caused by wild-type DnaA. These results suggested that DnaA Leu-422 and Pro-423 residues functioned in DnaA interaction with the DNA-loaded clamp⅐Hda complex.
DNA Binding Activities of DnaA Mutants-DnaA domain IV is a DNA-binding region that has a high affinity for the DnaA box. In the co-crystal structure reported previously (43), DnaA Leu-422 and Pro-423 residues interact with a 13-mer dsDNA fragment (5Ј-TGTTATCCACAGG-3Ј) containing the DnaA box R1 (bold). In that structure, the main chain NH group of Leu-422 is thought to interact with the phosphate group of T (T11) that is a complementary base to the 11th A (in the 13-mer shown above). The side chain of Pro-423 is postulated to interact with the methyl group of T11 by van der Waals force. In addition, the 1 H-15 N heteronuclear single quantum correlation spectrum of DnaA domain IV in NMR shows a prominent chemical shift perturbation on Leu-422 in the presence of the DnaA box (50). Thus, to determine whether the mutant proteins are affected in their affinities for the DnaA box, they were analyzed with 21-mer dsDNA fragments containing the DnaA box R1 or a nonspecific DNA sequence (nonsense DNA) as a control using SPR.
These results showed that L422G and P423A mutant proteins bound the DnaA box with kinetics similar to the binding observed with wild-type DnaA (Fig. 3A, C, and D). DnaA L422A had a slightly reduced affinity (Fig. 3B), which was consistent with the data of NMR analysis (50). The sustained binding activity of DnaA L422A/G mutant proteins might be caused by the interaction of the Leu-422 main chain (but not the side chain) with the DnaA box. The activity of Pro-423 could be complemented by the substitution of alanine bearing a side chain. Otherwise, the small contribution of Pro-423 to DnaA box binding could be explained by specific interactions supported by van der Waals force.
RIDA is dependent on the clamps loaded on duplex DNA (20), but a specific DNA sequence is not required. Using SPR analysis, the affinity of the DnaA mutants for a nonspecific DNA sequence was determined. Given that DnaA binding to nonspecific DNA is weaker than its binding to DnaA box DNA (48), large amounts of nonsense DNA (21-mer) were immobilized on a sensor chip. Wild-type DnaA protein (9 -150 nM) showed a dose-dependent increase in binding to nonsense DNA (Fig. 3E). When analyzed using an intermediate concentration (75 nM), the resonance patterns of DnaA mutant proteins were similar to those of wild-type DnaA (Fig. 3F). Taken together, the nonspecific DNA binding activity of DnaA mutants supports the idea that the RIDA insensitivity of DnaA mutants was not due to a decrease in their affinity for DNA. These results are consistent with the model (Fig. 1, B and D) and the results of NMR analysis (50), which show that these amino acid residues make no contact with nonspecific DNA.
Hda Binding Activities of DnaA Mutants-To more quantitatively analyze the DnaA interaction, pulldown assays using His-tagged Hda and Co 2ϩ -conjugated magnetic beads were performed. In this assay, DnaA is recovered in an ADP-Hda dose-dependent manner in the presence of the DNA-loaded clamps (19). Binding specificity is supported by the data showing that DnaA R334A (AAAϩ sensor II) and Hda Q6A (the clamp-binding site) mutants (which are defective in Hda interaction and clamp binding, respectively) severely reduce the recovery of DnaA in these assays (45) (Fig. 4A). When DnaA L422A/G proteins were similarly assessed, the recovery of these mutants was severely reduced compared with wild-type DnaA (Fig. 4A), suggesting that the DnaA Leu-422 residue is crucial for Hda interaction. The residual binding of these mutants was 10 -20% of wild-type DnaA-binding, which can be explained by possible interactions between AAAϩ domains of DnaA and Hda. The recovery of DnaA P423A was roughly 50% of that of the wild-type DnaA (Fig. 4A). Notably, these results are consistent with the results showing that RIDA sensitivities of DnaA L422A/G and P423A are 10 -15% and about 50% of that of the wild-type DnaA, respectively (Fig. 2C). Thus, these results are effective for quantitative understanding of DnaA domain IVdependent interactions with Hda.
