A novel mode of DnaA-DnaA interaction promotes ADP dissociation for reactivation of replication initiation activity.

ATP-DnaA is temporally increased to initiate replication during the cell cycle. Two chromosomal loci, DARS (DnaA-reactivating sequences) 1 and 2, promote ATP-DnaA production by nucleotide exchange of ADP-DnaA for timely initiation. ADP-DnaA complexes are constructed on DARS1 and DARS2, bearing a cluster of three DnaA-binding sequences (DnaA boxes I-III), promoting ADP dissociation. Although DnaA has an AAA+ domain, which ordinarily directs construction of oligomers in a head-to-tail manner, DnaA boxes I and II are oriented oppositely. In this study, we constructed a structural model of a head-to-head dimer of DnaA AAA+ domains, and analyzed residues residing on the interface of the model dimer. Gln208 was specifically required for DARS-dependent ADP dissociation in vitro, and in vivo analysis yielded consistent results. Additionally, ADP release from DnaA protomers bound to DnaA boxes I and II was dependent on Gln208 of the DnaA protomers, and DnaA box III-bound DnaA did not release ADP nor require Gln208 for ADP dissociation by DARS-DnaA complexes. Based on these and other findings, we propose a model for DARS-DnaA complex dynamics during ADP dissociation, and provide novel insight into the regulatory mechanisms of DnaA and the interaction modes of AAA+ domains.

INTRODUCTION

Initiation of chromosomal replication is rigidly regulated to take place in a timely manner during the cell cycle. In Escherichia coli, the initiator protein DnaA plays a crucial role in initiating replication at the oriC chromosomal origin (1–3). DnaA is a member of the AAA+ ATPase superfamily, and stably binds ATP or ADP (1,4–7). ATP-DnaA, but not ADP-DnaA, is the active form during initiation. DiaA (DnaA initiator-associating protein) is homotetrameric and binds a few molecules of DnaA, which stimulates ATP-DnaA oligomerization on oriC and replication initiation (8,9). The minimal oriC region consists of an AT-rich duplex unwinding element (DUE), and DnaA-oligomerization region (DOR) bearing 12 specific DnaA-binding sites (DnaA boxes) with various affinities (Figure 1A) (2,3). Also, integration host factor (IHF), a member of the nucleoid-associated protein family, binds to a specific IHF-binding region (IBR), which partly overlaps the τ1 DnaA box (10). IHF binding to IBR prevents DnaA binding to this site. Also, IHF binding introduces a sharp bend in the DNA, promoting DUE unwinding and replication initiation (11–13).

Figure 1.

Structure of oriC, DnaA, the initiation complex, DARS1, and DARS2. (A) Basic structure of oriC. The oriC region consists of the duplex unwinding element (DUE) and the DnaA-oligomerization region (DOR). The DUE bears AT-rich 13-mer repeats (L, M, R), while DOR bears a single IHF-binding region (IBR) and 12 DnaA boxes (R1 and 2, R4, R5M, τ1 and 2, I1−3, and C1−3). IBR includes the IHF-binding consensus sequence and partially overlaps DnaA box τ1 (10). Black triangles represent DnaA boxes with the complete consensus sequence (5′-TT[A/T]TNCACA). Dark triangles represent DnaA boxes with the consensus sequence containing one or a few mismatches. These triangles also show the directions of the DnaA box sequences. (B) Model of the initiation complex with unwound DUE (the open complex). DnaA domains are shown (also see panel C). DnaA oligomers are constructed via domain III-domain III interactions in a head-to-tail manner, in which bound ATP interacts with the Arg285 finger motif of the flanking protomer. IHF causes DNA to bend sharply. The T-rich strand of the single-stranded DUE (ssDUE) binds to the Val211 and Arg245 residues of DnaA protomers bound to the R1−I2 region. DiaA is omitted for simplicity. (C) Features of DnaA domains. Domains I−IV are shown with amino acid residue numbers. Domain III is subdivided into IIIa and IIIb. Domain IIIb (residues 296−373) is connected to the IIIa subdomain via a loop (17). The ATP-binding Walker motifs are indicated along with residues relevant to this study. Secondary structures are shown as cylinders for α-helices and arrows for β-strands. α6, the α6 helix. (D) Basic structures of DARS1 and DARS2. DnaA boxes are shown as described for panel A. The DARS core contains three DnaA boxes, of which two are oriented in a head-to-head manner. DARS2 has a regulatory region containing an IBR and a Fis-binding region (FBR). (E) Sequences of the DARS1 and DARS2 core regions. DnaA box sequences (I, II, and III) are highlighted by red characters and are underlined. DnaA box sequences different from the consensus are indicated by dark red characters.

DnaA consists of four functional domains (Figure 1B, C). Domain I bears specific sites for self-dimerization and interactions with other proteins such as DiaA and DnaB helicase (5,9,14). Domain II is a flexible linker (14,15). Domain III has specific AAA+ motifs for ATP recognition (Walker A/B, Sensor I/II, and Arg-finger) in addition to DnaA-specific motifs for stimulating ATP-DnaA oligomerization (AID-1/2) and DUE unwinding (H/B motifs) (4–6,16–22). Like typical AAA+ proteins, ATP-DnaA is believed to form spiral oligomers on oriC, in which Arg-finger residue Arg285 interacts with ATP bound to the adjacent protomer, resulting in a head-to-tail DnaA–DnaA interaction (Figure 1B, C) (10,20,23–25). DnaA AID-1 Arg227 and AID-2 Leu290 assist the DnaA–DnaA interaction in the DOR R1-I2, but not significantly in DOR R4-C3 (22). H/B motifs (Val211/Arg245) of DnaA bound to the R1 and R5M sites bind specifically to the T-rich strand of unwound DUE, stabilizing the single-stranded (ss) state, which promotes DnaB helicase loading (Figure 1B, C) (10,13,21,26). Val211 is included in the AAA+ initiator-specific motif (ISM) (4,5,26). Domain IV is a DNA-binding domain for specific recognition of the 9-mer DnaA box, which has the consensus sequence TT(A/T)TNCACA (27,28). In DnaA-DOR complexes, the binding direction of the DnaA protomer relative to this asymmetric sequence has been determined (Figure 1A, B) (24).

The cellular level of ATP-DnaA fluctuates with the peak at the time for replication initiation, which is tightly regulated by negative and positive regulatory systems (3,29–31). DnaA-bound ATP is hydrolyzed and converted to ADP-DnaA by at least two systems, regulatory inactivation of DnaA (RIDA) and datA-dependent DnaA-ATP hydrolysis (DDAH), both activated after replication initiation. RIDA is promoted by construction of the complex formed between Hda and the DNA-loaded clamp subunit of DNA polymerase III holoenzyme, which interacts with ATP-DnaA (32,33). DDAH is promoted by the complex formed between IHF and chromosomal locus datA, which contains a cluster of DnaA boxes (34,35). IHF-datA binding occurs temporally after replication initiation.

