Spo0J regulates the oligomeric state of Soj to trigger its switch from an activator to an inhibitor of DNA replication initiation

Authors


E-mail heath.murray@ncl.ac.uk; Tel. (+44) 191 208 3233; Fax (+44) 191 208 3205.

Summary

Control of DNA replication initiation is essential for bacterial cells to co-ordinate the faithful replication and segregation of their genetic material. The Bacillus subtilis ATPase Soj is a dynamic protein that regulates DNA replication initiation by either inhibiting or activating the DNA replication initiator protein DnaA. Here we report that the key event which switches Soj regulatory activity is a transition in its oligomeric state from a monomer to an ATP-dependent homodimer capable of DNA binding. We show that the DNA binding activity of the Soj dimer is required both for activation of DNA replication initiation and for interaction with Spo0J. Finally, we demonstrate that Spo0J inhibits Soj dimerization by stimulating Soj ATPase activity. The data provide a molecular explanation for the dichotomous regulatory activities of Soj, as well as assigning unique Soj conformations to distinct cellular localization patterns. We discuss how the regulation of Soj ATPase activity by Spo0J could be utilized to control the initiation of DNA replication during the cell cycle.

Introduction

Initiation of DNA replication must be co-ordinated with the cell cycle to ensure accurate chromosome segregation. In bacteria there are often multiple overlapping systems that orchestrate DNA replication initiation by regulating the activity of the DNA replication initiator protein DnaA (Katayama et al., 2010). In the model bacterium Bacillus subtilis the ATPase Soj acts as a molecular switch to regulate the activity of the DNA replication initiator protein DnaA (Murray and Errington, 2008). Soj is a member of the ParA family of ATPases and its gene is transcribed from a highly conserved module composed of soj(parA), spo0J(parB) and parS (Gerdes et al., 2000; Livny et al., 2007). parS sites are cis-acting DNA sequence motifs located proximal to the DNA replication origin that nucleate the spreading of Spo0J into flanking regions of DNA to create large nucleoprotein complexes (Lin and Grossman, 1998; Murray et al., 2006; Breier and Grossman, 2007).

Structural and biochemical analysis of Soj orthologues has led to the model that these proteins are ATP-dependent ‘sandwich’ dimers (Leonard et al., 2005; Hester and Lutkenhaus, 2007). In the presence of ADP (or in the absence of nucleotide) Soj proteins are monomeric. ATP binding by Soj facilitates dimerization and is required for Soj to bind DNA. Soj DNA binding activity appears to be non-specific, highly cooperative and mediated by conserved arginine residues located on one face of the dimer.

Under several conditions, such as in the absence of the regulator Spo0J, when artificially overexpressed, or as a mutant allele (SojD40A) that cannot hydrolyse ATP, Soj activates DnaA to stimulate DNA replication initiation (Ogura et al., 2003; Lee and Grossman, 2006; Murray and Errington, 2008). Under other conditions, such as when weakly expressed or when converted to a mutant allele (SojG12V) that cannot bind to DNA, Soj inhibits DnaA to repress DNA replication initiation (Ogura et al., 2003; Murray and Errington, 2008). Although the ability of Soj to differentially regulate DnaA has been established, the mechanism by which Soj switches between its regulatory states could not been determined because biochemical characterization of the Soj mutants utilized in these studies was incomplete.

In this report we have investigated the contributions of dimerization and DNA binding to the regulatory activities of B. subtilis Soj. Our data indicate that DNA replication initiation is inhibited by monomeric Soj and activated by dimeric Soj. We find that negative regulation by Soj does not require its DNA binding activity, while positive regulation by Soj does. We go on to show that the regulatory protein Spo0J stimulates the ATPase activity of Soj to drive it from a dimer to a monomer and that this stimulation is dependent on two positively charged residues in the N-terminus of Spo0J. Taken together, the results suggest that dimerization of Soj is the key molecular event that switches the protein from a repressor to an activator of DNA replication initiation. Moreover, these results are likely to have implications for other members of the large ParA/ParB protein family.

Results

Dimerization capability of wild-type and mutant Soj proteins

Previous work in vitro indicated that B. subtilis Soj is a dimer in the presence of ATP and a monomer in the presence of ADP or in the absence of nucleotide (Leonard et al., 2005; Hester and Lutkenhaus, 2007). To test whether the oligomeric state of Soj affects its regulatory activity in vivo, we characterized two mutant Soj proteins that were predicted to be dimerization deficient. The first mutant (SojG12V) was based on the structure of Thermus thermophilus Soj which suggested that substitution of the conserved glycine within the P-loop near the dimer interface would cause a steric clash in the active site between two Soj monomers (Leonard et al., 2005). The second mutant (SojK16A) substituted the conserved lysine residue that was shown to be required for ATP binding (Leonard et al., 2005).

