Genetic Organization of the Vibrio harveyi dnaA Gene Region and Analysis of the Function of the V. harveyi DnaA Protein in Escherichia coli

Abstract

The Vibrionaceae family is distantly related to Enterobacteriaceae within the group of bacteria possessing the Dam methylase system. We have cloned, sequenced, and analyzed the dnaA gene region of Vibrio harveyi and found that although the organization of the V. harveyi dnaA region differs from that of Escherichia coli, the expression of both genes is autoregulated and ATP-DnaA binds cooperatively to ATP-DnaA boxes in the dnaA promoter region. The DnaA proteins of V. harveyi and E. coli are interchangeable and function nearly identically in controlling dnaA transcription and the initiation of chromosomal DNA replication despite the evolutionary distance between these bacteria.

The DnaA initiator protein and the origin of replication, oriC, with its binding sites for the DnaA protein, the DnaA boxes, are the key elements in controlling the initiation of chromosomal DNA replication (15, 23, 29). The dnaA operon in Escherichia coli is transcribed from two promoters with a DnaA box located between them (16). dnaA transcription is autoregulated (1, 5, 19) by a mechanism involving cooperative binding of the ATP-DnaA to newly identified ATP-DnaA boxes (AGatct) (33) in combination with classical DnaA boxes [TT(A/T)TNCACA] (28).

Several chromosomal origins of replication from Enterobacteriaceae (9) and the origin from Vibrio harveyi—a luminous, motile, and facultatively anaerobic gram-negative rod belonging to the Vibrionaceae—are all functional in E. coli (42), whereas the Pseudomonas putida origin is not (40). In parallel, several dnaA genes from Enterobacteriaceae complement dnaA(Ts) mutations in E. coli (31, 32) but the P. putida dnaA gene does not (17). In the γ subdivision of Proteobacteria the families Enterobacteriaceae, Pasteurellaceae, and Vibrionaceae are distinguished from Pseudomonadaceae (10, 26) by having the Dam methylase system in common (4). The Dam methylase methylates the adenine in GATC sequences and plays a role in DNA replication, in mismatch repair, and in gene regulation (27). Overrepresentation of GATC sequences in the oriC and dnaA regions is a conserved feature in E. coli (22, 34), in other Enterobacteriaceae (9, 31), and in V. harveyi (Vibrionaceae) (42). It takes a much longer time in E. coli to remethylate GATC sites within the clusters located in the oriC and the dnaA regions than other GATC sites. This delay of remethylation is due to sequestration of the GATC sites, most likely to the cell membrane (7, 25). The period with the sequestered origins (the eclipse period) is important for the synchrony of initiation (15), and the SeqA protein (21, 37) is an absolute requisite for the eclipse period (38).

We sought an evolutionarily distant dnaA gene that is functional in E. coli. Because neither the dnaA_{P. putida} nor dnaA_{Haemophilus influenzae} gene complements the dnaA46 mutation in E. coli (17; O. Skovgaard, unpublished data), we chose a member of Vibrionaceae, V. harveyi, for this purpose. We investigated whether the basic elements in the control of dnaA transcription regulation and DNA replication in V. harveyi are similar to those of E. coli. After cloning and physical and physiological analysis of the V. harveyi dnaA gene region, we conclude that the basic elements are the same whereas the detailed organizations are different in V. harveyi and in E. coli, reflecting their evolutionary distance.

Isolation and sequencing of the dnaA gene and its flanking regions

Chromosomal DNA from V. harveyi strain B392 (42) was partially digested with TaqI and ligated to pBR322 digested with ClaI. The recombinant DNA was transformed into competent RUC663 (relevant markers, hsdR and dnaA46) cells, and a colony with a plasmid (pRUC666) complementing the temperature-sensitive phenotype of dnaA46 was selected. The presence of the dnaA gene on pRUC666 was verified by Southern hybridization with DNA of an M13 clone containing the E. coli dnaA gene as the probe. DNA from pRUC666 and from pFHC539 (39), which contains the E. coli dnaA gene, was used to probe Southern blots of genomic V. harveyi DNA digested with either SalI, PvuII, HindIII, or BamHI. The results of the Southern hybridizations with the E. coli and the V. harveyi probes were in mutual agreement and in agreement with restriction mapping of pRUC666 (data not shown). This indicated that there is only one copy of the dnaA gene in the V. harveyi genome and that only one DNA fragment was inserted in pRUC666. A second cloning of this region on a SalI fragment in pRUC886 also showed that a contiguous fragment was cloned.