Next, SPR analysis was used to determine direct binding modes of wild-type and mutant DnaA proteins to Hda. Histagged Hda and His-tagged GFP (used as a background control) were separately bound to a nickel-NTA-coated chip. The DnaA-Hda interaction was detected when the concentration of wild-type DnaA was 12.5-100 nM (Fig. 4B). Because the reaction was not saturated even at 100 nM DnaA, precise kinetic parameters could not be calculated. Thus, in agreement with the pulldown data, the direct binding of DnaA to Hda is probably weak in the absence of DNA-loaded clamps. At 50 nM (within the linear range for the reaction), DnaA L422A and L422G exhibited reduced Hda binding activity relative to that of wild-type DnaA (Fig. 4C). The affinity of DnaA P423A was moderately reduced (Fig. 4C), which was consistent with the RIDA activity data (Fig. 2C). Therefore, Leu-422 and Pro-423 residues had significant and specific roles in the direct DnaA-Hda interaction.
Residual activities of DnaA L422A/G and P423A in Hda binding (Fig. 4, A and C) can be explained by direct interactions between the AAAϩ domains of DnaA and Hda. Previous studies show that specific binding of DnaA to Hda requires several amino acid residues within the AAAϩ domains of both Hda and DnaA, including Hda Phe-118 (H-finger), Asn-122 (E-finger), Arg-153 (Arg finger), and DnaA Arg-334 (sensor II) (Fig. 1, B-D) (34,45). These interactions as well as possible nonspecific interactions conceivably cause the low levels of binding activity between DnaA and Hda even in the absence of DNA-loaded clamps. The DnaA Leu-422/Pro-423-mediated interaction might assist or regulate the functional interactions of these residues (see "Discussion").   AUGUST 19, 2011 • VOLUME 286 • NUMBER 33

JOURNAL OF BIOLOGICAL CHEMISTRY 29341
DnaA Mutants Are Active in Minichromosome Replication and Formation of Initiation Complexes and ADP Release-The function of the mutant proteins during replication initiation at oriC in vitro was first characterized using an in vitro DnaA complementation system with a crude protein extract and an oriC plasmid. The replication activity and its ATP-dependent regulation of DnaA L422A and P423A were similar to those of wild-type DnaA (Fig. 5A). DnaA L422G exhibited a slightly higher replication activity than wild-type DnaA (Fig. 5A) that might have been caused by inhibition of RIDA activity contained in the crude extract (17). Next, we used a reconstituted replication system using purified replication proteins and found that DnaA L422G and P423A had replication activities that were similar to those of the wild type (Fig. 5B). DnaA L422A had reduced replication activity (Fig. 5B), reflecting the decreased DnaA box binding activity of this protein (Fig. 3B).
Open complex formation was assessed using an oriC plasmid and P1 nuclease, which specifically cleaves single-stranded DNA. If the oriC-unwinding region on the plasmid were cleaved by P1 nuclease, subsequent digestion with AlwNI restriction enzyme would produce two DNA fragments of 3.8 and 4.1 kb (39). The results indicated that, like wild-type DnaA, DnaA L422G and P423A were active and regulated in an ATP binding-dependent manner during open complex formation (Fig. 5C). DnaA L422A exhibited a slightly reduced activity relative to wild-type DnaA (Fig. 5C) consistent with the results of the reconstituted replication assay (Fig. 5B). These results suggested that the DnaA mutants were active in the DNA replication initiation process, including formation of initial complexes and DnaB helicase loading on oriC.
The inactive ADP-DnaA yielded by RIDA is reactivated to ATP-DnaA by the nucleotide exchange activity of DnaA-reactivating sequences (DARS1 and -2) on the chromosome that are required for timely DNA replication initiation during the cell cycle (51). DARSs bear a DnaA box cluster promoting formation of a specific DnaA multimer and dissociation of ADP bound to DnaA (51). DnaA L422A/G and P423A proteins were substantially active in the DARS1-dependent DnaA-ADP dissociation in vitro (data not shown). The combined results support the proposed function of DnaA Leu-422 and Pro-423 residues specifically in RIDA-dependent DnaA-ATP hydrolysis.