To analyse the dimerization potential of the Soj mutant proteins each was purified and assayed using size exclusion chromatography (Fig. 1A). Both SojG12V and SojK16A were found to elute as monomers in the presence of ATP, indicating that these mutants are unable to form stable dimers. As positive controls wild-type Soj and a DNA binding mutant SojR189A (Hester and Lutkenhaus, 2007) were also analysed [we note that the ATP hydrolysis-deficient mutant SojD40A, which likely forms a stable dimer (Leonard et al., 2005), could not be included in these experiments due to problems encountered during protein purification]. In the presence of ATP both of these proteins eluted from the column earlier than they did in the presence of ADP or in the absence of nucleotide, thus indicating ATP-dependent dimerization. Far-UV circular dichroism spectroscopy showed that the purified Soj proteins had indistinguishable secondary structure compositions (Fig. 1B), indicating that the inability of SojG12V and SojK16A to dimerize was not due to protein misfolding. Moreover, UV cross-linking confirmed that SojG12V was able to bind ATP (Fig. 1C), consistent with the model that this mutant is monomeric because the substitution of glycine for valine introduces a steric clash at the dimer interface (Leonard et al., 2005).

Figure 1.

Dimerization capabilities of wild-type and mutant Soj proteins.
A. Size exclusion chromatography showing the ability of Soj proteins to dimerize in the presence of ATP, ADP, or without nucleotide (EDTA). Molecular weights were calculated from the peak retention volume by reference to Bio-Rad protein standards.
B. Circular dichroism spectroscopy indicating the secondary structure of purified Soj proteins. Each spectrum is the product of 10 individual scans.
C. Autoradiograph of [α-32P]-ATP UV cross-linked to Soj. Lane 1, Soj. Lane 2, SojG12V. Lane 3, SojK16A. Lane 4, SojR189A. Lane 5, Soj with MgCl2 omitted. The position of Soj proteins, as judged by Coomassie staining, is indicated by the arrow.
D. Bacterial two-hybrid analysis of Soj proteins. Results from β-galactosidase assays are shown as the average ± standard deviation from three independent transformants. Empty vector (pKT25+pUT18C), soj (pHM371+pHM376), sojG12V (pHM372+pHM377), sojK16A (pHM373+pHM378), sojR189A (pHM374+pHM379).

We found that the Soj dimerization deficient mutants were unable to stably interact with plasmid DNA using an electrophoretic mobility shift assay (Fig. S1A), consistent with previous results demonstrating that the DNA binding activity of Soj is ATP-dependent (Leonard et al., 2005; Hester and Lutkenhaus, 2007). The DNA binding activity of Soj appears highly cooperative and it has been found that homologous ParA proteins can either form nucleoprotein filaments (T. thermophilus; Leonard et al., 2005) or assemble into filaments in the absence of DNA (Caulobacter crescentus; Ptacin and Shapiro, 2010). We wondered whether the DNA binding activity of Soj might involve filament formation and therefore we evaluated the ability of Soj to form filaments in vitro using 90 degree angle light scattering (Fig. S1B) and electron microscopy (data not shown). We were unable to detect the assembly of Soj into higher-order complexes using either approach (with or without DNA), and we note that Soj complexes larger than dimers were not observed by size exclusion chromatography (data not shown). However, we cannot rule out the possibility that Soj forms filaments under different reaction conditions.

The ability of the Soj proteins to dimerize in vivo was probed using an Escherichia coli two-hybrid system that utilizes a split adenylate cyclase to activate lacZ gene expression when the two halves of the enzyme are brought together (Karimova et al., 1998). Wild-type and mutant soj alleles were fused to the adenylate cyclase fragments and the ability to dimerize was assayed using a β-galactosidase assay. An interaction was detected for the wild-type Soj protein and the SojR189A mutant, but not for the SojG12V and SojK16A mutants (Fig. 1D). We note that previously both the SojG12V and SojK16A mutants have been shown to interact with DnaA using this two-hybrid assay (Murray and Errington, 2008), suggesting that these mutants are properly expressed and folded. Taken together with the in vitro analyses, the results indicate that both the SojG12V and SojK16A mutants cannot dimerize and instead exist as monomers.

Activity of Soj proteins on DNA replication initiation

To analyse the activities of the wild-type and mutant Soj proteins on DNA replication initiation, each allele was transferred into the chromosome of a Δsoj strain under the control of a xylose-inducible promoter. Soj proteins were overexpressed and their effects on DNA replication initiation were determined by marker frequency analysis (we hypothesized that high levels of protein expression would have two advantages: accentuating the regulatory activity of Soj proteins and promoting Soj dimerization). Compared with the uninduced controls, both of the Soj proteins that were capable of dimerization (wild type and SojR189A) showed increased DNA replication initiation, while both of the Soj proteins that were unable to dimerize (SojG12V and SojK16A) inhibited DNA replication initiation (Fig. 2A). Western blot analysis showed that all of the Soj proteins were expressed to a similar level, signifying that the difference in protein activities was not due to variations in protein concentration (Fig. 2B). Interestingly, the SojK16A mutant was not as potent an inhibitor of DNA replication initiation as the SojG12V mutant, suggesting that nucleotide binding enhances the activity of the Soj monomer. Additionally, the SojR189A mutant was not as potent an activator of DNA replication initiation as the wild-type protein, indicating that DNA binding facilitates positive regulation of DNA replication by Soj (see next section below).

Figure 2.