Initial sequence analysis showed that, in contrast to what is observed with E. coli, the rpmH gene was not preceding the dnaA gene in V. harveyi. We hypothesized that the rpmH gene is located close to the dnaA gene as it is in Haemophilus influenzae (11) and designed PCR primers on the basis of the known sequence of pRUC886 and the rpmH gene. The sequence of the resulting 1.4-kb fragment confirmed this hypothesis.

A putative ABC transport operon is inserted between the dnaA and rpmH genes in V. harveyi

The DNA sequence upstream of the dnaA gene revealed that an operon with three putative open reading frames had been inserted between the dnaA and the rpmH genes in an orientation opposite to the dnaA gene. The sequence and the gene organization indicate that this operon encodes an ATP-binding cassette (ABC) transporter. ABC transporters in gram-negative bacteria include a periplasmic substrate-binding protein, one or two hydrophobic proteins forming a diffusion channel, and one or two ATP-binding hydrophilic proteins and are often organized in a single operon (35). The proteins with known functions that are most similar to the putative substrate-binding protein of V. harveyi are E. coli FliY (cysteine-binding protein; accession no. P39174) and GlnH (glutamine-binding protein; accession no. P10344). However, a protein from Clostridium acetobutylicum (accession no. NP_347516) with unknown substrate specificity is even more similar to this V. harveyi substrate-binding protein than any of these E. coli proteins. This suggests that this operon came into V. harveyi by a lateral transfer and that the hypothetical gene products transport a polar amino acid or a similar substance. We have consequently named it pat (mnemonic for polar amino acid transport) with the genes patH (encoding the periplasmatic binding protein), patM (encoding the membrane protein), and patP (encoding the ATP-binding protein). The pat operon and the putative ribosome-binding sites, a promoter in front of the first open reading frame, and two rho-independent terminators in the intercistronic region between patH and patM are indicated on the sequence figure on our website (http://virgil.ruc.dk/∼olesk/). The rpmH gene encoding ribosomal protein L34 is located downstream of the pat operon.

The dnaA promoter region

Three transcriptional start sites for the dnaA operon and one transcriptional start site for the pat operon were identified with primer extension analysis on V. harveyi total RNA (Table 1). Each start site for the dnaA operon was identified with two different primers to eliminate the possibility of false priming. These sites are shown in Fig. 1 along with the most likely promoter sequences. Three putative DnaA boxes were identified in the V. harveyi promoter region (Fig. 1). DnaA box 2 matches the DnaA box definition [TT(A/T)TNCACA] of Schaper and Messer (28). DnaA boxes 1 and 3, though they do not fulfill this definition of a DnaA box, were also investigated for their affinity for DnaA protein (see below).

TABLE 1.

Detected transcriptional start sites in V. harveyi

Sequence of primer	Primer positiona (5′ → 3′)	Transcription start(s)
CATTGAGCTCCGCTTGTAACGGACGAACCCACATACTG	3093 → 3056	2954
CGGACGAACCCACATACTGAATTCTGTAGCTGGTAGCTCTTC	3074 → 3033	2954
GAAGCTGCTGCAAACATTGCAACCAAAGCGAAGATG	3030 → 2995	2859; 2860
GCAACCAAAGCGAAGATGACACGGTGCCCCCACTGAG	3012 → 2976	2741; 2859
GTGAGATCAACACATCTTGCCCCAAATTTGCCCACAC	2857 → 2821	2741
CTTGCTTCACAAAAGTGAATGGGAAGTAGCGACCAGACATACC	2437 → 2479	2582

^aCoordinates refer to Fig. 1.

