Ordered and sequential binding of DnaA protein to oriC, the chromosomal origin of Escherichia coli.

DnaA protein of Escherichia coli acts in initiation of chromosomal DNA replication by binding specific sequences, termed DnaA boxes in the chromosomal origin, oriC. On binding, it induces a localized unwinding to create a structure recognized by other replication proteins that act subsequently in the initiation process. In this report, we examined the binding of DnaA protein to each of the DnaA boxes in oriC. By gel mobility shift assays, DnaA protein formed at least six discrete complexes. ATP or ADP included in the reaction mixture prior to electrophoresis was required. Chemical cleavage of isolated complexes with 1,10-phenanthroline-copper revealed that DnaA protein binds in an ordered manner to the DnaA boxes in oriC. Preferential binding to one DnaA box (R4) was confirmed by demonstration that a DNA fragment containing it was bound with greater affinity than another DnaA box sequence (R1). In vitro replication activity correlated with a complex formed at a ratio of 30 DnaA monomers/oriC in which all DnaA boxes are occupied. The last site bound is DnaA box R3. This event may be critical in promoting initiation from oriC as it correlates with in vivo observations that binding of DnaA protein to box R3 occurs at the time of initiation of chromosomal replication, whereas other DnaA boxes are bound by DnaA protein throughout the cell cycle (Cassler, M. R., Grimwade, J. E., and Leonard, A. C. (1995) EMBO J. 14, 5833-5841).

DnaA protein of Escherichia coli acts in initiation of chromosomal DNA replication by binding specific sequences, termed DnaA boxes in the chromosomal origin, oriC. On binding, it induces a localized unwinding to create a structure recognized by other replication proteins that act subsequently in the initiation process. In this report, we examined the binding of DnaA protein to each of the DnaA boxes in oriC. By gel mobility shift assays, DnaA protein formed at least six discrete complexes. ATP or ADP included in the reaction mixture prior to electrophoresis was required. Chemical cleavage of isolated complexes with 1,10-phenanthroline-copper revealed that DnaA protein binds in an ordered manner to the DnaA boxes in oriC. Preferential binding to one DnaA box (R4) was confirmed by demonstration that a DNA fragment containing it was bound with greater affinity than another DnaA box sequence (R1). DnaA protein of Escherichia coli is a sequence-specific DNAbinding protein, proposed to recognize 9-mer sequences termed DnaA boxes, present in four copies within the chromosomal origin, oriC ( Fig. 1) (1). At oriC, its binding promotes an ordered series of events to result in the initiation of chromosomal DNA replication (reviewed in Ref. 2). The binding of DnaA protein to oriC has been examined by a variety of methods. By electron microscopy (3,4) and DNase I footprinting (1), a large nucleoprotein structure containing 20 -30 monomers of DnaA protein is formed at oriC. Experiments to examine its interaction with individual DnaA boxes led to the conclusion that it bound to the two centrally located DnaA boxes in oriC (R2 and R3) with greater affinity than to the flanking DnaA boxes (R1 and R4) (5). This was based on an indirect assay that assessed the activity of DnaA protein as a transcriptional terminator in vivo. Expression from the lac promotor, located upstream to the mutant DnaA box being examined, was measured.

In vitro replication activity correlated with a complex formed at a ratio of 30 DnaA monomers/oriC in which all
By contrast, gel mobility shift assays with oligonucleotides of 21 base pairs containing various DnaA boxes with natural flanking sequences indicated that DnaA protein binds to DnaA boxes R1 and R4 of oriC with higher affinity than R2 (6). DnaA box R3 was bound as poorly as nonspecific oligonucleotides. In addition to these in vitro findings, in vivo footprinting of oriC plasmids with dimethyl sulfate in exponentially growing cells revealed protection of DnaA boxes R1, R2, and R4 with little binding to R3 (7,8). These observations, suggesting that the binding of DnaA protein to R3 is critical for the initiation process, is supported by the observation that occupancy of R3 occurs at the time of initiation of DNA replication in synchronized cultures (7). Another study reported that mutations in single DnaA boxes (R1 and R4) of oriC reduced binding of DnaA protein to respective sites (5) but only reduced replication activity when both mutant sequences were present together (9). The replication activity of oriC may tolerate an alteration of one of the binding sites (9), perhaps by the speculated cooperative binding of DnaA protein, or occupancy of all four DnaA boxes is not required for replication.
