Site-directed Mutational Analysis for the Membrane Binding of DnaA Protein

DnaA protein, the initiator of chromosomal DNA replication in Escherichia coli, interacts with acidic phospholipids, such as cardiolipin, and its activity seems to be regulated by membrane binding in cells. In this study we introduced site-directed mutations at the positions of hydrophobic or basic amino acids which are conserved among various bacteria species and which are located in the putative membrane-binding region of DnaA protein (from Asp357 to Val374). All mutant DnaA proteins showed much the same ATP and ADP binding activity as that of the wild-type protein. The release of ATP bound to the mutant DnaA protein, in which three hydrophobic amino acids were mutated to hydrophilic ones, was stimulated by cardiolipin, as in the case of the wild-type protein. On the other hand, the release of ATP bound to another mutant DnaA protein, in which three basic amino acids were mutated to acidic ones, was not stimulated by cardiolipin. These results suggest not only that the region is a membrane-binding domain of DnaA protein but also that these basic amino acids are important for the binding and the ionic interaction between the basic amino acids and acidic residues of cardiolipin and is involved in the interaction between DnaA protein and cardiolipin.

DnaA protein, the initiator of DNA replication in Escherichia coli, specifically binds to oriC, the unique sequence for the initiation of chromosomal DNA replication, and recruits other replication proteins (1)(2)(3)(4). DnaA protein exhibits a high affinity for ATP and ADP; the ATP-binding form is active in an oriC replication system in vitro, while the ADP-binding form is inactive (5). Synthesized organic compounds designed to block ATP binding to DnaA protein specifically inhibited oriC DNA replication in vitro (6). dnaA46 and dnaA5 mutants show recessive lethality and DnaA46 and DnaA5 proteins lose the affinity for ATP and ADP (7)(8)(9). These results suggest that adenine nucleotides bound to DnaA protein regulate the initiation of chromosomal DNA replication in E. coli cells.
In order to couple the initiation of DNA replication with cell division, the initiator protein should be inactivated soon after the initiation of DNA replication to prevent overinitiation and it should be activated after an appropriate period for the next replication cycle. The inactivation seems to be mediated by the intrinsic ATPase activity of DnaA protein and its activation factor based on the following evidence: (i) the induction of a mutant DnaA protein with decreased ATPase activity led to overinitiation of DNA replication in cells, resulting in a dominant lethal phenotype (10), and (ii) the stimulation factor for the ATPase was found and shown to be identical to the inactivation factor for DnaA protein (11), which was previously suggested to play an important role in the inactivation of DnaA protein (12,13). As for the activation of DnaA protein for initiation of DNA replication, membrane phospholipids were reported to activate the inactive ADP-binding form of DnaA protein.
Acidic phospholipids, in particular cardiolipin (CL), 1 decrease the affinity of DnaA protein for adenine nucleotides (14)(15)(16)(17), and activate the DnaA protein from the ADP-binding form to the ATP-binding form in the presence of high concentrations of ATP by stimulating the exchange of ADP with ATP (14). We previously reported that artificial membranes consisting of a mixture of anionic and cationic phospholipids which mimic biological membranes can decrease the affinity of DnaA protein for adenine nucleotides under conditions in which acidic phospholipids form cluster structures (17). Garner and Crooke (18) suggested that the region around Lys 372 of DnaA protein is responsible for membrane binding based on their biochemical analysis using degraded products of DnaA protein.
A potential amphipathic helix (from Asp 357 to Val 374 of DnaA protein) located in this region may be a membrane-binding domain (1,18). A recent study using a cross-linking technique with a photoactivatable phospholipid analog provided further evidence that the amphipathic helix is a membrane-binding domain (19).
In order to better understand how DnaA protein interacts with phospholipids and how it is involved in the regulation of DNA replication, we attempted to identify the amino acids of DnaA protein that are involved in membrane binding. To do this, we examined the effects of site-directed mutations on the membrane binding to DnaA protein.

EXPERIMENTAL PROCEDURES
Materials-A crude extract for oriC complementation assay was prepared from the WM433 strain of E. coli as described previously (20,21). CL was purchased from Sigma. [␣-32 P]ATP (10 mCi/mmol) and * This work was supported in part by Grants-in-Aid for Scientific Research from the Ministry of Education, Science, Sports, and Culture, Japan. 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.