Effect of wild-type and mutant Soj proteins on DNA replication initiation.
A. The oriC-to-terminus ratio of each Soj protein was determined using marker frequency analysis. Cells were grown in CH medium at 30°C with or without xylose (1%). The average ± standard deviation from three technical replicates is shown. The dashed line represents the average ratio for the parent strain.
B. Western blot analysis of Soj protein overexpression. Strains were grown as described above except that xylose was added after the cultures had reached the exponential growth phase. Empty vector (HM359), soj (HM558), sojG12V (HM240), sojK16A (HM625), sojR189A (HM626).

To confirm the results of the marker frequency analysis, DNA replication was assayed by observing nucleoid localization within single cells. Overexpression of either SojG12V (Fig. S2C) or SojK16A (Fig. S2D) caused a dramatic decrease in the number of nucleoids per cell, consistent with severe inhibition of DNA replication initiation. Conversely, overexpression of wild-type Soj (Fig. S2B) led to cells containing multiple nucleoids, consistent with activation of DNA replication initiation. Overexpression of SojR189A (Fig. S2E) did not appear to appreciably affect nucleoid size or distribution, consistent with the modest increase in DNA replication initiation determined by marker frequency analysis.

The DNA binding activity of Soj is required for activation of DnaA

The overexpression analysis described above suggested that the DNA binding activity of Soj facilitates maximal activation of DNA replication initiation. In order to further assess the requirement of DNA binding for positive regulation by Soj, we investigated both the localization and the regulatory activity of the SojR189A mutant expressed from its endogenous locus.

Wild-type GFP–Soj localizes at septa and forms foci at the oriC region of the chromosome (Fig. 3A). In a Δspo0J mutant GFP–Soj relocalizes to the nucleoid (Fig. 3B). The localization of the GFP–SojR189A mutant was indistinguishable from GFP–Soj in the presence of Spo0J (Fig. 3C), but in contrast GFP–SojR189A did not relocalize to the nucleoid in a Δspo0J mutant (Fig. 3D). These results indicate that GFP–SojR189A maintains the ability to interact with factors required for its localization at septa and at oriC, but that it is unable to bind to the nucleoid, consistent with the model that SojR189A is unable to bind to DNA in vivo.

Figure 3.

Soj DNA binding activity is required to activate DNA replication initiation.
A–D. The localization of GFP–SojR189A in the presence and absence of Spo0J was observed using epifluorescence microscopy. Cells were grown in CH medium at 30°C. (A) gfp-soj (HM4), (B) gfp-sojΔspo0J (HM13), (C) gfp-sojR189A (HM308), (D) gfp-sojR189AΔspo0J (HM310).
E. The oriC-to-terminus ratio of SojR189A was determined using marker frequency analysis. Cells were grown in CH medium at 30°C. The average ± standard deviation from three technical replicates is shown. The dashed line represents the average ratio for the parent strain. Wild type (HM34), Δspo0J (HM42), Δsoj (HM161), ΔsojΔspo0J (HM183), sojR189A (HM416), sojR189AΔspo0J (HM417).
F. The oriC-to-terminus ratio of SojD40A,R189A was determined using marker frequency analysis. Cells were grown in CH medium at 30°C. The average ± standard deviation from three technical replicates is shown. The dashed line represents the average ratio for the parent strain. Wild type (HM34), sojD40A (HM36), sojR189A (HM416), sojD40A,R189A (HM696).
G. The oriC-to-terminus ratio of SojR189A (HM626) expressed at increasing concentrations was determined using marker frequency analysis. Cells were grown in CH medium at 30°C with increasing concentrations of xylose (0%, 0.0016%, 0.008%, 0.04%, 0.2%, 1%). The average ± standard deviation from three technical replicates is shown. Below the amount of SojR189A expressed at each xylose concentration was determined by Western blot and compared with the expression of SojR189A from its native locus (HM416; see hatched box). Western blot analysis of DivIVA was used as a loading standard.

Next, marker frequency analysis was used to determine the effect of SojR189A on DNA replication initiation in the absence of Spo0J (Fig. 3E). In a Δspo0J mutant the wild-type Soj protein causes an increase in the ratio of DNA at the origin of replication compared with DNA at the terminus of replication, indicating that Soj activates DNA replication initiation. In contrast, the SojR189A mutant did not activate DNA replication in a Δspo0J mutant, indicating that the DNA binding activity of Soj is required for positive regulation. Moreover, the SojR189A mutant displayed an ori/ter ratio that was below the wild-type ratio, rather than the increased ori/ter ratio characteristic of a Δsoj mutant. Although the ∼10% decrease in the ori/ter ratio observed in the SojR189A mutant is modest, it has a profound effect on cell cycle regulation and produces a large number of cells that lack DNA (0.49% for wild type compared with 6.8% for SojR189A; Fig. S3A). We suggest that this phenotype arises due to delayed DNA segregation as a consequence of a decreased rate of DNA replication initiation; consequently, as the cell wall elongates it creates a gap between the cell pole and the nucleoid where the cell division machinery can assemble, leading to the production of ‘stumpy’ cells lacking DNA. Taken together these results suggest that the DNA binding activity of Soj is not required for it to inhibit DnaA. To substantiate this hypothesis we created a sojG12V,R189A double mutant under the control of a xylose-inducible promoter. Overexpression of the SojG12V,R189A protein caused the same inhibition of DNA replication initiation as SojG12V (Fig. S2C and F), verifying that the DNA binding activity of Soj is not required for negative regulation of DNA replication initiation.