Other proteins have been characterized to bind to specific regions of oriC. IciA protein was isolated by its ability to bind specifically to 13-mer motifs near the left boundary of oriC (10) (Fig. 1). Its binding inhibits the initiation process. IHF 1 and Fis binding sites in oriC have been described (11)(12)(13)(14)(15). Rob protein binds to a region near the right boundary of oriC, but its significance is unknown (16).
The studies summarized above do not provide a clear understanding of whether DnaA protein binds to sites in oriC randomly or in an ordered and sequential manner in the process of initiation of chromosomal replication. If this event is ordered, the binding of other proteins to oriC may inhibit or augment the initiation process. In this report, we use gel mobility shift and DNA footprinting techniques to characterize complexes of DnaA protein bound to oriC. Results indicate that DnaA protein binds to oriC in an ordered manner. DnaA box R4 is bound first, then R1, and finally the two inner boxes. Formation of these discrete complexes was dependent on ATP or ADP. Replication activity correlates with binding to all four DnaA boxes.

EXPERIMENTAL PROCEDURES
Gel Mobility Shift Assays (17,18)-Unless noted, a SmaI-Xhol fragment containing oriC, gel-purified from pBSoriC (19) with a Qiaex DNA extraction kit (Qiagen) and quantified by absorbance at 260 nm, was 3Ј-end-labeled with the large fragment of DNA polymerase I (Boehringer Mannheim), and [␣-32 P]dTTP, then combined with the same unlabeled fragment to adjust its specific radioactivity to 4 ϫ 10 3 cpm/25 fmol of DNA. Reactions (10 l) with the labeled oriC fragment (25 fmol) and indicated amounts of DnaA protein were incubated in buffer containing 20 mM HEPES-KOH, pH 8.0, 5 mM magnesium acetate, 1 mM EDTA, 4 mM dithiothreitol, 0.2% Triton X-100, 5 mg/ml bovine serum albumin, and 0.5 M ATP (unless noted otherwise) at 20°C for 5 min * This work was supported by Grant GM33992 from the National Institutes of Health and by the Michigan Agricultural Experiment Station. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
In Situ Cleavage with 1,10-Phenanthroline-Copper (20)-Gel mobility shift assays were performed as above but scaled up 10-fold. After electrophoresis, the wet gel was immersed for 2.5 to 4.5 min in 200 ml of 42 mM Tris-HCl, pH 8.0, 0.2 mM phenanthroline, 38 M CuSO 4 , and 5 mM 3-mercaptopropionic acid at room temperature. To quench the reaction, 2,9-dimethylphenanthroline was added to 2.3 mM followed by incubation for 2 min. The gel was quickly washed with water and autoradiographed for 1 h at room temperature. The developed film was used to guide excision of the complexes. DNA from the gel slices was eluted overnight at 37°C in 500 l of elution buffer (0.5 M ammonium acetate, 0.2% SDS, 1 mM EDTA, 10 g/ml Proteinase K, and 100 g/ml tRNA). After recovering the elution buffer, an additional 200 l of elution buffer was used to wash the gel slices, and both were pooled. The eluted DNA was ethanol-precipitated, washed with 70% ethanol, dried, and resuspended in 5 l of 80% (v/v) formamide, 10 mM NaOH, 1 mM EDTA, 0.1% bromphenol blue, and 0.1% xylene cyanol, heated to 95°C for 2 min, and electrophoresed at 50 watts on a prerun 6% sequencing gel. After autoradiography, beta emission scanning was with a Molecular Dynamics PhosphorImager. Graphed with Excel (Microsoft), the radioactivity in each lane was normalized to the cleavage pattern of unbound DNA that was isolated from the gel.
DNA Replication Assays-Reactions (25 l) were performed as described (21) with a crude protein fraction deficient in DnaA protein activity, M13oriC26 DNA (25 fmol) as a template, and the indicated amounts of DnaA protein. Incubation to measure DNA synthesis was at 30°C for 20 min. Acid-insoluble incorporation of [ 3 H]TTP was quantified by liquid scintillation counting.