Site-directed Mutagenesis and Plasmid Construction-Site-specific mutation was performed using the method of Kunkel (22). In brief, uracil-containing single-stranded DNA of M13 phage which contains the coding region of the dnaA gene (10) was hybridized with each of the oligonucleotide primers, 5Ј-TTTTCCYGCAATGCAGAAGAGTCGCGA-GACGCCTCACGC-3Ј, 5Ј-GGTGACCAGTTCTTCCTGGAATGCCAGC-AAGTCTTCCAGCGCCTCTTCCACGAAGTCG-3, or 5Ј-GGTGACCAG-TTCTTCCTGGAATGAGAAGAGTCTTCAGACGCCTCTTCCACGAAG-TCG-3. The first primer contained a mismatch sequence for replacement of Leu 363 , Leu 366 , and Leu 367 with Ser, the second replaced Arg 360 , Arg 364 , and Lys 372 with Glu, and the third made all six of the above substitutions (i.e. Leu 363 , Leu 366 , and Leu 367 were replaced with Ser and Arg 360 , Arg 364 , and Lys 372 were replaced with Glu). (The changed bases are indicated by underlines). The complementary DNA strand was synthesized in vitro and the resultant double-stranded DNA was introduced into JM109 cells. The mutation was confirmed by DNA sequencing, and double-stranded DNAs that contains the mutations (pMZ000-30, pMZ000-31, and pMZ000-32) were prepared.
Purification of DnaA Protein-The mutant DnaA protein and the wild-type DnaA protein were purified, as described (23)(24)(25), but with some modifications. Strain KA450 (12) transformed with pMZ001-30, pMZ001-31, or pMZ001-32 was grown in 20 liters of LB medium containing 25 g/ml thymine at 37°C until the optical density at 595 nm reached 0.5, and then arabinose was added to 1%. After 1 h of incubation at 37°C, the cells were harvested by centrifugation, resuspended in buffer C (23) containing 250 mM KCl to an optical density at 595 nm of 220, and stored at Ϫ80°C. The thawed cell suspension was diluted 2-fold with buffer C containing 250 mM KCl, and spermidine-HCl and egg white lysozyme were added to final concentrations of 20 mM and 400 g/ml, respectively. After incubation at 0°C for 30 min, the suspension was centrifuged for 30 min in a Beckman 50.2 Ti rotor. The supernatant was precipitated with 0.22 mg/ml ammonium sulfate and re-suspended with buffer C. Samples were dialyzed against buffer C and the precipitates that developed were collected by centrifugation at 100,000 rpm in a Beckman TLA100.3 rotor. The precipitates were washed twice with buffer C containing 0.6 M ammonium sulfate and finally resuspended in buffer C containing 4 M guanidine-HCl and 0.6 M ammonium sulfate. Insoluble materials were removed by centrifugation at 100,000 rpm in the TLA100.3 rotor. The supernatant was fractionated by gel filtration on a Superose-12 column (Pharmacia fast protein liquid chromatography HR10/30) equilibrated with buffer D (23) at a flow rate of 0.3 ml/min and the active fractions were pooled.
Influence of CL on Release of ATP from ATP⅐DnaA Complex-The release of ATP from DnaA⅐ATP complex was examined as described (26). Liposomes of CL were prepared from dried phospholipids on the bottom of glass tubes by vigorous vortex mixing in water. The amount of phosphorus in the phospholipid fraction was determined by the method of Chen et al. (27). DnaA protein was preincubated with 2 M [␣-32 P]ATP (10 mCi/mmol) in 40 l of buffer G (50 mM Tricine-KOH (pH 8.25), 0.5 mM magnesium acetate, 0.3 mM EDTA, 7 mM dithiothreitol, 20% (v/v) glycerol, and 0.007% Triton X-100) at 4°C for 15 min. CL liposome was added and the mixture was incubated at 37°C. Samples were passed through nitrocellulose membranes (Millipore HA, 0.45 m) and washed by ice-cold wash buffer (50 mM Tricine-KOH (pH 8.25), 0.5 mM magnesium acetate, 0.3 mM EDTA, 5 mM dithiothreitol, 17% (v/v) glycerol, 10 mM ammonium sulfate, and 0.005% Triton X-100). The radioactivity remaining on the filters was counted in a liquid scintillation counter.