Interestingly, we observed that the effect of the SojR189A mutant on DNA replication initiation was dependent upon the assay: SojR189A activated DNA replication initiation when overexpressed (Fig. 2A) but inhibited DNA replication initiation when expressed from its native locus (Fig. 3E). Although this mutant clearly has the ability to dimerize (Fig. 1), its localization and activity suggested that it was actually monomeric under the latter condition (Fig. 3). To investigate this possibility we constructed a sojD40A,R189A double mutant that was expressed from the native locus; this aspartic acid is required for ATP hydrolysis and it has been shown that mutation of this residue locks T. thermophilus Soj as a dimer (Leonard et al., 2005). If natively expressed SojR189A is dimeric in vivo, then the SojD40A,R189A double mutant should have properties similar to the SojR189A single mutant. We tested the effect of the double mutant on DNA replication initiation using marker frequency analysis and on protein localization using a GFP tag. In contrast to the SojR189A single mutant, we found that the SojD40A,R189A double mutant displayed a significantly elevated initiation rate (Fig. 3F) and its localization was clearly different from the GFP–SojR189A single mutant (Fig. S3B). The SojD40A,R189A double mutant was expressed to a similar level as the wild type and the Soj single mutants (Fig. S4A), indicating that the observed increase in DNA replication was not due to changes in protein expression. We did find that the GFP–SojD40A,R189A double mutant was consistently underexpressed compared with other GFP–Soj proteins (Fig. S4B). However, we did not detect any degradation of this fusion protein, indicating that the observed diffuse localization pattern was not due to free GFP. Moreover, we note that low basal expression of GFP–Soj from the xylose-inducible promoter does not preclude this chimera from clearly localizing at cell poles (Murray and Errington, 2008; Fig. S1A), suggesting that underexpression cannot account for protein mislocalization.

The results using the SojD40A,R189A double mutant are inconsistent with the model that SojR189A is dimeric when expressed from its native locus, and they suggested that the overexpression of SojR189A might alter the oligomeric status and activity of the protein. To test this hypothesis we titrated the level of inducer to incrementally increase the expression of SojR189A from the xylose-inducible promoter and correlated this with the rate of DNA replication initiation. Figure 3G shows that at low induction levels SojR189A decreases the rate of DNA replication initiation. Importantly, the level of protein expression observed under these conditions was similar to the expression of SojR189A from its native locus (hatched box in Fig. 3G). As the expression level of SojR189A was further increased, this resulted in a marked upshift in the rate of DNA replication initiation. Taken together with the results using the SojD40A,R189A double mutant shown above, the data suggests that SojR189A is a monomer when expressed from its native locus, but that overexpression drives the protein into a dimer.

Regulation of Soj through the activation of its ATPase activity by Spo0J

Since Soj is a monomer either in the presence of ADP or without nucleotide, one way for Soj to switch from a dimer to a monomer could be for it to hydrolyse its bound ATP molecules. To begin examining this possibility the intrinsic ATPase activity of wild-type Soj was determined using a malachite green assay. Figure 4 shows that Soj has a low level of intrinsic ATPase activity that is just above the background level.

Figure 4.

Efficient stimulation of Soj ATPase activity requires Spo0J and DNA. Reaction components are described in the lower table and are associated with symbols shown in the time-courses above. The DNA substrate was a 55-base-pair fragment from the spo0J gene containing a single parS site. Error bars represent the standard error of the mean (n = 3). Rate of phosphate released was calculated using a potassium phosphate standard along with a linear regression analysis (coloured lines) for each data set. A control reaction containing ATP alone released 0.8 moles of Pi per hour.

It has been found that T. thermophilus Spo0J stimulates the ATPase activity of T. thermophilus Soj (Leonard et al., 2005). To determine whether B. subtilis Spo0J could act in a similar manner to regulate Soj, the Spo0J protein was purified and the ATPase activity of Soj was determined in the presence of Spo0J. In sharp contrast to the result with Soj alone, Spo0J stimulated the ATPase activity of Soj more than 50-fold (Fig. 4), much greater than the degree of stimulation observed for T. thermophilus Soj. The ATPase activities of SojG12V and SojK16A were not stimulated by Spo0J, suggesting that Soj must form a dimer to become competent for Spo0J-mediated ATPase stimulation.