FIG. 1.

Nucleotide sequence of the dnaA_{V. harveyi} promoter region. GATC sequences are underlined. Promoters (black arrows), DnaA boxes (thin gray arrows), transcriptional start sites (thick gray arrows), ribosome-binding sites (rbs), and translational start sites (PatH; DnaA) are shown above the sequence. A presentation of the entire sequence is available at our website (http://virgil.ruc.dk/∼olesk/)

The dnaA genes of V. harveyi and E. coli are nearly identical in transcription regulation and in controlling the initiation of DNA replication

Initially we found that the dnaA_{V. harveyi} gene complements the temperature sensitivity of an E. coli dnaA46 strain. In order to investigate the functional identity of the V. harveyi and E. coli dnaA genes, we constructed similar transcriptional fusions of the dnaA_{E. coli} and dnaA_{V. harveyi} promoter regions to lacZ, which were integrated into the chromosome (Table 2). We also constructed plasmids which expressed the dnaA_{E. coli} (pRUC978) and dnaA_V. harveyi (pRUC1020) genes from the IPTG (isopropyl-β-d-thiogalactopyranoside)-inducible promoter P_A1/O3/O4 (Table 2). Synthesis of DnaA protein from both these plasmids was very low without induction: (i) in some sodium dodecyl sulfate-polyacrylamide gel electrophoresis gels the DnaA_{E. coli} (from the chromosomal gene) separated clearly from the DnaA _{V. harveyi} and no DnaA_{V. harveyi} signal was visible without induction on Western blots of these gels (data not shown); (ii) the permissive temperature of dnaA46 strains was altered by these plasmids only when they were induced, as no colonies formed at 39°C unless IPTG was added (apart from recombinant colonies with pRUC978; data not shown).

TABLE 2.

DnaA protein concentration, level of dnaAp repression, and number of origins per cell at different IPTG concentrations^a

IPTG concn (μM)	DnaA protein concnb		β-Galactosidase activity (Miller units)				Origins per cellc		Asynchrony indexd
			dnaApE. coli′-′lacZ		dnaApV. harveyi′-′lacZ
	pRUC978	pRUC1020	pRUC978	pRUC1020	pRUC978	pRUC1020	pRUC978	pRUC1020	pRUC978	pRUC1020
100	8.4	16.2	2.7	2.7	5.5	4.3	ND	ND	ND	ND
75	4.2	9.4	3.5	3.9	7.6	5.1	4.23	4.01	0.49	0.54
60	2.5	7.1	4.7	4.7	8.5	6.3	3.94	3.79	0.33	0.50
45	2.1	4.7	6.0	5.9	10.0	7.5	3.26	3.31	0.19	0.39
30	1.4	2.4	7.4	7.0	10.8	8.6	3.10	3.11	0.16	0.23
0	1.2	0.8	8.0	8.2	12.0	10.9	2.76	2.83	0.12	0.16

^aStrains with indicated plasmids and chromosomally integrated dnaAp′-′lacZ fusions were derived from RUC1000 (MC1000, leu⁺). The E. coli (pRUC978) and V. harveyi dnaA (pRUC1020) genes were set under control of the IPTG-inducible P_A1-O3/O4 promoter (18) in plasmid pFHC2102 (accession number AY070365 [2]). Transcriptional fusions of dnaA promoters to lacZ were constructed and integrated by the method of Atlung et al. (3) and Brøndsted and Atlung (6). Strains were pregrown in AB medium (8) supplemented with 0.2% glucose and 1 μg of thiamine per ml at 35°C for four generations. They were then diluted into the same prewarmed medium with different IPTG concentrations and grown exponentially for at least five generation times before samples were taken for analysis. Generation times were between 57 and 59 min for all cultures except for cultures with pRUC1020 induced with 100 μM IPTG, which had a generation time of 61 to 65 min.