By contrast, the DARS system produces ATP-DnaA before replication initiation by promoting nucleotide exchange of ADP-DnaA to ATP-DnaA (36–39). This system is based on at least two chromosomal loci termed DARS1 and DARS2 (Figure 1D, E). The two DARSs share a highly conserved core region containing three DnaA boxes (I, II, and III) that are essential for nucleotide exchange. ADP-DnaA proteins construct specific dynamic complexes on these loci, which promotes ADP dissociation in the complexes. The resultant apo-DnaA dissociates from the DARS-bound complexes and binds ATP, reactivating DnaA molecules (36). DARS2, the predominant DARS element in vivo, is activated in a timely manner by nucleoid proteins Fis and IHF (Figure 1D) (37). DARS2-IHF binding occurs temporally before replication initiation.

Whereas many AAA+ proteins are involved in various mechanisms including DNA replication, recombination, and membrane dynamics, they typically form oligomers via a head-to-tail interaction of the AAA+ domains (4,16). However, in the essential cores of DARSs, DnaA boxes I and II are oppositely oriented (Figure 1D, E), unlike DnaA boxes II and III, which are oriented in the same direction, like DnaA boxes R1 to I2 and R4 to C3 in the DOR (Figure 1B). DnaA self-associates in a head-to-tail manner at these DnaA boxes facing in the same direction, and the direction of the DnaA boxes appears to play an important role in facilitating the head-to-tail interaction. Although DnaA boxes I and II are essential for the DARS system (36), the fundamental mechanisms by which ADP-DnaA molecules bind to these sites, and possibly interact with each other, largely remain to be explored.

While carrying out this study, a head-to-head dimer structure within the c-di-GMP-bound FleQ AAA+ hexamer from Pseudomonas aeruginosa was reported (40). FleQ is a transcriptional regulator that interacts with both a specific enhancer DNA and the sigma factor. Apo-, ADP- and ATP-γ-S-bound FleQ hexamers are constructed via typical head-to-tail interactions of AAA+ domains, but when c-di-GMP binds to the hexamer, structural changes occur, resulting in head-to-head dimers of the AAA+ C-terminal subdomains within the AAA+ hexamer. However, these features are largely different from those of DARS-DnaA complexes that promote ADP dissociation and do not construct stable hexamers, nor bind c-di-GMP.

In this study, we constructed a structural model of a head-to-head dimer of DnaA AAA+ domains using previous crystal structure analysis (21). In the model, the head-to-head dimer is constructed from AAA+ N-terminal subdomains (domain IIIa; Figure 1C). Based on the model, we analyzed the functions of specific residues predicted to reside on the protein-protein interface, and found that DnaA Gln208 is required for DARS-dependent ADP dissociation. In vivo analysis using the DnaA Q208A mutant supported the significance of this residue in DARS systems, rather than initiation mechanisms at oriC per se. Furthermore, we investigated the specific roles of DnaA protomers bound to DARS1 and found that DnaA bound to DnaA boxes I and II, but not DnaA box III, promotes ADP dissociation dependently on Gln208. In addition, DiaA moderately stimulated DARS1-dependent ADP dissociation. Based on these and other findings, we propose a mechanistic model of DnaA–DnaA interactions on DARS for dissociating ADP. The results provide novel insight on the regulatory mechanisms of DnaA for replication initiation, and the head-to-head interaction of AAA+ domains.

MATERIALS AND METHODS

Bacterial strains, plasmids, and DNA fragments

Strain KH5402-1 (relevant genotype: wild-type [WT] dnaA) and its derivative KA451 (dnaA750::Tn10 rnhA::cat) have been described previously (19,21), as have derivatives of mini-R plasmids pRRNH (rnhA) and pOZ18 (rnhA, WT dnaA) which bear the pemI-pemK stable maintenance system (9,19,22). The pRS1 (dnaA Q208A) derivative of pOZ18 was constructed by PCR using mutagenic primers Q208A-1 and Q208A-2 (Supplementary Table S1). The pBR322-derivative pKX11 and pACYC177-derivative pOA61, both of which bear DARS2, have been described previously (36). A 1.2 kb DNA fragment carrying a kanamycin resistance gene was amplified using pACYC177 and primers BamHI-kan-f and SphI-kan-r (Supplementary Table S1), digested with BamHI and SphI, and ligated to BamHI-SphI fragments of pBR322 and pKX11, resulting in pBR322-kan and pKX11-kan, respectively. MD1-8 DNA (WT DARS1) and its derivatives, as well as Nonsense DNA (MD1-10) bearing a nonsense sequence instead of DnaA boxes I−III (Supplementary Table S1), were constructed by annealing synthetic DNA with complementary sequences. The Nonsense DNA sequence, which excludes DnaA-specific binding, was analyzed previously (10,24,27).

DnaA overproduction and purification

Plasmids for overproducing mutant DnaA protein were derivatives of pKA234, described previously (41). Mutations were introduced into pKA234 using mutagenic primer pairs Q208A-1 and Q208A-2, and Q215A-1 and Q215A-2 (Supplementary Table S1), resulting in pKA234-Q208A and pKA234-Q215A, respectively. Plasmids pKA234-L214R and pKA234-F239A have been previously described (21,42). Overproduction and purification of DnaA proteins were performed as previously described (21,24), as was expression and purification of DnaA V211A and DnaA L290A (21,22).

ADP dissociation assay

The ADP dissociation assay was performed essentially as described previously (36,37). Briefly, [3H]ADP-DnaA was prepared by incubation of apo-DnaA on ice for 15 min in buffer N (50 mM HEPES–KOH pH 7.6, 2.5 mM magnesium acetate, 0.3 mM EDTA, 7 mM dithiothreitol, 20% v/v glycerol, 0.007% v/v Triton X-100) containing 3 μM [3H]ADP. When DARS1 or chimeric DARS1 was used, the resultant [3H]ADP-DnaA (2 pmol) was incubated in 25 μL of dissociation buffer (20 mM Tris–HCl pH 7.5, 100 mM potassium glutamate, 10 mM magnesium acetate, 2 mM ATP, 8 mM dithiothreitol, 100 μg/mL bovine serum albumin) containing 150 ng of poly(dI-dC) and the indicated amounts of DARS1 or chimeric DARS1. DnaA-bound [3H]ADP was recovered on nitrocellulose filters and analyzed using a liquid scintillation counter. When DARS2 was analyzed, the indicated amounts of DARS2 plasmid pOA61 and 50 fmol each of IHF and Fis were similarly incubated in the absence of DARS1. Where indicated, DiaA (0.25 pmol as tetramer, 1 pmol as monomer) was also co-incubated.