We observed that the dramatic stimulation of Soj ATPase activity by Spo0J was dependent upon DNA in the reactions (Fig. 4). To further investigate the role that DNA plays in the Soj:Spo0J interaction we used a surface plasmon resonance (SPR) assay. Double-stranded DNA encoding a single parS site was immobilized onto the surface of a SPR sensor chip and Spo0J was injected at a concentration that rapidly reached equilibrium (Fig. S5). Subsequently, the Spo0J injection was terminated and Soj was immediately injected over the Spo0J:DNA complexes in the presence and absence of various nucleotides. Figure 5A shows that wild-type Soj only interacted with the Spo0J:DNA complexes in the presence of ATP, indicating that Soj needs to dimerize in order to form a complex with Spo0J. We then repeated this experiment using the SojR189A DNA binding mutant. Figure 5B shows that SojR189A did not significantly interact with Spo0J, even though this mutant is able to dimerize in the presence of ATP. This result is consistent with the observation that Spo0J-dependent foci formation by GFP–SojD40A is dependent upon the DNA binding activity of Soj (Fig. S3B; Murray and Errington, 2008). Taken together, the ATPase and SPR experiments indicate that Soj must form a dimer and bind to DNA in order to be properly orientated so that it can efficiently interact with Spo0J:DNA complexes to stimulate its ATPase activity.

Figure 5.

Interaction of Soj with Spo0J:DNA complexes. Spo0J proteins (500 nM) were injected over DNA substrates for 200 s, followed by an injection of Soj proteins (1500 nM) with or without nucleotide for 300 s during the Spo0J dissociation phase.
A. Injection of wild-type Spo0J followed by wild-type Soj.
B. Injection of wild-type Spo0J followed by SojR189A.
C. Injection of Spo0JK3A,K7A followed by wild-type Soj.
D. Injection of Spo0JK3R,K7R followed by wild-type Soj.

Lysines 3 and 7 of Spo0J are required to control Soj activity

Thermus thermophilus Spo0J, as well as many orthologous plasmid proteins, contain arginine residues within the N-terminal tail of the protein that have been shown to be required for activating their cognate ATPases (Leonard et al., 2005; Barilla et al., 2007; Ah-Seng et al., 2009). B. subtilis Soj does not contain an arginine residue within this region; however, it does contain two lysine residues at amino acid positions 3 and 7.

To determine whether these lysines are involved in the stimulation of Soj ATPase activity in vitro, a double alanine substitution was constructed and purified. Addition of Spo0JK3A,K7A to Soj resulted in an approximately ninefold decrease in ATPase activity compared with wild-type Spo0J, indicating that these lysine residues play an important role in regulating Soj (Fig. 4). Interestingly, we found that these lysine residues could be substituted with arginine residues with almost no loss of ATPase stimulation (Fig. 4). This result differs from previous reports for plasmid ParA homologues, in which the analogous arginine residues could not be substituted with lysines (Barilla et al., 2007; Ah-Seng et al., 2009).

To further understand the role of the Spo0J N-terminal lysine residues, we investigated the ability of Soj to interact with the Spo0JK3A,K7A and Spo0JK3R,K7R mutants using SPR. Consistent with results from the ATPase assays, the Spo0JK3A,K7A mutant was unable to form a complex with Soj (Fig. 5C). In contrast the Spo0JK3R,K7R mutant interacted with Soj similarly to wild-type Spo0J (Fig. 5D). In spite of these differences, both Spo0J mutants bound to DNA in a similar manner to wild-type Spo0J.

To determine whether these lysine residues in Spo0J are required to regulate Soj in vivo, we replaced the endogenous spo0J gene with both single and double alanine substitutions and assessed the activity of Soj using marker frequency analysis and the localization of GFP–Soj. Marker frequency analysis revealed that each of the single mutants overinitiated DNA replication, with the K3A mutation having the strongest effect (Fig. 6A). The degree of over-replication observed in the Spo0JK3A,K7A double mutant appeared to be the additive effect of the two single substitutions and was indistinguishable from that of the Δspo0J null mutant (Fig. 6A), indicating that these lysine residues are critical for the regulation of Soj in vivo. We found that the lysine mutations did not affect the ability of Spo0J–GFP to localize as foci, indicating that these mutations do not interfere with Spo0J dimerization or DNA binding (Fig. S6).

Figure 6.

N-terminal lysine residues of Spo0J regulate Soj localization and activity.
A. The oriC-to-terminus ratio of Spo0J mutants was determined using marker frequency analysis. Cells were grown in CH medium at 30°C. The average ± standard deviation from three technical replicates is shown. The dashed line represents the average ratio for the parent strain. Wild type (HM34), Δspo0J (HM42), spo0JK7A (HM329), spo0JK3A (HM337), spo0JK3A,K7A (HM338).
B. The localization of GFP–Soj in the presence of Spo0J N-terminal lysine mutants was observed using epifluorescence microscopy. Cells were grown in CH medium at 30°C. Patches of GFP–Soj are indicated by yellow asterisks. gfp-soj spo0J+ (HM4), gfp-sojΔspo0J (HM13), gfp-soj spo0JK7A (HM391), gfp-soj spo0JK3A (HM392), gfp-soj spo0JK3A,K7A (HM393). Note that because the end of soj overlaps the start of spo0J, the K3A mutation eliminates the natural Soj stop codon and results in the addition of the following residues: CRPWSRD.
C. The localization of GFP–SojR189A in the presence of Spo0JK3A was observed using epifluorescence microscopy. Cells were grown in CH medium at 30°C. Patches of GFP–Soj are indicated by yellow asterisks and polar localization is indicated with arrows. gfp-soj spo0JK3A (HM392), gfp-sojR189A spo0JK3A (HM695).
D. Bacterial two-hybrid analysis of Soj in the presence of Spo0J. Strains were streaked onto nutrient agar plates supplemented with antibiotics and X-gal. All strains contained the low-copy-number vector pHM371 with cya′-soj, which was paired with the following plasmids: empty vector (pUT18C), soj (pHM376), soj spo0J (pHM393), soj spo0JK3A,K7A (pHM394).