Six Discrete Complexes Are Formed on Binding of DnaA
Protein to oriC-By footprint analysis, and electron microscopy, the nucleoprotein complex of DnaA protein bound to oriC is estimated to contain 20 -30 monomers organized to occupy the four DnaA boxes (1,3,4)). Sequence comparison of these DnaA boxes reveal that R1 and R4 are identical, whereas R2 and R3 differ at the fifth and seventh positions, respectively (Table I). Inasmuch as other studies indicated that several nucleotide changes at each of the positions only marginally affected the ability of DnaA protein to bind to oriC and that mutant DnaA boxes of oriC did not affect in vivo replication activity when singly present (9), the binding of DnaA protein to oriC may not show a strong preference to one site relative to others. To investigate this, we examined the binding of DnaA protein to sites in oriC by use of a gel mobility shift assay. With increasing amounts of DnaA protein, six discrete complexes were observed (Fig. 2). Complex I was observed at the lowest ratio of DnaA protein to oriC fragment. Complexes of slower mobility were observed at higher ratios. Compared with other complexes, the minor abundance of Complexes IV and V suggests their lesser stability. Complex VI was predominant at a ratio of 30 monomers of DnaA protein per DNA fragment, consistent with electron microscopic measurements (1). In this lane, other material near the well of the gel corresponds to that formed by a self-aggregate of DnaA protein. 2 Formation of Complex VI Correlates with Formation of an Active Replication Complex-Quantitative analysis was done of complexes in an experiment comparable to that of Fig. 2, but with a larger range of added protein (Fig. 3). In parallel, the replication activity of DnaA protein was measured. Replication activity correlated with the level of DnaA protein that formed Complex VI. At this level, complexes of greater mobility were relatively minor in abundance (as in Fig. 2). Higher levels of DnaA protein resulted in material that remained near the wells of the gel (denoted as "well" in Fig. 3). This material, whose appearance correlated with inhibition of replication activity, could be resolved into two species on longer electrophoresis (data not shown). For simplicity, quantitative analysis of Complexes II-V has not been shown in Fig. 3 as their abundance at lower amounts of DnaA protein are represented in Fig.  2, and at higher DnaA protein levels, they were rare.
DnaA protein bound to ATP is active in unwinding the ATrich 13-mers located in oriC (Fig. 1), whereas the ADP-bound form is relatively inert (22,23). The importance of the nucleotide-bound form led us to investigate the effect of ATP and ADP on complex formation. Formation of discrete complexes was dependent on inclusion of ATP or ADP in the reaction prior to electrophoresis (Fig. 4), although with ADP, the amount of Complex VI was reduced. The minor species resolved between complexes V and VI in Fig. 4 were rarely observed in other comparable experiments. In the absence of nucleotide, trace amounts of Complex I and VI were observed. Instead, the labeled DNA remained near the wells of the gel. The effect of nucleotide on formation of discrete complexes may relate to the observation that DNase I protection of DnaA boxes in oriC by the ATP-bound form of DnaA protein was localized to the DnaA boxes, whereas binding by the nucleotide-free form was less specific (24). By electron microscopy (4), the classes of structures formed in the presence of either nucleotide may correlate to some of the complexes resolved here (Figs. 2 and 4).
Relative Binding Affinities of DnaA Protein to Supercoiled and Linear DNA-DNA replication activity of DnaA protein is dependent on a supercoiled plasmid containing oriC (25). By comparison, the gel mobility shift assays were with a linear DNA fragment. Nitrocellulose filter binding assays showed  (36) is also shown. Binding sites for IHF (11,12) and Fis (13-15), 13-mer motifs recognized by IciA protein (10), and restriction enzyme sites (S, SmaI; Hf, HinfI; A, AvaII; H, HinD III; Ac, AccI; X, XhoI; and P, PstI) of pBSoriC are indicated. that DnaA protein bound with about 3-fold higher affinity to a supercoiled oriC plasmid than to the linearized or relaxed form (26). To examine this issue with the gel mobility shift assay, unlabeled competitor DNA was added to reactions containing a fixed level of DnaA protein and radioactively labeled oriC fragment (Fig. 5). Addition of either supercoiled (estimated to be contaminated by ϳ10% nicked DNA by resolution by agarose gel electrophoresis and quantitative densitometry), linearized oriC plasmid, or the same unlabeled restriction fragment resulted in a comparable reduction of DNA binding to the labeled DNA, measured by densitometric analysis of the autoradiogram. By contrast, poly(dI-dC) was an ineffective competitor. These results indicate a comparable binding affinity of DnaA protein to supercoiled or linear DNAs containing oriC. The apparent discrepancy between these observations and the cited study may be due to the absence of ATP in the filter binding assays (26), whereas it was present in the gel mobility shift experiments (Fig. 5). We have not examined the effect of ATP on binding affinity to different topological forms of oriC-containing plasmids nor compared the effect of supercoiled and linear oriC-containing DNAs by each assay method.