In vitro oriC replication assay

The in vitro oriC replication assay was performed essentially as previously described (9,35,43). Briefly, reaction mixtures (25 μL) contained crude replicative extract prepared from strain WM433 (dnaA203), M13KEW101 oriC plasmid RF I (200 ng; 600 pmol as nucleotide), rNTPs, dNTPs including [α-32P]dATP and DnaA.

Electrophoretic mobility shift assay (EMSA)

EMSA experiments were essentially performed as described previously (10,24,37). Briefly, DNA fragments (50 nM) were incubated at 30°C for 10 min in 10 μL of GS buffer (20 mM HEPES-KOH pH 7.6, 100 mM potassium chloride, 20 mM magnesium acetate, 1 mM EDTA, 8 mM dithiothreitol, 100 μg/mL bovine serum albumin, 10% glycerol) containing ADP-DnaA and 50 ng of λ phage DNA as a competitor, followed by 6% polyacrylamide gel electrophoresis (PAGE) at 100 V for 85 min at room temperature in Tris–borate buffer. DnaA was visualized using GelStar staining.

Flow cytometry analysis

These experiments were essentially performed as described previously (8,10,37). Briefly, cells were grown exponentially for at least ten generations at 37°C in LB medium. Portions of the cultures were withdrawn for analysis of cell mass (cell size) and other potions were further incubated for four hours in the presence of rifampicin and cefalexin, followed by DNA staining with SYTOX Green and analysis with a FACS Calibur flow cytometer.

RESULTS

Structural model of the putative head-to-head DnaA complex

Unlike DnaA boxes on oriC, the opposite orientation of DnaA boxes I and II in DARS1 and DARS2 suggests that a head-to-head interaction by DnaA molecules occurs at these sites (Figure 1D, E). Given that ADP dissociation basically takes place even with truncated DnaA bearing only domains III and IV (36), DnaA domain III regions bearing AAA+ motifs likely interact with each other to form the predicted head-to-head homodimer. Notably, we previously observed a head-to-head orientation in a crystal structure of an ADP-bound DnaA domain III from the hyperthermophile Thermotoga maritima (21; PDB 2Z4R). In this structure, the two DnaA molecules are closely associated with each other, primarily through the α6 helix residing on the N-terminal subdomain (domain IIIa) of the conserved AAA+ domain (Figure 1C). Distance between the C-terminal regions of the monomers in the model complex was 6.5–9 nm which corresponds to the length of 19–26 bp dsDNA (Supplementary Figure S1). Considering that 9-bp DnaA boxes I and II are separated by 5-bp space and that domain IV is connected with domain III via a short flexible linker, this distance could be reasonable in allowing a head-to-head dimer formation of the DnaA molecules. We therefore focused our attention on the amino acid residues that constitute the monomer-monomer interface, and investigated the biological significance of the corresponding residues in Escherichia coli DnaA, using a homology model of the E. coli DnaA domain III head-to-head homodimer (Figure 2A).

Figure 2.

Structural model for the head-to-head dimer of DnaA domain III (AAA+ domain). (A) Homology model of the head-to-head dimer of E. coli DnaA AAA+ domain III. The crystal structure of the head-to-head dimer of T. maritima DnaA AAA+ domain III (21; PDB 2Z4R) was used as a template for homology modeling with an analyzing software MOLFEAT. The overall protein structure is shown in ribbon representation, with different colors for each structurally identical protomer. The monomer-monomer interface is enlarged (insert). Residues positioned at the monomer-monomer interface are shown as sticks and colored differently, as are functionally important residues. ADP is also shown in stick representation. The nucleotide-binding pocket is located at the boundary of AAA+ N-terminal and C-terminal subdomains IIIa and IIIb. The same model is shown also in space filling representation (Supplementary Figure S1). (B) Homology model of the head-to-tail dimer of E. coli DnaA AAA+ domains. The overall protein structure and residues residing at the monomer-monomer interface are similarly shown (22; PDB 3R8F).

Residues exposed at the interface of the putative head-to-head complex

We examined the DnaA–DnaA interface in the model, and searched for amino acid residues that may play important roles in the head-to-head interaction. Gln208, Val211, Leu214, Gln215, Phe239 and Lys243 were identified as potential candidates, all of which are exposed at the interface of the head-to-head complex in the model structure (Figure 2A and Supplementary Figure S1). A model of the head-to-tail dimer of DnaA domain III suggests that these residues are not present on the interacting surface of the complex (21) (Figure 2B). Among these residues, DnaA V211A and DnaA K243A inhibit DUE unwinding (21), while Val211 (H-motif or ISM) is crucial for ssDUE binding, and K243A is suggested to cause intrinsic conformational changes in ATP-DnaA oligomers constructed on oriC DNA.

Affinity of mutant DnaA proteins for ADP

DnaA proteins bearing Q208A, L214R, Q215A or F239A, as well as WT DnaA, were purified, and only Leu, a relatively small residue, was substituted with Arg bearing a large side chain. Filter retention assays showed that the affinities of DnaA Q208A, L214R and Q215A for ADP were comparable to those of WT DnaA (Supplementary Table S2). DnaA F239A appeared to exhibit slightly impaired ADP binding. The flanking residue of Phe239 is another Phe, and the bulky side chains of Phe239 and Phe240 are located only two residues away from the Walker B motif (Asp235−Asp236; Figure 1C). Thus, the F239A substitution may indirectly affect the structure of this motif. DnaA V211A and DnaA K243A have been purified previously with the intact affinity for ADP (21).

DnaA Q208A displays impaired ADP dissociation but is active in initiation

To analyze responses to DARS, we used reconstituted systems of DARS1- and DARS2-dependent ADP dissociation of DnaA that included a 70 bp DARS1 fragment or a 455 bp DARS2 fragment supporting full ADP dissociation activity (36,37). First, to select specific mutants we used DARS1, a simpler system than DARS2. ADP-bound DnaA proteins were incubated with DARS1 DNA, then subjected to filter retention assays. Compared with WT DnaA, DnaA Q208A displayed considerable inhibition in ADP dissociation (Figure 3A); 25 fmol of DARS1 dissociated ∼90% of ADP from WT DnaA, but only ∼40% from DnaA Q208A. We previously reported that DnaA L290A is inactive in DARS2-dependent ADP dissociation (37). The present data showed that this residue is also crucial for DARS1 (Figure 3A). By contrast, ADP dissociation activities of DnaA V211A, L214R, and Q215A were essentially sustained (Figure 3B). DnaA F239A and DnaA K243A showed moderate inhibition, suggesting that these residues might play a stimulatory role in ADP dissociation. As mentioned above (Supplementary Table S2), DnaA F239A could partially alter the intact structure of the nucleotide-binding pocket, causing indirect inhibition of ADP dissociation.