Next the localization of GFP–Soj was examined in the context of the Spo0J lysine mutants (Fig. 6B). The trend observed followed the pattern of the marker frequency analysis. While GFP–Soj normally colocalizes with the cell septa and forms faint foci at oriC, the Spo0JK3A,K7A double mutant caused GFP–Soj to accumulate into large patches that colocalized with the nucleoid, similar to the pattern observed in a Δspo0J null mutant. The single lysine mutants caused GFP–Soj to accumulate as smaller patches most often located proximal to a cell pole. We suspected that these smaller patches represented GFP–Soj binding to DNA. To test this idea we localized the DNA binding mutant GFP–SojR189A in the context of Spo0JK3A. Figure 6C shows that the fluorescent patches were dependent upon the DNA binding residues of GFP–Soj, indicating that they represent GFP–Soj binding to the nucleoid. Western blot analysis confirmed that all of the GFP–Soj proteins were expressed to the same level (Fig. S4B), confirming that altered protein localization was not a consequence of changes in protein concentration.

Activation of Soj ATPase activity inhibits Soj dimerization

The data presented above suggest a model in which Spo0J controls Soj localization and activity by altering the oligomeric state of Soj via stimulation of its ATPase activity. To test this model we utilized the two-hybrid assay to investigate Soj self-interaction in the presence of Spo0J. Figure 6D shows that wild-type Spo0J efficiently inhibits the Soj self-interaction (white colonies), whereas the Spo0JK3A,K7A mutant does not (blue colonies). Taken together, the results presented in this work are consistent with a model in which Spo0J regulates the localization and regulatory activity of Soj by driving Soj from an ATP-bound dimer to an ADP-bound monomer.

Discussion

Molecular basis for the dichotomous regulatory activities of Soj

The B. subtilis ATPase Soj was known to act as either an activator or an inhibitor of the DNA replication initiator protein DnaA under different conditions; however, the molecular basis underlying this switch in regulatory activities had not been established. Here we report that DNA replication initiation is inhibited by monomeric Soj and activated by dimeric Soj, thereby indicating that the key event which switches Soj regulatory activity in vivo is the monomer-to-dimer transition. We go on to show that specific amino acids of the regulatory protein Spo0J required for stimulating the ATPase activity of Soj are required for regulating Soj dimerization, thus accounting for the mechanism by which Spo0J regulates Soj localization and activity.

The results suggest that Spo0J plays a key role in flipping the Soj switch to control DNA replication initiation. We hypothesize that during the cell cycle or in response to an undefined cellular cue the regulation of Soj by Spo0J is overcome. One model is that the concentration of Soj relative to Spo0J increases significantly to shift the equilibrium of Soj protein from a monomer to a dimer. This could occur through increased protein synthesis or through an increase in the local concentration of Soj relative to Spo0J (i.e. near the cell pole). Alternatively, it was recently shown that Spo0J recruits condensin to the origin region of the chromosome to facilitate DNA segregation (Gruber and Errington, 2009; Sullivan et al., 2009). We wonder whether the binding of Condensin occludes Spo0J from interacting with Soj, thereby promoting Soj to activate DNA replication initiation as the Condensin chromosome segregation machinery is being recruited. A further possibility is that the Spo0J nucleoprotein complexes may be displaced during DNA replication by the action of DNA polymerases. The dispersed Spo0J may not be able to effectively interact with and/or regulate Soj, thus allowing a window of time before Spo0J reassembles for Soj to accumulate as a dimer and activate DnaA. This could help to synchronize the initiation of multiple replication origins since DnaA is sequestered from newly activated origins by YabA-dependent tethering to the replication machinery (Soufo et al., 2008), thereby transiently leaving DnaA bound only to origins that have yet to fire.

Soj DNA binding activity is required to activate DNA replication initiation and to interact with Spo0J

We have found that the DNA binding activity of Soj is required for positive regulation of DNA replication initiation, but not for negative regulation. Furthermore, Soj DNA binding activity is required for both the Spo0J interaction and the Spo0J-mediated ATPase stimulation. We imagine that Soj must be properly orientated by DNA so that it sits adjacent to Spo0J in order to efficiently interact. As noted previously (Hester and Lutkenhaus, 2007), the location of the Soj DNA binding surface on the structure T. thermophilus Soj (Leonard et al., 2004) would allow Spo0J to contact Soj dimers that are located either upstream or downstream.

Interestingly, we have found that the SojR189A protein displays the localization pattern and regulatory activity of a monomeric Soj protein, even though this mutant is capable of dimerization either as a purified protein in vitro or when overexpressed in vivo. We propose that Soj dimerization is carefully regulated in B. subtilis through, at the very least, protein expression and protein localization to ensure proper control of Soj activity. We also suspect that additional factors are present in vivo that regulate the monomer-to-dimer transition which were not present in our in vitro and two-hybrid experiments (see below).