Filter Binding Assay for ATP or ADP Binding to DnaA Protein-ATP and ADP binding activity of DnaA protein was determined by a filter binding assay (6). DnaA protein (2 pmol) was incubated with [␣-32 P]ATP or [ 3 H]ADP at 0°C for 15 min in 40 l of buffer G. Samples were passed through nitrocellulose membranes and washed and counted as described above.
oriC DNA Replication in a Crude Extract-Replication of minichromosomes in a crude extract (Fraction II) was assayed as described (20,21). Template DNA (M13E10) (200 ng, 600 pmol at nucleotides), 240 g of Fraction II from WM433 (dnaA204), and DnaA protein were mixed with reaction mixtures (20, 21) and incubated for replication at 30°C for 20 min. The reaction was terminated by chilling on ice and adding 10% trichloroacetic acid. Samples were passed through Whatman GF/C glass-fiber filters. The amount of radioactivity on the filters was measured in a liquid scintillation counter, and the amount of DNA synthesized (picomoles of nucleotides) was calculated (20,21).

Site-directed Mutation for the Functional Interaction between
Acidic Phospholipids and DnaA Protein-Since, in general, protein-lipid interactions are mediated by hydrophobic interactions between the hydrophobic moiety of lipids and hydrophobic amino acids in proteins, the amino acids most likely to be involved in the interaction of DnaA protein with phospholipids are hydrophobic ones. The potential amphipathic helix (from Asp 357 to Val 374 of DnaA protein) contains many hydrophobic amino acids (Fig. 1). A comparison of the amino acid sequences of DnaA protein from various bacteria species revealed that Leu 363 , Leu 366 , and Leu 367 of E. coli DnaA protein are well conserved as hydrophobic amino acids (1). Thus, these three leucines are considered to be important for the interaction between DnaA protein and phospholipids.
It has been reported that anionic residues in phospholipids are indispensable for their interaction with DnaA protein (14 -17). Thus, besides the hydrophobic interaction, there is also a possibility that ionic interaction between anionic residues of phospholipids with basic amino acids in DnaA protein are involved in the functional interaction of the protein with phospholipids. As shown in Fig. 1, the amphipathic domain contains three basic amino acids (Arg 360 , Arg 364 , and Lys 372 ), which are well conserved among DnaA protein from various bacteria species (1). Thus, these three basic amino acids may also be involved in the interaction between DnaA protein and phospholipids.
We introduced site-directed mutations into the dnaA gene to convert Leu 363 , Leu 366 , and Leu 367 to Ser (a hydrophilic amino acid) or Arg 360 , Arg 364 , and Lys 372 to Glu (an acidic amino acid) to construct the mutant dnaA genes, dnaA430, or dnaA431, respectively. Considering the possibility that both groups of amino acids are required for the interaction of DnaA protein with phospholipids, a mutant dnaA gene (dnaA432) which has mutations of both Leu 363 , Leu 366 , and Leu 367 to Ser and Arg 360 , Arg 364 , and Lys 372 to Glu was also prepared. The coding region of the dnaA430, dnaA431, or dnaA432 gene was conjugated with the promoter of the arabinose operon to construct a plasmid for overproduction of the mutated DnaA protein (DnaA430, DnaA431, or DnaA432). To avoid contamination of the wild-type DnaA protein in the fraction of the mutant DnaA protein, the KA450 strain (⌬oriC1071::Tn10, rnhA199(Am), dnaA17(Am), trpE9828(Am), tyrA(Am), thr, ilv, and thyA) in which the dnaA gene on chromosomal DNA was deleted (12) was used as host cells for overproduction. The viability is not dependent on the function of DnaA protein in KA450. Addition of 1% arabinose caused overexpression of each mutant DnaA protein (data not shown). Purification of each mutant DnaA protein was done as described under "Experimental Procedures." At the final step of purification (gel- filtration column chromatography) both the monomer and aggregated forms of DnaA430 and DnaA432 proteins were recovered (data not shown), as was the case of wild-type DnaA protein (Fig. 2) (23). However, only the monomer form of the DnaA431 protein was recovered (Fig. 2). The recovery of DnaA 431 protein after column chromatography was not significantly different from the recoveries of the other types of DnaA protein. The aggregated form of DnaA protein has been suggested to be a complex of DnaA protein with phospholipids, based on the finding that the aggregated form of DnaA protein is activated to the level of the monomer form by phospholipase (28). Thus, the result in Fig. 2 suggests that DnaA431 protein does not have an affinity for phospholipids.