^bTotal DnaA protein concentration was quantified from Western blots (14). The values are the averages of three different experiments and are normalized to the DnaA protein concentration in RUC1000.

^cThe number of origins per cell was calculated from the flow cytometry analysis of rifampin-treated samples (Fig. 2). ND, not determined due to incomplete DNA replication during rifampin treatment.

^dThe asynchrony index was calculated as the fraction of cells with 3, 5, 6, or 7 chromosomes.

The amount of DnaA protein expressed, the levels of homologue and heterologue autorepression, and the numbers of origins per cell as a function of induction of pRUC978 and pRUC1020 were determined in the same growth experiment (Table 2; Fig. 2).

FIG. 2.

Flow cytometry analysis of origin distributions in cells containing different amounts of DnaA_{E. coli} protein (pRUC978) (A) and DnaA_{V. harveyi} protein (pRUC1020) (B). Samples were withdrawn from cultures growing with the indicated concentrations of IPTG and incubated with 300 μg of rifampin per ml plus 36 μg of cephalexin per ml to terminate ongoing DNA replications without cell division (20).

Expression of both DnaA proteins was induced by IPTG (Table 2). However, DnaA_{V. harveyi} was more strongly expressed than DnaA_{E. coli} with the same IPTG concentration. For example, the total DnaA protein concentration was increased 4.2-fold when pRUC978 was induced with 75 μM IPTG compared to 4.7-fold when pRUC1020 was induced with 45 μM IPTG. Both DnaA proteins repressed both dnaAp′-′lacZ fusions (Table 2). By comparing the repression by pRUC978 at 75 μM IPTG (3.5 U) with the repression by pRUC1020 at 45 μM IPTG (5.9 U), it is seen that the dnaAp_{E. coli} is repressed more strongly by DnaA_{E. coli} than by DnaA_{V. harveyi}. In contrast, dnaAp_{V. harveyi} is repressed equally well by DnaA_{V. harveyi} and DnaA_{E. coli} (7.6 and 7.5 U at these concentrations of IPTG).

We determined the origin number distribution after completion of DNA synthesis in the presence of rifampin and cephalexin by flow cytometry to measure the effects of induced DnaA protein synthesis on the control of DNA replication (Fig. 2; Table 2). Without induction there were 2.76 to 2.83 origins per cell. At 75 μM IPTG the number of origins per cell increased to 4.23 (DnaA_{E. coli}) and 4.01 (DnaA_{V. harveyi}). At 100 μM IPTG DNA replication in the presence of rifampin was incomplete with both plasmids (Fig. 2). It can be seen that more DnaA_{V. harveyi} protein than DnaA_{E. coli} protein is required for the same initiation capacity (Table 2).

Initiation of DNA replication is synchronous in E. coli wild-type cells, so most cells have 2ⁿ (n = 1, 2, 3) origins per cell, and this synchrony may be disturbed by manipulating the dnaA gene (30). Indices of asynchrony have been calculated and are shown in Table 2. Considering the difference in expression of the two proteins, we conclude that the induced DnaA_{V. harveyi} protein disturbs the synchrony less than the DnaA_{E. coli} protein does, probably because of its lower initiation capacity.

This experiment demonstrated that the dnaA_{V. harveyi} gene can replace the dnaA_{E. coli} gene and control the DNA replication in E. coli in a manner similar to that of the dnaA_{E. coli} gene. By contrast, the dnaA genes of P. putida (17) or H. influenzae (Skovgaard, unpublished) cannot complement the dnaA46 mutation and cannot control the initiation of chromosomal DNA replication in E. coli.

Stoichiometry and equilibrium binding constants of the heterologous dnaA-dnaAp complexes

We used the heterologous DnaA_{E. coli} protein to obtain information about complex formation at the dnaA_{V. harveyi} promoter region, since the DnaA_V. harveyi and the DnaA_E. coli proteins function nearly identically. We used the BIAcore instrument to analyze equilibrium binding constants and stoichiometry of the complexes (33). BIAcore experiments are based on surface plasmon resonance (SPR), which measures the change of mass in real time at the surface of a chip. Single DnaA boxes or different combinations of binding sites on biotin-labeled oligonucleotides were bound to a streptavidin-coated chip, and a flow of E. coli ATP- or ADP-DnaA was applied past the chip. Results of representative binding experiments are shown in Fig. 3, and stoichiometries are summarized in Table 3.