Ordered Binding of DnaA Protein to oriC-To determine the sites bound by DnaA protein in the separated complexes, in situ footprint analysis was performed with 1,10-phenanthrolinecopper (20). Complexes formed with DNA labeled in the top or bottom strand were examined (Fig. 6, A and B). Regions protected from chemical cleavage were identified by quantitative analysis of the resultant autoradiograms (Fig. 6, C and D). Complex I consisted of DnaA protein bound to R4. R1 was additionally protected in Complex II. Complex III differed from Complex II by protection of R2 as well as sequences to the left of R4 in vicinity of the AccI site. Binding to the region encompassing the AccI site is likely not responsible for the electrophoretic position of this and more slowly migrating complexes (described below). Gel mobility shift experiments with a DNA fragment lacking the AccI site resulted in a similar number and proportion of complexes (data not shown). The scarcity of Com-  (bottom strand, panel B). The cleavage pattern of respective complexes was analyzed by beta emission scanning plexes IV and V relative to other complexes (Fig. 2, lanes with 10 -40 ng of DnaA protein) suggests that these may be of lesser stability. Nonetheless, greater protection of R2 was observed in Complex IV compared with Complex III (as well as protection of the region containing the AccI site). Whereas some protection of R3 was seen in Complex IV (Fig. 6D), it was enhanced in Complexes V and VI. Complex VI consisted of protection of the four DnaA boxes as well as flanking sequences in the region from R2 to R4. The altered mobility of Complexes IV and V relative to III may be due to more stable binding of DnaA protein to R2, and R3, possibly by interaction among proteins bound to different DnaA boxes by bending or looping of the DNA. In summary, these findings indicate that DnaA protein binds, in order, to R4, R1, then R2 and R3. These methods do not distinguish the possible contribution that binding to one site may have on subsequent binding to another site due to cooperativity.
DnaA Protein Binds to DnaA Box R4 with Greater Affinity than to R1-Results from gel retardation and footprinting experiments indicated that DnaA box R4 is bound with greater affinity than R1 by DnaA protein. To determine the relative affinity of DnaA protein to restriction fragments containing only box R1 or R4, gel shift assays were performed with unlabeled DNAs as competitors. Addition of increasing amounts of unlabeled fragment containing R4 resulted in proportional inhibition of binding to the labeled R4 fragment present at a constant level (Fig. 7). By comparison, addition of unlabeled R1 fragment was less effective. These results, the average of three independent experiments, confirm that DnaA protein binds to R4 in oriC with greater affinity than R1. As the 9-mer sequence of R4 is identical to that of R1, sequences that flank the DnaA box appear to contribute to binding affinity. The protection of residues at Ϫ1 and Ϫ2 positions (Table I) outside the 9-mer sequence of R4 in Complex I (Fig. 6C) and of R1 in Complex II (Fig. 6D) supports this conclusion. Also, the binding affinity to R2 and R3 apparently is less and may be due to sequence differences in respective 9-mers in addition to the influence of flanking sequences (Table I). DISCUSSION By use of a gel retardation assay in conjunction with protection from cleavage by 1,10-phenanthroline-copper, the binding of DnaA protein to sequences in oriC was found to be ordered and sequential. The six discrete complexes (and material near the wells seen at higher levels of DnaA protein) may correlate with the seven unique structures detected by electron microscopy (4). At higher levels of DnaA protein, the inhibition of replication activity correlated with formation of material that entered the gel poorly (Fig. 3). Complex VI (Fig. 2) may correspond to structure 3 (4) as the formation of both correlated with optimal replication activity. Second, both structure 3 and complex VI formed more efficiently with ATP than with ADP in the reaction mixture. In the absence of nucleotide DnaA protein failed to form discrete complexes. This apparently is due to aggregation of DnaA protein that occurs on its incubation without ATP. 2 By comparison, similar experiments with a dnaA promoter fragment containing a DnaA box identical to R4 of oriC (Table  I) and a weak DnaA box have been performed. 3 We observed two prominent complexes and two minor, more slowly migrating species.