Characterization of ATP and ADP Binding Activity of the Mutant DnaA Proteins-The functional interaction of DnaA protein with phospholipids is monitored by phospholipid-dependent stimulation of the release of ATP (or ADP) from DnaA protein (23). The activation of the ADP-binding form of DnaA protein by phospholipids is mediated by this function of phospholipids on DnaA protein (23). Thus, the mutant DnaA proteins were required to maintain their ATP binding activity in order to examine their functional interaction with phospholipids. The ATP binding activities of these mutant proteins were examined by a filter-binding assay (6). As shown in Fig. 3A, each mutant DnaA protein (DnaA430, DnaA431, or DnaA432) showed nearly the same ATP binding activity as the wild-type protein. The K d values of DnaA430, DnaA431, DnaA432 and the wild-type protein for ATP were determined to be 28, 19, 28, and 30 nM, respectively. The numbers of ATP-binding sites per DnaA430, DnaA431, DnaA432, and wild-type proteins were calculated to be 0.40, 0.46, 0.39, and 0.42, respectively. The K d value and the number of ATP-binding sites of wild-type protein were nearly the same as reported previously (5). We also examined the binding of ADP by the mutant DnaA proteins in the same manner. As shown in Fig. 3B, each of the mutant DnaA proteins maintained its ADP binding activity. These results enabled us to examine their functional interaction with phospholipids. The results also suggest that these amino acids in the amphipathic helix domain are not necessary for the adenine-nucleotide binding activity of DnaA protein.
Functional Interaction of the Mutant DnaA Proteins with CL-To examine the functional interaction of DnaA430, DnaA431, and DnaA432 proteins with CL, we investigated the effect of CL on the release of ATP from DnaA⅐ATP complex.
DnaA protein bound to [␣-32 P]ATP was incubated with CL and the remaining ATP was determined by the filter-binding assay. As shown in Fig. 4, the rates of release of ATP from each of the mutant proteins in the absence of CL were nearly the same as that of the wild type protein, suggesting that the K d values for ATP of DnaA protein were not significantly affected by these mutations (see above). On the other hand, the release of ATP in the presence of CL was greatly affected by the mutations. CL did not stimulate the release of ATP from DnaA431 protein, whereas it stimulated the release from DnaA430 to nearly the same extent that it stimulated the release from the wild-type protein (Fig. 4). CL slightly stimulated the release of ATP from DnaA432, which has mutations of both DnaA430 and DnaA431 (Fig. 4). These results suggest that the basic amino acids (Arg 360 , Arg 364 , and Lys 372 ) but not the hydrophobic amino acids (Leu 363 , Leu 366 , and Leu 367 ) in the amphipathic helix domain of DnaA protein are involved in the functional interac- tion between DnaA protein and CL, and that the interaction is mediated by an ionic interaction between the anionic residues of phospholipids and the basic amino acids of DnaA protein.
These results also support the hypothesis that the amphipathic helix domain (from Asp 357 to Val 374 of DnaA protein) is the membrane-binding domain of DnaA protein (1, 18, 19).
Replication Activity of the Mutant DnaA Proteins in Vitro-We measured the activities of DnaA430, DnaA431, and DnaA432 proteins for initiation of DNA replication in an oriC complementation assay (20,21). As shown in Fig. 5A, each of the mutants showed activity for DNA replication, although the activities were less than the activity of the wild-type protein.
The specific activities of DnaA430 and DnaA431 were about one-third and one-fifth, respectively, the activity of the wildtype protein, whereas the activity of DnaA432 was less than one-tenth that of the wild-type protein. DnaA A184V, DnaA46, and DnaA5 required longer incubation periods for expression of their replication activity; a time lag for the DNA replication reaction has been previously reported for these mutant DnaA proteins (7)(8)(9). In the case of these mutants, the time course of DNA replication was approximately linear as in the case of the wild-type protein (data not shown). Preincubation with each of the mutant proteins with 1 M ADP inhibited the replication activity (data not shown) as in the case of the wild-type protein (6).