FIG. 3.

Examples of binding reactions of ATP- and ADP-DnaA protein measured by SPR. dnaA_{E. coli} protein (18) was incubated with two different dnaAP_{V. harveyi} fragments (fragments I and VII from Table 3). ATP and ADP DnaA proteins bind with a typical 1:1 interaction at fragment I with DnaA box 3. For oligonucleotide VII including DnaA boxes 1 and 2 plus surrounding sequence, we found weak cooperative binding of ADP-DnaA and strong cooperative binding of ATP-DnaA. Lengths of the oligonucleotides and relative positions of the DnaA boxes are indicated. One resonance unit (RU) corresponds to a change in mass by 1 pg/mm². Within each sensorgram individual curves were obtained with the following protein concentrations (bottom to top): 0.39, 0.78, 1.6, 3.2, 6.3, 12.5, 25, 50, and 100 nM. Determinations of stoichiometry and equilibrium binding constants were performed as described in reference 33.

TABLE 3.

Binding of E. coli ATP- and ADP-DnaA protein to different oligonucleotides with different binding sites from the V. harveyi dnaA promoter region measured with SPR

^aSequences of the 5′-biotinylated strand are shown. Bold, DnaA box sequences; bold and italic, potential ATP-DnaA boxes; underlined, mutated sequence. Coordinates refer to Fig. 1. The complementary strand was not biotinylated.

^bThe stoichiometry factor corresponds to the number of DnaA proteins per DNA molecule at a concentration of 100 nM DnaA, approaching saturation of the DnaA box(es) with protein.

DnaA box 3, overlapped by dnaAp₃, has one mismatch to the E. coli consensus sequence and was bound by ATP- and ADP-DnaA protein in an identical manner (Fig. 3). The data perfectly matched the Langmuir binding model describing a 1:1 interaction. The equilibrium binding constant was 4 nM, as determined by using the BIAevaluation software. This is similar to binding constants found for the DnaA boxes of oriC of E. coli (28). Oligonucleotide II, including DnaA box 3 plus eight additional bases, gave identical results. Only oligonucleotide III, which was more than 30 bp longer, showed a strong difference in ATP- and ADP-dependent binding. With oligonucleotide III we observed at 100 nM a stoichiometry of 1 for ADP-DnaA protein. ATP-DnaA showed complex binding kinetics, compatible with cooperative binding and a stoichiometry of 3.5, although saturation of the binding reaction was not reached. To demonstrate that, in addition to the DnaA box 3, ATP-DnaA recognizes another site specifically, we scrambled the sequence surrounding the DnaA box 3. Oligonucleotide III-mut showed a significantly reduced stoichiometry for ATP-DnaA but not for ADP-DnaA, indicating that ATP-DnaA recognizes specifically DNA sequences upstream and downstream of DnaA box 3.