A supercoiled template containing oriC is required for in vitro replication (25). Relating this requirement to structure 3, it was observed more frequently with supercoiled DNA than linear DNA (4). We have not determined whether complex VI forms more efficiently on supercoiled DNA than linear or relaxed DNA, despite attempts to resolve complexes formed on supercoiled DNA in low percentage agarose gels. In addition, we have been unsuccessful in demonstrating ordered binding of DnaA protein on a supercoiled oriC-containing plasmid. The method used was quantitative footprint analysis with DNase I or 1,10-phenanthroline of complexes formed in solution, followed by primer extension. Footprinting in solution provides an averaged picture of complexes formed. At lower ratios of DnaA protein to oriC where we expected to see preferential binding to DnaA box R4 then to R1, we presume that the amount of free DNA masks the protection pattern resulting from ordered binding. Also, we presume that this reason explains why ordered binding was not observed in previous reports (1,24). Although solution footprinting on supercoiled DNA failed to detect ordered binding, competition experiments demonstrated that DnaA protein binds with a similar affinities to supercoiled or linear DNAs containing oriC (Fig. 5). This suggests that DnaA protein binds to either topological form by a similar mechanism.
Despite the identical 9-mer sequences of R1 and R4 (Table I), DnaA protein bound to R4 with about 3-fold higher affinity than to R1 (Fig. 7). Presumably, sequences that flank the 9-mer in R4 contribute to its higher binding affinity. Indeed, the protection of flanking sequences at Ϫ1 and Ϫ2 positions of R4, and R1, clearly seen in Complexes I and II (Fig. 6, C and D), indicates that DnaA protein binds to residues outside of the core sequence. Whether the differences in sequences at the Ϫ2 position of R4 compared with R1 is responsible for the different binding affinities can be tested directly. Other evidence supports the notion that flanking sequences contribute to binding affinity. With a nitrocellulose filter binding assay, we found that DnaA protein bound 4-fold greater to the DnaA box in a 3 C. Margulies and J. M. Kaguni, unpublished results. and compared with the cleavage pattern of the corresponding unbound fragment that was treated similarly (panels C and D). Maxam-Gilbert G and G ϩ A reactions performed on the appropriate oriC fragment served as size markers. The sequence of R4 (panel C) and R1 (panel D) with sequences at Ϫ1 and Ϫ2 positions (Table I) is presented at the bottom.
FIG. 7. DnaA protein binds to DnaA box R4 with greater affinity than to R1. An EcoO1091-PstI restriction fragment (25 fmol) containing DnaA box R4 from pBSR4, radioactively labeled by end-filling at the EcoO1091 site with [␣-32 P]dGTP, was incubated with the indicated amounts of the same unlabeled fragment or an unlabeled SmaI-AvaII restriction fragment containing DnaA box R1. pBSR4 was constructed by insertion of the HindIII-PstI fragment (gel-purified) containing DnaA box R4 of oriC into corresponding sites of pBluescript II SKϩ (Stratagene). DnaA protein (3 ng) was added and complexes resolved from free DNA as described under "Experimental Procedures." After autoradiography, the amount of free DNA remaining in each lane was quantified by ␤ emission scanning, then the amount of DNA bound was calculated. The amount of radiolabeled DNA bound in the absence of competitor was normalized to 1.
dnaA promoter-containing fragment than to a synthetic DnaA box (9-mer) inserted into the multiple cloning site of M13mp18 (Table I). 4 These observations are also supported by the 50-fold difference in binding affinity to a specific DnaA box when flanking sequences were varied (6).