Regulation of the monomeric Soj/ParA proteins

Soj is a member of a large class of proteins referred to as ParA ATPases. These proteins were originally identified as stability determinants on low-copy-number plasmids and have now been identified in the majority of sequenced bacterial genomes (Gerdes et al., 2000; Livny et al., 2007). Although all of the ParA orthologues share homology, there are also significant differences between them, both at the sequence level and potentially at the functional level. Based on structural and biochemical studies of plasmid-encoded ParA proteins from P1, P7 and pSM19035, one interesting difference appears to be that these plasmid ParA proteins do not undergo a monomer to dimer switch, but rather they form stable dimers independent of the identity of their bound nucleotide (Pratto et al., 2008; Dunham et al., 2009). We note that because Soj/ParA proteins from the distantly related organisms B. subtilis (Gram-positive) and T. thermophilus (Gram-negative) appear to share the ATP-dependent dimerization switch, this mechanism for Soj/ParA regulation was likely derived from a deeply rooted common ancestor and we suspect that the majority of chromosomally encoded Soj/ParA proteins will retain this activity.

We speculate that the Soj monomer-to-dimer transition may provide the opportunity for regulatory systems to influence Soj activity. One candidate for such a mechanism would be the Soj polar localization determinant(s) which appears to only interact with monomeric Soj (Murray and Errington, 2008). In the current study we observed that the patches of GFP–Soj observed in the Spo0J N-terminal lysine mutants were most often located adjacent to septa (Fig. 6B) and we have previously shown that in filamentous cells GFP–Soj preferentially accumulates near cell poles (Autret and Errington, 2003). These results suggest that factors localized at the cell pole could either spatially or temporarily affect Soj regulation by sequestering monomeric Soj. Although MinD is required for Soj to localize to the cell pole (Murray and Errington, 2008), we have not been able to detect a direct interaction between Soj and MinD using the bacterial two-hybrid assay, nor have we observed a change in the regulation of DNA replication by Soj in a ΔminD mutant (data not shown). It will be highly informative to identify the direct polar localization determinant of Soj and to evaluate its role regulating Soj dimerization.

Experimental procedures

Strains and plasmids

Construction of strains and plasmids used in this study are described in Supporting information and listed in Tables S1 and S2. E. coli strain DH5α (Invitrogen) was used for the construction of all plasmids, and strain BL21 (DE3) pLysS (Stratagene) was used to express all proteins. The plasmid pET21-d (Invitrogen) was used as the expression vector for all proteins, and either pUC18 or pBSoriC4 (Krause et al., 1997) was used for biochemical assays.

Media

Nutrient agar (Oxoid) was used for routine selection and maintenance of both B. subtilis and E. coli strains. For experiments in B. subtilis cells were grown in casein hydrolysate (CH) medium, except for the experiment to calculate the frequency of cells lacking DNA where defined minimal medium was used [Spizizen minimal medium supplemented with glycerol (0.5%), MgSO4 (6 mM), CaCl2 (0.1 mM), MnSO4 (130 µM) and Fe-NH4-citrate (0.01 mg ml−1)]. Supplements were added as required: 20 µg ml−1 tryptophan, 5 µg ml−1 chloramphenicol, 2 µg ml−1 kanamycin, 50 µg ml−1 spectinomycin. For protein expression in E. coli cells were grown in Luria–Bertani (LB) medium or Nutrient Broth (Oxoid) and supplemented with 30 µg ml−1 (for single copy plasmids) or 75 µg ml−1 ampicillin, 10 µg ml−1 chloramphenicol, 50 µg ml−1 kanamycin. For the bacterial two-hybrid analysis cells were grown in M9 salts supplemented with glucose (0.8%), casamino acids (0.4%), MgSO4 (1 mM), CaCl2 (0.1 mM) and thiamine (2 µM).

Microscopy

To visualize cells during exponential growth starter cultures were grown in CH medium overnight, then diluted 1:100 into fresh medium (with inducer where indicated) and allowed to achieve at least three doublings before observation. Cells were mounted on ∼1.2% agar pads (containing 0.5× CH) immobilized within a Gene Frame (ABgene) using a 0.13–0.17 mm glass coverslip (VWR). To visualize nucleoids the DNA was stained with 2 µg ml−1 4′-6-diamidino-2-phenylindole (DAPI) (Sigma). To visualize individual cells the cell membrane was stained with either 2 µg ml−1 Nile Red (Sigma) or 0.4 µg ml−1 FM5-95 (Molecular Probes). Microscopy was performed on an inverted epifluorescence microscope (Zeiss Axiovert 200M) fitted with a Plan-Neofluar objective (Zeiss 100×/1.30 Oil Ph 3). Light was transmitted from a 300 Watt xenon arc-lamp through a liquid light guide (Sutter Instruments) and images were collected using a Sony CoolSnap HQ cooled CCD camera (Roper Scientific). All filters were Modified Magnetron ET Sets from Chroma and details are available upon request. Digital images were acquired and analysed using metamorph software (version V.6.2r6).