At dnaAp₂ two potential DnaA boxes, 1 and 2, were found in tandem orientation with 5 bp between them. Oligonucleotide IV contains only the DnaA box 2 plus some downstream sequence. There was almost no difference in the stoichiometry for ATP- and ADP-DnaA with this oligonucleotide. With ADP-DnaA we measured a binding constant of 8 nM. The neighboring DnaA box 1, located on oligonucleotide V including more upstream sequence, was not bound by DnaA protein. Interestingly, DnaA boxes 1 and 2 on oligonucleotide VI gave a stoichiometry of 2.3 with ATP-DnaA and a stoichiometry of 1.4 with ADP-DnaA, indicating that DnaA box 2 promotes binding to DnaA box 1. However, ADP-DnaA seems to bind DnaA box 1 less effectively. The long oligonucleotide VII, consisting of oligonucleotides IV and V, showed a stoichiometry of 1.6 with ADP-DnaA, comparable to the two-box oligonucleotide VI, and a stoichiometry of 4.9 with ATP-DnaA (Table 3). Saturation was not reached at the highest concentration used (100 nM), as was seen with oligonucleotide III. These results indicate that binding to sequences upstream of DnaA box 1 depends on the consensus DnaA box 2, which therefore serves as an anchor. Because ATP-DnaA binds with a higher stoichiometry than ADP-DnaA to sequences flanking the 9-mer DnaA boxes and because these sequences are strongly similar to 6-mer ATP-DnaA boxes from the dnaA_{E. coli} promoter (33), we conclude that the dnaA_V. harveyi promoter also contains ATP-DnaA boxes. The proposed ATP-DnaA boxes are indicated in Table 3. The consensus of those sequences is aGAtcg, which is the potential ATP-DnaA box sequence of V. harveyi. Since the binding patterns of DnaA to the promoters of E. coli and V. harveyi are very similar and dnaA gene regulations in vivo are nearly identical (Table 2), our results suggest that the ATP/ADP switch of DnaA controls dnaA transcription in both bacteria.

Evolutionary aspects of the gene organization, GATC sites, and the ATP-DnaA box sequence

Figure 4 compares the organization of genes in the dnaA region of P. putida, E. coli, V. harveyi, and H. influenzae. The gene organization of the P. putida dnaA region is believed to be very similar to the ancestor of these species (41). The different translocations suggest that the acquisition of Dam methylase in the ancestor to Enterobacteriaceae, Vibrionaceae, and Pasteurellaceae must have been the primary event, followed by the integration of the Dam methylase into the mechanism controlling that each origin is initiated only once per cell cycle. This allowed separation of oriC from dnaA and thereafter insertions between rpmH and dnaA. Simultaneously, acquisition of Dam methylase led to the introduction of many GATC sites at the dnaA promoter and the oriC. Two clusters of GATC sites are found in the V. harveyi dnaA gene region (Fig. 1) (http://virgil.ruc.dk/∼olesk/). One cluster containing 10 GATC sites is located in the dnaA promoter region. A second cluster containing six GATC sites within 130 bp is located in the middle of the structural gene (http://virgil.ruc.dk/∼olesk/). These clusters are conserved in the dnaA genes of E. coli, Salmonella enterica serovar Typhimurium, and Serratia marscescens, whereas Proteus mirabilis and H. influenzae have lost this second cluster (11, 13, 16, 31, 32). The clusters of GATC sites influence cell cycle-dependent regulation of the dnaA gene in E. coli (36). The GATC sequence then merged with the ATP-DnaA box recognition sequence, so the consensus ATP-DnaA box sequence of E. coli became AGatct (33) and the consensus ATP-DnaA box sequence of V. harveyi became aGAtcg, as predicted from Table 3. The sequences are very similar, and both contain the Dam methylation sequence. On the other hand, Bacillus subtilis, which lacks the Dam system, has no GATC sequence in its ATP-DnaA boxes (C. Speck and W. Messer, unpublished data), indicating that the merging of the GATC sequence with the ATP-DnaA box was an event secondary to the acquisition of the Dam methylase.

FIG. 4.

Comparison of the dnaA regions from P. putida, E. coli, V. harveyi, and H. influenzae. Data are from Yee and Smith (40), Fujita et al. (12), Ogasawara and Yoshikawa (24), Hansen et al. (13, 16), Fleischmann et al. (11), and this work. Arrows indicate direction of transcription.

In this report we demonstrate that the two DnaA proteins can be interchanged, the transcription of V. harveyi dnaA gene is autoregulated, ATP-DnaA boxes are present in the V. harveyi dnaA promoter, and GATC sites exist in the promoter region of the V. harveyi dnaA gene. This suggests that the control of the dnaA gene expression in V. harveyi and E. coli is conserved.

Nucleotide sequence accession number

The nucleotide sequence reported in this paper has been deposited in the GenBank database under accession no. L47617.