The observations described here are in contradiction to the conclusion that DnaA protein bound with higher affinity to central DnaA boxes (R2 and R3) relative to the flanking sites (R1 and R4) (5). This deduction was based on an indirect method that measured expression of galK dependent on transcription from the lac promoter. The DnaA box sequence being assessed was positioned between the promoter and the galK gene. The relative ability of DnaA protein to function as a transcriptional terminator to affect galK expression was the basis of the assay. It is possible that differences in mRNA stability and/or translational efficiency may have influenced the results obtained.
Four lines of evidence correlate the initiation of DNA replication to the binding of DnaA protein to DnaA box R3. First, dimethyl sulfate treatment of an oriC plasmid carried in an exponentially growing strain revealed that DnaA box R3 was not bound whereas the remaining sites were (8). Such minichromosomes in which plasmid replication occurs from oriC are duplicated once per generation (27) and synchronously with the bacterial chromosome (28). Assuming that replication fork movement is the same as that of the bacterial chromosome, oriC plasmid replication should be completed within a few seconds. As dimethyl sulfate treatment was for 2 min, most plasmids should not be active in replication. These observations suggest two possibilities. One is that the level of DnaA protein throughout most of the cell cycle may be insufficient to bind to this site and that a critical level must be attained to promote initiation. Alternatively, this site may be occluded (see below). Second, in synchronous cultures, initiation of oriC plasmid replication correlated with the binding of DnaA protein to R3 (7). Third, elevated expression of DnaA protein stimulated initiation (29 -31), possibly by increased occupancy of DnaA box R3. Fourth, formation of Complex VI in which DnaA box R3 is occupied last correlated with replication activity (Figs. 3  and 6).
The method used here does not provide a clear picture of the structure of Complex VI except for additional protection of sequences flanking the DnaA boxes. The protected site between R1 and R2 (Fig. 6D) contains the sequence TTATACGGT that resembles the DnaA box sequences (TTAT(A/C)CA(A/C)A) of oriC and presumably is bound for this reason. The protected region between R2 and R3 contains a Fis binding site (Fig. 1) (13)(14)(15). That fis null mutants maintain poorly oriC-dependent plasmids (13,15) and are asynchronous in initiation (32) suggest a positive role for Fis binding. However, certain in vitro conditions demonstrate that Fis is inhibitory to oriC plasmid replication (33), a finding consistent with the report that binding of DnaA protein to R3 is mutually exclusive to the binding of Fis at its respective site (13). Indeed, footprinting studies of oriC minichromosomes in synchronous cultures suggest that Fis blocks the binding of DnaA protein to R3 (7). At initiation, the protection pattern attributed to Fis was not observed. Instead, DnaA box R3 was protected. Shown here, occupancy of R3 by DnaA protein correlates with optimal replication activity ( Figs. 3 and 6). DnaA protein and Fis may compete for binding with contrasting effects on replication activity.
HU protein or IHF act in initiation (34,35) to facilitate unwinding of the 13-mers near the left boundary of oriC (24). A preferred IHF binding site between R1 and R2 (11,12) suggests the possibility that IHF may enhance one or more steps in formation of Complex VI as a prerequisite to unwinding. In synchronously growing cells, IHF occupies this site just before initiation (7). IHF and DnaA protein were concluded to bind independently to oriC, based on the lack of effect of IHF on the DNase I protection pattern by DnaA protein (Fig. 5 of Ref. 24). However, the level of DnaA protein examined was in excess of the level optimal for formation of Complex VI (Fig. 2) and would have obscured detecting enhancement of DnaA protein binding.
In addition to the four DnaA boxes described above, a fifth site, R5, has been proposed as a site of binding of DnaA protein (36). We observed little, if any, protection at this site. This may be partly due to the reduced ability of phenanthroline-copper to cleave in this region.
Except in Complexes I and II, all other complexes revealed a protected region encompassing the AccI site. DNase I footprinting with the nucleotide-free form of DnaA protein showed that it bound to this region (24). However, this region was not protected by DnaA protein bound to ATP, ADP, or ATP␥S. Although the studies presented here involved incubation of DnaA protein with the oriC-containing restriction fragment and 0.5 M ATP prior to electrophoresis, protection of the region containing the AccI site suggests that dissociation of DnaA protein and rebinding of the nucleotide-free form occurs during electrophoresis. The significance of binding to this site is unclear as it is not part of the functional oriC sequences (37).