The relatively low specific activities of the mutant DnaA proteins seems not to be due to denaturation as a result of the purification procedures, because the specific activities of the crude extract fractions of the mutant DnaA proteins were also lower than the specific activity of the wild-type protein. The mutated amino acids may affect the higher order structure of DnaA protein, resulting in the low specific activities.
We also examined the effect of CL on the replication activity of the mutant and the wild-type DnaA proteins. It was shown that CL inhibited the replication activity of the nucleotide-free form of DnaA protein through inhibition of the ATP-binding to DnaA protein (23). As shown in Fig. 5B, the replication activity of DnaA431 was more resistant to CL than DnaA430 and the wild-type DnaA protein. The concentrations of CL needed to 50% inhibition of the replication activity of DnaA430, DnaA431, and the wild-type DnaA protein were determined to be 1.9, 6.5, and 1.8 M, respectively. The result supports the notion that the functional interaction of DnaA431 but not DnaA430 with CL is weakened by the mutations and thus, the basic amino acids are important for the functional interaction. DISCUSSION The replication of chromosomal DNA is initiated in coordination with cell division (membrane division). A replicon model was proposed to explain this coordination (29). DNA replication in bacterial cells is initiated on membranes, and the activity of the initiator protein for DNA replication is regulated by membrane components (29). Thus, the interaction between DnaA protein and acidic phospholipids has received much attention because it may provide biochemical support for the replicon model. Acidic phospholipids, such as CL, enhance the exchange reaction of ADP bound to DnaA protein with ATP, which results in activation of DnaA protein (14 -17). Thus, it was proposed that acidic phospholipids activate DnaA protein to initiate the chromosomal DNA replication in cells (14). To test this hypothesis, identification of amino acids involved in the interaction is important. In this study, we showed that DnaA431, which has mutations (Arg 360 , Arg 364 , and Lys 372 ) in the amphipathic helix (from Asp 357 to Val 374 of DnaA protein) did not have a functional interaction with CL. The behavior of the DnaA431 protein in gel filtration chromatography is consistent with the notion that DnaA431 protein cannot form a complex with CL. These results strongly suggest that the amphipathic helix is the membrane-binding domain of DnaA protein as suggested previously (1,18,19).
We also showed that DnaA430 protein, which has mutations of hydrophobic amino acids in the amphipathic helix, maintained an activity for the functional interaction with CL. Thus, the basic amino acids (Arg 360 , Arg 364 , and Lys 372 ) but not the hydrophobic amino acids (Leu 363 , Leu 366 , and Leu 367 ) seem to be involved in the interaction. It is natural to suppose that the positively charged basic residues interact with the negatively charged phosphates of CL through an electrostatic interaction. Fig. 6 depicts a plausible form of the ionic interaction between DnaA protein and CL. The location of each amino residue is putatively defined by considering the general structural parameters for an ␣-helical wheel. It is noteworthy that all three of the basic residues are placed on one side of the helical wheel and thus, provide a cationic domain. This configuration allows both of the two phosphates of CL to bind all or part of the three basic residues. A multisite electrostatic interaction may perturb the affinity of DnaA protein for adenine nucleotides.
Garner and co-workers (19) recently proposed a two-step interaction of DnaA protein with acidic phospholipids. At first, DnaA protein adsorbs onto the anionic lipid bilayer surface, and then some part of the protein penetrates into the hydrophobic interior of the membrane (19). The results of this study support the hypothesis that the first step of the interaction is mediated by an ionic interaction between the anionic residues of CL with the basic amino acids of DnaA protein. The second step may be achieved by an interaction between the hydrophobic region of the phospholipids with the hydrophobic amino acids of DnaA protein. However, the present study suggests that hydrophobic amino acids in the amphipathic helix are not involved in this interaction. Besides these amino acids (Leu 363 , Leu 366 , and Leu 367 ), there are some other hydrophobic amino acids that are conserved among DnaA proteins from various bacteria spices (1). Some of these amino acids may be involved in the second step of the interaction.