Figure 3.

ADP dissociation activities of mutant DnaA proteins. (A and B) Activities of DARS1. [3H]ADP-bound forms of WT and mutant DnaA (2 pmol) were incubated for 15 min at 30°C with the 70 bp MD1−8 DNA including the DARS1 core (DARS1) or the 70 bp Nonsense DNA excluding the DARS1 core (Non). The amount of DnaA-bound ADP was assessed by a nitrocellulose filter retention assay, and the proportions (%) of the amounts present in the absence of DNA are plotted. Experiments were performed at least twice, and averages with standard deviations (SDs) are shown with error bars. (C) Activities of DARS2. [3H]ADP-bound forms of WT and DnaA Q208A (2 pmol) were incubated for 15 min at 30°C with pOA61 DNA including the DARS2 (DARS2) or empty plasmid (vector) in the presence of IHF and Fis. The amount of DnaA-bound ADP was similarly assessed. (D and E) Replication initiation activity. In vitro complementation assays were performed using a plasmid bearing oriC (200 ng, 600 pmol as nucleotide) and a crude extract prepared from a dnaA mutant strain (WM433) containing all replication proteins except DnaA. ATP and ADP forms of WT and Q208A DnaA were incubated at 30°C for 20 min (D) or various durations (E). For time-course experiments (E), 1 pmol of DnaA was used for each reaction.

Based on these results, we further analyzed DnaA Q208A using a reconstituted DARS2 system including IHF and Fis (37). DnaA Q208A displayed moderately impaired ADP dissociation (Figure 3C), consistent with the response to DARS1. In addition, we analyzed this mutant using an in vitro oriC replication system using a crude replicative extract. Whereas ADP forms of WT DnaA and DnaA Q208A were inactive, ATP forms of the two proteins were fully active (Figure 3D, E). These results demonstrate the specific importance of the Gln208 residue in ADP dissociation by DARS systems, and are essentially consistent with the head-to-head complex model (Figure 2A).

DiaA diminishes DnaA Q208 activity in ADP dissociation

DiaA has a high affinity for DnaA molecules, and thereby stimulates assembly of DnaA molecules on oriC (8,9). This protein forms a homotetramer, and each protomer bears a specific binding site (including DiaA Leu190) for DnaA domain I (8). Binding of the DiaA tetramer to multiple DnaA molecules stimulates the cooperative binding of DnaA molecules on the oriC low-affinity DnaA boxes (9). DiaA itself does not substantially bind DNA or nucleotides. In cells, DiaA might affect the function of DARSs in addition to oriC. While DnaA domains I and II are basically dispensable for DARS1-dependent ADP dissociation, the presence of those moderately stimulates the ADP dissociation (37). DnaA domain I bears weak activity to stimulate self-oligomerization (45). Thus, we analyzed DARS-dependent ADP dissociation activities of WT and mutant DnaAs in the presence of DiaA (Figure 4).

Figure 4.

ADP dissociation activities of DnaA proteins in the presence of DiaA. (A−C) Activities of WT DnaA, DnaA Q208A, and L290A with DARS1 in the presence of DiaA. Experiments were performed as described for Figure 3A and B except for the addition of DiaA (0.25 pmol as tetramer, 1 pmol as monomer). DiaA does not directly bind DNA or ADP. These experiments were performed at the same time using WT DnaA and for visibility, the data are shown separately using the identical data of WT DnaA with or without DiaA in each panel. (D) Activities of WT DnaA and DnaA Q208A with DARS2 in the presence of DiaA. Experiments were performed as described for Figure 3C except for the addition of DiaA (0.25 pmol as tetramer). (E−H) Activities of other DnaA mutants with DARS1 in the presence of DiaA. Experiments were performed at the same time as described for panels A−C using the indicated DnaAs. For visibility, the data are shown separately in each panel using the identical data of WT DnaA. For all panels, at least two independent experiments were carried out, and mean values are plotted with SD error bars (some SDs are too small to be shown).

First, the ADP form of WT DnaA was incubated with WT DiaA or DiaA L190A that is defective for DnaA binding. Unlike DiaA L190A, WT DiaA moderately stimulated ADP dissociation; whereas 50% ADP dissociation required ∼20 pmol of DARS1 without DiaA, ∼10 pmol of DARS1 was sufficient in the presence of DiaA (Figure 4A and Supplementary Figure S2A). The inactivity of DiaA L190A indicates that this stimulation is dependent on specific DnaA–DiaA binding. Analogous to the case for oriC, this stimulation might be due to enhanced assembly of DnaA molecules on DARS1 by DiaA.

Second, similar analysis of the ADP form of DnaA Q208A indicated that the residual activity of ADP dissociation of DnaA Q208A was diminished by DiaA, resulting in severe inactivity (Figure 4B). This might be because the rate of DnaA assembly is important for construction of a functional DnaA assembly on DARS1, and an enhanced assembly rate may negatively impact the residual functional assembly of DnaA Q208A. Alternatively, DiaA may interact weakly with DnaA domain III or IV, thereby affecting the complex structure of DnaA assembly on DARS1, which could have positive and negative effects on WT DnaA and DnaA Q208A, respectively.

Third, when DnaA L290A, which is defective in the DARS2 function (37), was similarly analyzed, full inhibition of ADP dissociation remained, even in the presence of DiaA (Figure 4C). This suggests that DnaA Leu290 sustains structural changes essential for the function of both DARS1 and DARS2 (see Discussion).

Fourth, similar analysis was performed using a DARS2 reconstitution system with ADP forms of WT DnaA or DnaA Q208A (Figure 4D). ADP dissociation activity of WT DnaA was basically similar in the presence or absence of DiaA, although DiaA might slightly stimulate the activity. Unlike the case for DARS1, the DARS2 reconstitution system included IHF and Fis, which enhances DnaA assembly (37), and could reduce the effect of DiaA on WT DnaA. By contrast, the residual activity of ADP dissociation by DnaA Q208A was diminished by DiaA, resulting in severe inhibition, consistent with the results obtained for DARS1. This suggests that DnaA Gln208 plays a similar role in DARS1 and DARS2 DnaA complexes.

Fifth, we extended the analysis using DARS1 and DnaA V211A, L214R, Q215A and F239A (Figure 4E−H). Like WT DnaA, these mutant DnaAs were stimulated in ADP dissociation by DiaA.