DnaA L422A/G and P423A Proteins Inhibit Cell Growth-Certain dnaA mutants, such as dnaAcos and dnaA29 (R334A), are inactive in RIDA and cause oriC-dependent overinitiation of chromosomal replication, resulting in inhibition of cell growth and colony formation (22,52,53). In the absence of rnhA, an alternative oriC-independent replication system operates that allows cells without oriC to grow even in the presence of overinitiating dnaA mutant alleles (18,22,52,54). To investigate in vivo RIDA activity of DnaA L422A/G and P423A, we asked whether these DnaA mutants exhibit oriC-dependent inhibition of colony formation ( Table 1). Introduction of the dnaA L422A/G and P423G mutant alleles on a pING1 (vector) derivative inhibited or impeded colony formation when the host cells (KH5402-1, YT411, and KA451) were active for chromosomal replication from oriC even in the absence of rnhA (Table 1). In particular, colony formation was severely inhibited in the chromosomal dnaA-disrupted cells (KA451), indicating  30 min (B). C, oriC unwinding activity. The indicated amounts of the ATP-or ADP-bound forms of wild-type and mutant DnaA proteins were incubated at 38°C for 3 min in buffer containing M13KEW101 oriC plasmid (7.9 kb) and HU protein followed by further incubation with P1 nuclease. The resultant DNA was purified, digested with AlwNI, and analyzed using 1% agarose gel electrophoresis and ethidium bromide staining. The amounts of 3.8and 4.1-kb fragments detected were normalized to the amount of total DNA and were plotted as an open complex (%).
that the inhibition was caused only by expression of the mutant dnaA alleles but not by co-expression of the wild-type and mutant dnaA alleles.
By contrast, introduction of those alleles did not inhibit colony formation when the host cells lacked oriC in the absence of rnhA (KA429 and KA450) ( Table 1). In addition, expression levels of wild-type and mutant DnaA proteins in KA450 were comparable (supplemental Fig. 1), suggesting that the inhibition of colony formation was not due to simple overexpression of the mutant proteins.
These results suggested that DnaA L422A, L422G, and P423A mutant proteins specifically affected DNA replication initiation at oriC and were consistent with the idea that the mutants promoted overinitiation in vivo due to a specific defect in RIDA. The dnaA P423A mutant inhibited cell growth at a level that was moderate compared with that caused by dnaA L422A/G. This observation was also consistent with the reduced activity of this allele in RIDA in vitro.
Even when the incubation time after transformation of cells was extended to 20 h, colony formation efficiency of KH5402-1 bearing pL422A or pL422G was only 5% of that by pKA234, and even the colonies that were formed contained cells that could not form new colonies (data not shown). Colony formation of KH5402-1 bearing pP423A was also inhibited (Table 1); cells in the colonies could grow but only at a 2-fold slower rate compared with KH5402-1 cells bearing pING1 or pKA234 (data not shown).
dnaA L422A/G and P423A Mutants Promote Accumulation of ATP-DnaA-In cultures of wild-type cells, the ATP-bound form of DnaA represents 10 -20% of the total ATP/ADP-DnaA content (21,22). As constitutive expression of the mutant genes inhibits the growth of KH5402-1 host cells (Table 1), controlled expression of the mutant alleles was achieved by placing them downstream of the lac promoter on a pBR322 derivative in lacI q cells. To assess the relative ratio of ATP/ADP forms of the mutant DnaA proteins in vivo, cells were grown in a synthetic medium containing [ 32 P]orthophosphate. DnaA protein was isolated by immunoprecipitation, and the nucleotides bound to the protein were analyzed by thin-layer chromatography (22).
When DnaA L422A expression was induced for 90 min in the presence of 1 mM isopropyl ␤-D-(Ϫ)-thiogalactopyranoside, the proportion of ATP-DnaA increased from 21 to 61% (Fig. 6). Similarly, the levels of the ATP forms of DnaA L422G and P423A increased from basal levels to 70 and 53%, respectively, in the presence of isopropyl ␤-D-(Ϫ)-thiogalactopyranoside ( Fig. 6). However, induction of the wild-type protein was not associated with an increase in the level of ATP-DnaA (Fig. 6), which is consistent with previous data (22). These results suggested that the mutant proteins were defective in RIDA in vivo.

DISCUSSION
A structural model of the Hda⅐DnaA domains III-IV⅐DNA complex was used to ascertain amino acid residues within DnaA domain IV that specifically interact with Hda. Mutant analyses revealed that DnaA Leu-422 and Pro-423 residues are important for RIDA-dependent ATP hydrolysis, formation of a complex with the DNA-loaded clamp⅐Hda complex, and Hda interaction (Figs. 2 and 4). Consistently, these amino acid residues were found to be required for DnaA-ATP hydrolysis in vivo ( Fig. 6 and Table 1). These mutants were not defective in    (45). In the DnaA homologues of these species, the amino acid residues that correspond to E. coli DnaA Leu-422 and Pro-423 are highly conserved (data not shown) (57). Hence, the proposed interaction and regulation models might apply to other bacterial species that possess an Hda orthologue. The proposed structural model indicates that DnaA Leu-422 and Pro-423 residues are located near the Hda sensor I motif (Ser-127 to Pro-140) (Fig. 1, B-D). In particular, Hda Arg-138, Gln-142, and Asn-144 residues might be exposed on the surface of Hda and be able to interact with DnaA Leu-422 and Pro-423. The sensor I motif of AAAϩ chaperones, such as Saccharomyces cerevisiae Hsp104 and the p97/ valosin-containing protein-like ATPase from Thermoplasma acidophilum, play a specific role in catalyzing ATP hydrolysis (58 -60). In addition, the Arg finger (Arg-153) and its spatially neighboring residues in Hda play a crucial role in RIDA and the interaction with ATP-DnaA (most likely with ATP itself and the ATP-binding Walker B motif (34,45)). Thus, it is plausible that the interaction between DnaA Leu-422/Pro-423 and the areas near Hda sensor I in addition to that between ATP-DnaA domain III and the Hda Arg finger supports a specific conformation of the DnaA⅐Hda complex that allows DnaA-ATP hydrolysis (Fig.  7B). The interactions might indirectly promote a conformational change of the Hda Arg finger, thereby affecting DnaA-ATP hydrolysis and the affinity of the two proteins. The x-ray and NMR analyses of the DnaA domain IV⅐DnaA box DNA complex show that DnaA Leu-422 is involved in specific binding to the DnaA box DNA (43,50). In the crystal structure, the main chain of Leu-422 also interacts with a phosphate group of the DnaA box DNA, which is mediated by DNA bending at 28° (Fig. 7A) (43). Thus, substitution of the side chain of the Leu residue still allowed DnaA box binding activity (Fig. 3, B and C). DNA bending by DnaA binding to the DnaA box is also observed in solution (48). A model in which the DnaA box DNA acts as a physical obstacle that inhibits the interaction between DnaA Leu-422/Pro-423 residues and the surface of the Hda AAAϩ domain is proposed (Fig. 7, A and C). In fact, ATP-DnaA molecules are hydrolyzed by RIDA in vivo, and ϳ20% (200 -400 molecules) of 1000 -2000 DnaA molecules present in the cell are present as ATP-DnaA (21). This could be explained if ATP-DnaA molecules bound to the DnaA box DNA are insensitive to RIDA in vivo. There are ϳ300 DnaA boxes on the E. coli chromosome (61). Maintaining a basal level of ATP-DnaA may be relevant to efficient origin firing in the next round of replication initiation and ATP-DnaA-specific transcriptional regulation of genes like dnaA and the nrd operon (47,62).
There are examples where a DNA-binding region in a protein plays a role in protein-protein interactions other than DNA binding. The DNA-binding region of Cdc6, a replication-licensing factor in eukaryotic and archaeal cells, interacts with and regulates the MCM helicase (63,64). Additionally, the DNA-binding regions of OmpR and PhoB, which are transcriptional regulators in E. coli, are suggested to contain an interaction site for a subunit of RNA polymerase and to regulate directly RNA polymerase activity (65,66). The protein-protein interaction on a DNA-binding region may function for the direct recognition of conformational changes within the region induced by DNA binding. In the RIDA reaction, DnaA domain IV likely interacts with the clamp-loaded DNA to form an active RIDA complex (20). The cross-talk between the Hda AAAϩ domain and DnaA Leu-422/Pro-423 within domain IV may serve to allow Hda to recognize the interaction of DnaA domain IV with the clamp-loaded DNA.