DnaA Q208A is active in vivo in initiation at oriC but is impaired in reactivation at DARS

After establishing the specific significance of DnaA Gln208 for DARS systems in vitro, we analyzed the in vivo significance. To this end, we first analyzed replication initiation activity using plasmid complementation tests with dnaA::Tn10 rnhA::cat mutant cells. The rnhA::cat sequence activates alternative origins outside oriC, allowing chromosomal replication in a dnaA-independent manner and slow growth of cells (doubling time ∼100 min) (44). The WT dnaA or dnaA Q208A gene was inserted into a low copy number mini-R plasmid bearing the rnhA gene (pRRNH). These plasmids (pOZ18 for WT dnaA, pRS1 for dnaA Q208A) and the pRRNH vector were introduced into a WT dnaA strain, and cells were incubated at 30°C for 22 h, and the number of visible colonies was counted. WT dnaA cells were transformed with three plasmids at a comparable efficiency, although transformation of pRS1 (dnaA Q208A) might be slightly less efficient (Table 1).

Table 1.

DnaA Q208A is active in replication initiation in vivo

	Relevant genotype
Strain	rnhA	dnaA	mini-R–rnhA plasmid	dnaA allele on plasmid	Colony formation efficiency (×105 /μg DNA)
KH5402-1	WT	WT	pOZ18	WT	1.0
	WT	WT	pRS1	Q208A	0.70
	WT	WT	vector	none	1.0
KA451	::cat	::Tn10	pOZ18	WT	1.1
	::cat	::Tn10	pRS1	Q208A	0.90
	::cat	::Tn10	vector	none	<1.0 × 10-3

Cells of the wild-type (WT) dnaA strain (KH5402-1) and the dnaA rnhA double mutant (KA451) were transformed with mini-R plasmid derivatives (10 ng) harboring only WT rnhA (pRRNH vector), WT rnhA, and WT dnaA (pOZ18), or WT rnhA and dnaA Q208A (Q208A; pRS1). Transformed cells were incubated on LB agar plates containing 100 μg/mL ampicillin at 30°C for 22 h, and visible colonies were counted.

Next, these plasmids were introduced into dnaA::Tn10 rnhA::cat cells. Since dnaA::Tn10 rnhA::cat cells grow slowly, cells complemented with WT dnaA could form visible colonies within 22 h under these conditions (19,21). Introduction of the pRRNH vector did not stimulate growth, resulting in a severely reduced number of colonies (Table 1). However, introduction of pOZ18 bearing WT dnaA stimulated the growth of cells by activating oriC, resulting in much higher transformation efficiency (and hence colony formation), as previously reported (19,21). Similar activation was shown for pRS1 bearing dnaA Q208A. In addition, we analyzed the DnaA expression level in transformed cells using immunoblot analysis (Supplementary Figure S3A). DnaA levels in cells bearing pOZ18 (WT dnaA) were similar to those in cells bearing pRS1 (dnaA Q208A). These results suggest that DnaA Q208A also activates replication initiation at oriC in vivo.

To analyze the in vivo role of Gln208 in the DARS system, we used a pBR322-derived multicopy plasmid bearing DARS2. Introduction of this plasmid into cells severely inhibits colony formation in a manner dependent on oriC (36). In this case, oversupply of DARS2 most likely causes over-activation of DnaA and severe over-initiation of replication, resulting in inhibition of cell growth. Consistent with these previous results, when a pBR322 derivative bearing DARS2 (pKX11-kan) was introduced into cells expressing WT dnaA, colony formation was severely inhibited (Table 2). By contrast, when the DARS2 plasmid was introduced into cells expressing dnaA Q208A but not WT dnaA, colony formation was fully sustained (Table 2). Given that DnaA Q208A is essentially active for initiation (Table 1), these results support the idea that DnaA Gln208 is fundamentally instrumental in reactivation by DARS2 in vivo.

Table 2.

Inhibition of cell growth by DARS2 oversupply is suppressed in dnaA Q208A cells

Plasmid
1st plasmid bearing WT rnhA	dnaA allele	2nd plasmid	DARS2	Colony formation efficiency (× 104/μg DNA)
pOZ18	WT	pBR322-kan	−	1.1
		pKX11-kan	+	<1.6 × 10−3
pRS1	Q208A	pBR322-kan	−	1.1
		pKX11-kan	+	1.0

Cells of the dnaA::Tn10 rnhA::cat double mutant (KA451) bearing Amp-resistant mini-R plasmid derivative pOZ18 (WT rnhA and WT dnaA) or pRS1 (WT rnhA and dnaA Q208A) (1st plasmid) were transformed with Kan-resistant pBR322 derivative (pKX11-kan or pBR332-kan) with (+) or without (−) DARS2 (2nd plasmid). Portions of transformed cells were incubated on LB agar plates containing 50 μg/mL kanamycin and 50 μg/mL ampicillin at 37°C for 22 h, and visible colonies were counted.

In addition, we performed flow cytometry analysis of dnaA::Tn10 cells bearing pOZ18 (WT dnaA) or pRS1 (dnaA Q208A), which were used for experiments of Table 1. Cells were grown exponentially, and portions of the cultures were further incubated in the presence of cefalexin and rifampicin for run-out replication of the entire chromosomes (8,10,37). The resultant number of chromosomes per cell (determined by flow cytometry) corresponds to the number of oriC copies at the time of drug addition, thereby indicating activity of replication initiation in cells. In those cells bearing WT dnaA, the eight-chromosomes peak predominated with a minor peak for four chromosomes (Supplementary Figure S3B). Additional minor peaks for five and six chromosomes indicate asynchronous initiations, likely due to expression of dnaA from the mini-R plasmid. In those cells bearing dnaA Q208A, the four-chromosomes peak, rather than the eight-chromosomes peak, predominated, and peaks for five and six chromosomes were enhanced. These results are consistent with the aforementioned results.

DnaA Q208A assembly on DARS sites in the presence and absence of DiaA

To analyze the role of DnaA Gln208 in DnaA-DARS interactions, we performed EMSA experiments under various conditions. When WT ADP-DnaA and the 70 bp DARS1 fragment were used, four discrete complexes were detected, depending on the amount of ADP-DnaA, and WT apo-DnaA was inefficient for formation of Complexes II−IV (Figure 5A), consistent with our previous data (36). In addition to the DARS core bearing three crucial DnaA boxes, we found a degenerated DnaA box sequence flanking DnaA box III within the 70 bp DARS1 fragment (Supplementary Figure S4). Substitution of the sequence (defined as DnaA box III’) with a nonsense sequence lacking specific affinity for DnaA prevented formation of Complex IV, but not Complexes I−III. However, ADP dissociation activity was almost fully sustained with DnaA box III’-substituted DARS1, which suggests that DnaA box III’ and formation of Complex IV are largely dispensable for ADP dissociation activity (Supplementary Figure S4).

Figure 5.

EMSA experiments using DARS1. (A and B) ADP-bound and apo forms of WT DnaA and DnaA Q208A (A) or DnaA L290A (B) were incubated at 30°C for 5 min with MD1−8 DNA (0.40 pmol), followed by EMSA analysis using 6% PAGE. Left, Gel image. Right, Intensity plot derived from scanning EMSA data obtained using 2 pmol of DnaA. Free, Protein-free DNA; C I−C IV, Complex I−IV. λ phage DNA was used as a competitor. (C and D) Proteins were similarly incubated in the presence of DiaA and analyzed. Data are also presented similarly, except that data obtained using 2 pmol of DnaA and 4 pmol (as monomer) of DiaA were scanned. In addition, migrating positions of complexes formed in the presence of DiaA (Complexes I’−IV’) are labeled as C I’−C IV’. As differences in positions of those and C I−C IV are slight, those complexes are labeled collectively in these panels. For details, see Supplementary Figure S2B.

When ADP-DnaA Q208A was similarly analyzed, Complexes I−III were formed at a level similar to WT ADP-DnaA (Figure 5A). These results suggest that functional DnaA assemblies of ADP-DnaA Q208A molecules are largely intact.

When ADP-DnaA L290A was similarly analyzed, Complexes I and II were formed at a level similar to WT ADP-DnaA, but Complex III was decreased slightly (Figure 5A). These results suggest that Complexes I and II of ADP-DnaA L290A molecules are stable, but higher complexes are not very stable. Our previous analysis indicated that DnaA L290A assembly on DARS2 was unstable, even for Complexes I and II (37). This difference is reasonable because DnaA box II is a complete consensus sequence for DARS1 and has a single mismatch for DARS2 (Figure 1E). Stability of DARS2-DnaA complexes is likely more dependent on DnaA–DnaA interactions than is stability of DARS1-DnaA complexes.

EMSA experiments were also performed in the presence of DiaA. Formation of higher complexes (putative Complexes III and IV) of WT ADP-DnaA was stimulated by DiaA (Figure 5C, D), basically consistent with the functional stimulation with DARS1 (Figure 4A) as well as the role for DiaA in oriC. Slight shift up of the bands corresponding to Complexes I-IV was detected depending on DiaA, and those were indicated as Complexes I’-IV’ (Figure 5C, D and Supplementary Figure S2B). Although the numbers of DnaA and DiaA molecules in these complexes have not been determined, ADP dissociation experiments suggest that these complexes respectively include DnaA molecules corresponding to Complexes I–IV. Retardation of the migration rates by DiaA in electrophoresis was slight, which would be due to enhanced migration by acidic nature (theoretical pI of 5.28) of DiaA contrast to basic nature (theoretical pI of 8.77) of DnaA. Addition of DiaA L190A, a DnaA binding-defective mutant, did not substantially change the band patterns in EMSA (Supplementary Figure S2B). Formation of Complex IV was only faint or not detected in the absence of DiaA because of the limited amounts of DnaA. Even for DnaA Q208A and DnaA L290A, formation of Complex III’ was stimulated by DiaA compared with WT DnaA, but formation of Complex IV’ was not. These results are consistent with those in the absence of DiaA, and suggest that DARS1 assemblies of DnaA Q208A and DnaA L290A are not fully intact, although overall assemblies on the DARS core DnaA boxes I−III are constructed. DnaA–DnaA interaction modes and/or local structures of DnaA-DARS1 complexes might be altered by the Q208A and L290A mutations.

Addition of DiaA moderately impeded the decease of the DnaA-free DNA levels (Figure 5C, D). Considering that a DiaA homotetramer stably binds multiple DnaA molecules, free diffusion of each DnaA molecule might be restricted, decreasing the chance to encounter the ligand DNA compared to DiaA-free DnaA monomers under the present conditions.

DnaA molecules arranged in head-to-head interactions dissociate ADP

To analyze the specific function of DnaA protomers bound to the DARS core, we used chimeric DnaA (chiDnaA) consisting of E. coli DnaA domains I−III and Thermotoga maritima DnaA domain IV (10,24) (Figure 6A). T. maritima DnaA specifically binds to a sequence (TmaDnaA box) different from the E. coli DnaA box sequence with a comparable affinity (46), and similarly chiDnaA binds specifically to TmaDnaA box, but not the E. coli DnaA box (10,24). We separately introduced TmaDnaA box into the equivalent position of DnaA boxes I, II, and III (Figure 6B), and the functions of these chimeric DARS1 derivatives (DARS1 TmaI−III) were assessed in the presence of E. coli DnaA and chiDnaA. Although the consensus sequence of TmaDnaA box is 12 mer (AAACCTACCACC) according to sequence analysis (46), previous experiments assessing functional initiation complexes using chiDnaA and oriC bearing TmaDnaA box suggest that the 5′ side of the 9 mer of the TmaDnaA box consensus sequence (AAACCTACC) corresponds to the position of the E. coli 9-mer DnaA box consensus sequence (TT[A/T]TNCACA) (10,24). Additionally, when DnaA boxes I−III were substituted with TmaDnaA box at different positions and ADP dissociation activities were assessed, similar specificity was observed for TmaDnaA box (Supplementary Figure S5). These results are consistent with the fact that spaces between DnaA boxes are highly conserved in DARS1, DARS2, and predicted DARS-candidate sequences in various bacterial species (36). Based on these results, we used optimal DARS1 derivatives (DARS1 TmaI−III) for further experiments.

Figure 6.

Analysis of the specific role of the DnaA protomer using chiDnaA. (A) Schematic representation of the chiDnaA structure. T. maritima DnaA and ChiDnaA domains are shown with the amino acid residue numbers. ChiDnaA consists of E. coli DnaA domains I−III and T. maritima DnaA domain IV (10,24). (B) Schematic representation of chimeric DARS1 with E. coli and T. maritima DnaA boxes. E. coli DnaA box I, II or III (black triangles) was replaced with T. maritima DnaA box (blue triangles), resulting in DARS1 TmaI, DARS1 TmaII, and DARS1 TmaIII, respectively. Sequences are shown in Supplementary Figure S5. (C) [3H]ADP-EcoDnaA dissociation assay using various DNA concentrations. [3H]ADP-EcoDnaA (2 pmol) and non-radioactive ADP-chiDnaA (1 pmol) were co-incubated at 30°C for 15 min with the indicated amounts of chimeric DARS1 TmaI−III or Nonsense DNA excluding DnaA boxes, then subjected to nitrocellulose filter retention assay (Left). Potential DnaA complexes are also illustrated (Right). Only DnaA domains III and IV are shown. Domain IV is colored green for EcoDnaA (Eco) and red for chiDnaA (Chi). Similar illustrations are also shown for panels D−F. (D) [3H]ADP-EcoDnaA dissociation assay using various DnaA concentrations. Similar experiments were performed using 0.4 pmol of the indicated DNA, [3H]ADP-EcoDnaA (2 pmol), and non-radioactive ADP-chiDnaA (0−1 pmol). (E) [3H]ADP-chiDnaA dissociation assay. Similar experiments were performed using various amounts of the indicated DNA, non-radioactive ADP-EcoDnaA (2 pmol), and [3H]ADP-chiDnaA (1 pmol). Data averaged from at least two independent experiments are plotted using mean values and SD error bars. (F) Time course of [3H]ADP-chiDnaA dissociation. Similar experiments were performed using 0.4 pmol of chimeric DARS1 TmaI−III, non-radioactive ADP-ecoDnaA (2 pmol), and [3H]ADP-chiDnaA (1 pmol). Data averaged from at least two independent experiments are plotted using mean values and SD error bars.

First, E. coli [3H]ADP-DnaA and ADP-chiDnaA were incubated with DARS1 TmaI−III (Figure 6B−D), and [3H]ADP was efficiently dissociated from DnaA when DARS1 TmaIII was used, whereas DARS1 TmaI and DARS1 TmaII promoted moderate [3H]ADP dissociation. These results suggest that DnaA bound to DnaA boxes I and II efficiently dissociates ADP, while DnaA bound to DnaA box III does not. When chiDnaA was used, higher amounts of DARS1 derivatives were required for ADP dissociation compared with WT DnaA and WT DARS1, possibly due to retarded turnover rates of DnaA molecules per DARS1 in the chiDnaA system.

Second, E. coli ADP-DnaA and [3H]ADP-chiDnaA were incubated with DARS1 TmaI−III (Figure 6E, F), and [3H]ADP dissociation was barely detected when DARS1 TmaIII was used, whereas DARS1 TmaI efficiently promoted [3H]ADP dissociation, and DARS1 TmaII also promoted [3H]ADP dissociation with moderate efficiency. These results are consistent with the aforementioned proposal that DnaA binding to boxes I and II releases ADP but binding to box III does not.

In addition, DnaA assemblies on DARS1 TmaI−III were assessed by EMSA experiments (Supplementary Figure S6). DARS1 TmaIII formed Complex III with E. coli DnaA less efficiently than WT DARS1, and formation of Complex IV was severely inhibited. Formation of Complexes III-IV might be caused by enhancement of non-specific DNA binding of DnaA by cooperative binding processes. However, formation of these complexes was efficiently promoted in the presence of chiDnaA. These results suggest that DnaA complexes were constructed even on DARS1 TmaIII in the presence of both E. coli DnaA and chiDnaA, which further implies that DnaA bound to box III does not dissociate ADP. Similar results were also obtained for DARS1 TmaI (Supplementary Figure S6). Higher complexes of DARS1 TmaII including chiDnaA also were formed but were moderately unstable (Supplementary Figure S6). However, this unstability in EMSA experiments would not be functionally very significant in that the activity in ADP dissociation was comparable between DARS1 TmaI and DARS1 TmaII (Figure 6).

DnaA Gln208 is instrumental for head-to-head interactions of DnaA

To explore the roles of Gln208 and Leu290 in the DnaA protomers, we further analyzed the reconstituted systems using E. coli mutant DnaA, chiDnaA, and DARS1 derivatives bearing TmaDnaA box. First, DARS1 TmaIII was incubated with ADP-chiDnaA in addition to the [3H]ADP form of E. coli WT DnaA, DnaA Q208A, or DnaA L290A (Figure 7A and Supplementary Figure S7). In this system, E. coli DnaA should bind to DnaA boxes I and II. In contrast to E. coli WT DnaA, DnaA Q208A was impaired in ADP dissociation, supporting the importance of these residues in the head-to-head interaction of DnaA protomers bound to DnaA boxes I and II. Also, DnaA L290A was impaired in ADP dissociation (also see below).

Figure 7.

Role of Gln208 in head-to-head DnaA dimers. [3H]ADP dissociation assay using DARS1 TmaI−III. [3H]ADP-EcoDnaA (WT, DnaA Q208A, or DnaA L290A; 2 pmol) and non-radioactive ADP-chiDnaA (1 pmol) were co-incubated at 30°C for 15 min with DARS1 TmaIII (A), DARS1 TmaII (B), or DARS1 TmaI (C). Predicted DnaA complexes are also illustrated, as shown for Figure 6 (Left).

Second, similar experiments were performed using DARS1 TmaII (Figure 7B and Supplementary Figure S7). In this system, E. coli DnaA should bind to DnaA boxes I and III. Unlike the case with DARS1 TmaIII, ADP dissociation was sustained even for DnaA Q208A. This suggests that ADP dissociation is inhibited only when both DnaA protomers bound to DnaA boxes I and II lose the Gln208 residue. DnaA L290A was impaired in ADP dissociation, including with DARS1 TmaIII.

Third, similar experiments were performed using DARS1 TmaI (Figure 7C and Supplementary Figure S7). ADP dissociation of DnaA Q208A was sustained, consistent with the proposed role for this residue. DnaA L290A was inactive in ADP dissociation, including with DARS1 TmaII. Taken together, the results for DnaA L290A suggest that this residue is important for structural changes of DnaA complex bound to DnaA boxes I and II, rather than head-to-tail interactions between DnaA protomers in the DARS1 system.

DISCUSSION

In this study, based on a structural model of a head-to-head dimer of DnaA AAA+ domains, we analyzed the roles of specific residues residing at the interface of the modeled dimer in ADP dissociation by the DARS system, and revealed that Gln208 is crucial for the reaction. The DnaA Q208A mutant essentially sustained activity for replication initiation from oriC in vitro and in vivo. Further experiments using chimeric DnaA and DARS derivatives indicated that DnaA protomers bound to DARS DnaA boxes I and II, rather than the box III-bound DnaA, predominantly released ADP, and ADP dissociation was inhibited when Gln208 residues of both DnaA protomers were substituted. Furthermore, we found that DnaA-interacting protein DiaA stimulates DARS1-dependent ADP dissociation from WT DnaA, which inversely exaggerated impairment of the DnaA Q208 mutant in DARS1-dependent ADP dissociation. Taken together, these results indicate an essential role for head-to-head interactions in the construction of DARSs-DnaA complexes competent for ADP dissociation, as well as a second role for DiaA in sustaining timely initiation by stimulating the DARS system. However, we do not exclude the possibility that other structures of DnaA AAA+ head-to-head dimers may also be consistent with the mutant data.

In the DARS1 core region, DnaA box II has a complete consensus sequence. Based on this, it is reasonable to presume that, during DnaA assembly, DnaA binding to box II is more efficient than to box I or III (Figure 8). Next, the box II-bound DnaA would recruit the additional DnaA molecules to boxes I and III via head-to-head and head-to-tail interactions with DnaA AAA+ domain III respectively. If DnaA box III’ is present, another DnaA molecule would be recruited, but this is not a prerequisite for DnaA-ADP dissociation activity. Interactions involving Gln208 in the head-to-head DnaA dimer are essential for the ADP dissociation function. In addition, DnaA Phe239 and Lys243, both residing at Gln208-flanking sites in the tertiary structure model (Figure 2), could assist in the functional head-to-head interaction of AAA+ domain III (Figure 3A). At least one of the two bonds involving Gln208 between monomers is required for functional structural changes during ADP dissociation. In addition, cooperative processes in DnaA assembly on DARS1 would be enhanced by DiaA.

Figure 8.

Modeling of molecular dynamics in DARS-DnaA complexes. A mechanistic model is proposed based on the present and previous studies. DARS1 DnaA box II includes the complete consensus sequence of the 9-mer DnaA box. We assume that ADP-DnaA binds to this site, recruiting the second and third molecules to DnaA boxes I and III, via head-to-head or head-to-tail interactions of DnaA domain III. The distance between the C-terminal regions of the domain III monomers in the head-to-head dimer might be fitted to that including DnaA boxes I and II and the 5-bp space between the two (Supplementary Figure S1). DiaA homotetramer binding to multiple domain I molecules accelerates these cooperative interaction processes. Interactions between ADP-DnaA molecules promote ADP dissociation from the head-to-head DnaA dimer in a manner dependent on Gln208 in both protomers. This might be caused by induced structural changes in the nucleotide-binding pockets of the head-to-head DnaA dimer, which decreases the affinity for the nucleotide. Leu290 might support these structural changes and might enlarge the relative angle of the boundary between domain III N-terminal and C-terminal subdomains (i.e. domains IIIa and IIIb), relaxing the structure of the nucleotide-binding pocket. Leu290 is located at the domain IIIa C-terminus (Figure 1B). The resulting apo-DnaA dissociates, enabling further interactions of ADP-DnaA molecules, and a new reaction cycle. DnaA domains I (dark circles), II (black lines), III (orange rectangles), and IV (dark green rectangles), ADP (blue circles), Q208 (red arrows), L290 (brown triangles), and DiaA tetramer (pale green circles) are shown.

The head-to-head interaction of the two DnaA AAA+ domains via Gln208 might cause structural changes in the domain that adversely affect the structure of the nucleotide-binding pocket, abolishing ADP-binding activity (Figure 8). The nucleotide-binding pocket is surrounded by AAA+ subdomains IIIa and IIIb in the tertiary structure, and these subdomains are linked via a short loop (Figures 1C and 2). For example, in DnaA molecules bound to boxes I and II, the head-to-head interaction promoted by AAA+ subdomain IIIa bearing Gln208 could swivel the subdomains, expanding the space within the nucleotide-binding pocket (Figure 8). Consistently, a recent structural study using cryo-electron microscopy and an AAA+ protein construct with head-to-tail oligomerization suggests that interaction of an effector protein allosterically induces structural changes in the AAA+ protein, expanding the space within the nucleotide-binding pocket, resulting in ADP dissociation (47).

DARS1-DnaA complexes are indicated to require Leu290 for ADP dissociation activity (Figures 3 and 7). However, construction of DnaA assemblies was essentially sustained, and was only slightly inhibited with DnaA L290A (Figure 5). This difference suggests that this residue may play an important role in sustaining structural changes in the DnaA complex that support ADP dissociation. This hydrophobic residue resides in the vicinity of AAA+ subdomains IIIa and IIIb, which are linked via a short loop; hence this residue might be important for physically sustaining structural changes around the linker (Figures 1C and 8). In addition to the requirement for Leu290 in DnaAs bound to boxes I and II, there is a possibility that box III-bound DnaA may also require Leu290 for ADP dissociation by DARS1-DnaA complexes. This could be explained by a possibility that structural changes in box III-bound DnaA L290A may adversely affect structural changes in box I- and II-bound DnaA complexes. This idea is consistent with results showing that construction of DARS1 Complex III was slightly inhibited with DnaA L290A (Figure 5). It should be noted that the structure of DnaA domains III and IV can differ using a short flexible linker between the two domains, which enables swiveling of domain IV (17). Molecular dynamics modeling of oriC-IHF-DnaA oligomer complexes indicated that each DnaA protomer is structurally different in the complex due to swiveling of domain IV to varying degrees and orientations (25). These various modes for domains III and IV might underlie the possible differences in dependency on Leu290 for complex formation in DARS1 core DnaA boxes, as well as oriC subregions (R1−I2 regions vs. R4−C3 regions) (22).

Given these results and the conserved structure of the core region, we can infer that the fundamental mechanisms of DnaA–DnaA interactions and ADP dissociation are likely common between DARS1 and DARS2. In addition, a minor difference between the two systems was apparent since DnaA L290A is less capable of constructing higher oligomers (>Complex II) on DARS2 (37). This is likely caused by a difference in the affinity of DARS core DnaA box II for DnaA (i.e., DARS1 has a complete DnaA box consensus sequence, but DARS2 has a sequence with one mismatch; Figure 1E). Also, the DARS2 DnaA complex is activated for ADP dissociation by binding of Fis and IHF (37). These findings can explain the differences in the dependency on Leu290 in DnaA assembly between DARS1 and DARS2. In addition, the stimulatory effect of DiaA in ADP dissociation was minimum for DARS2 (Figure 4D). This can also be explained by the previous result, since the DARS2-specific factors Fis and IHF play a stimulatory role in DnaA assembly (37). However, detailed roles for these factors in DARS2 functional mechanisms remain to be elucidated.

Typical AAA+ proteins construct stable homo-oligomers with a ring or spiral configuration that include 5−7 protomers engaging in head-to-tail interactions (4,16). In these complexes, ATP is hydrolyzed, the resulting ADP dissociates, and apo-protomers remain in the complex and bind ATP again to initiate a new reaction cycle. The structures of complexes are altered during nucleotide transfer reactions. Unlike these typical cases, DnaA molecules bound to DARSs are released from the complex when ADP dissociates. Releasing apo-DnaA protomers is crucial for recycling DARS elements, and promoting efficient ADP release and reactivation of DnaA proteins, all of which must take place at specific times before replication initiation during the cell cycle. Head-to-head interaction of AAA+ domains, rather than head-to-tail interactions, might be advantageous for constructing dynamic complexes that undergo rapid disassembly and assembly. The DnaA protein and DARS elements might have evolved to capitalize on such an advantage.

Click here for additional data file.