Marker frequency analysis

Cells were grown in CH medium at 30°C as described for microscopy. Sodium azide (0.5%; Sigma) was added to exponentially growing cells (A600 = 0.2–0.4) to prevent further growth. Chromosomal DNA was isolated using a DNeasy Blood and Tissue Kit (Qiagen). Power SYBR Green PCR Master Mix was used for PCR reactions (Applied Biosystems). Q-PCR was performed in a LightCycler 480 Instrument (Roche). By use of crossing points and PCR efficiency a relative quantification analysis was performed using Light-Cycler Software version 4.0 (Roche) to determine the ori/ter ratio of each sample. These results were normalized to the ori/ter ratio of a DNA sample from B. subtilis spores in which the ori/ter ratio is 1.

Bacterial two-hybrid assay

Escherichia coli strain BTH101 was transformed using a combination of complimentary plasmids. For quantitative β-galactosidase assays strains were incubated overnight at 30°C in M9 medium containing ampicillin and kanamycin, diluted 1:50 into fresh medium with antibiotics and grown until they reached an A600 of 0.4–0.6. Samples were placed on ice for 20 min to stop growth, after which the culture density was determined. Each culture was mixed with Z-buffer, then BugBuster (Novagen) was added to permeabilize the cells for 20 min at 30°C. β-Galactosidase activity was measured as previously described (Miller, 1972; Gourse et al., 1986). For qualitative β-galactosidase assays single colonies were streaked onto nutrient agar plates containing ampicillin, kanamycin and the indicator X-gal (0.008%). Plates were incubated at 30°C overnight and imaged using a digital camera.

Protein purification

Purification of Soj-His6 and Spo0J-His6 was performed as described below. The molecular weight, theoretical pI and extinction coefficient for Soj-His6 and Spo0J-His6 were calculated using ProtParam, an online tool present on the ExPASy proteomics sever. All affinity chromatography was performed at 4°C at a flow rate of 0.5 ml min−1 (unless otherwise stated) using an AKTA Purifier FPLC system (GE Healthcare).

Purification of Soj-His6

A BL21 (DES) pLysS strain containing the appropriate soj-his6 plasmid (see Table S2) was inoculated into LB containing ampicillin and chloramphenicol and grown overnight at 37°C. The culture was diluted 1:100 into LB containing ampicillin and incubated at 37°C until the A600 reached ∼0.6. Cultures were then supplemented with 1 mM IPTG and shifted to 30°C for 3 h to induce protein expression. Cells were collected by centrifugation at 5000 g (4°C) for 10 min and then resuspended in Soj purification buffer (30 mM Tris pH 8, 300 mM NaCl, 20 mM MgCl2, 40 mM imidazole and 20% glycerol) such that they were concentrated 25-fold. One complete EDTA-free protease inhibitor tablet (Roche) and lysozyme (32 mg ml−1) were added and cells were incubated on ice for 1 h with gentle agitation. Cells were lysed using a Vibra-Cell sonicator (Sonics) on ice for approximately 5 min (power setting 80; 4 s pulse, 1 s rest). Cell debris was removed by centrifugation at 31 000 g (4°C) for 45 min. The supernatant was applied to a 1 ml HisTrap FF column (GE Healthcare), washed with 20 ml of Soj purification buffer and then washed with 20 ml of Soj buffer supplemented with NaCl (200 mM). Bound protein was eluted with 5 ml of Soj purification buffer supplemented with imidazole (500 mM). The eluate was applied to a HiLoad Superdex 75 gel filtration column (GE Healthcare) pre-equilibrated with Soj gel filtration buffer (30 mM Tris pH 7.6, 200 mM potassium glutamate, 1 mM DTT, 1 mM EDTA and 20% glycerol) and run overnight at 0.3 ml min−1. Fractions were then analysed by SDS-PAGE and relevant fractions pooled. The protein solution was concentrated using an Amicon Ultra-15 centrifugal filter spin column (Millipore). All Soj mutants were purified in an identical manner.

Purification of Spo0J-His6

A BL21 (DES) pLysS strain containing the appropriate spo0J-his6 plasmid (see Table S2) was inoculated, grown and induced as described above. Cells were collected by centrifugation at 5000 g (4°C) for 10 min and then resuspended in Spo0J purification buffer (50 mM HEPES pH 7.6, 300 mM NaCl, 10 mM imidazole and 20% glycerol). Cells were lysed as described above. The supernatant was applied to a 1 ml HisTrap FF column (GE Healthcare), washed with 20 ml of Spo0J wash buffer (50 mM HEPES pH 7.6, 300 mM NaCl and 40 mM imidazole) and then eluted with a linear gradient of Spo0J wash buffer supplemented with imidazole (500 mM). Fractions were analysed by SDS-PAGE and relevant fractions were pooled, diluted to 50 ml with Spo0J heparin buffer (20 mM HEPES pH 7.6 and 1 mM EDTA) and applied to a 1 ml HiTrap Heparin column (GE Healthcare). Proteins were eluted using a linear gradient of Spo0J heparin buffer supplemented with NaCl (1.5 M). Fractions were analysed by SDS-PAGE and relevant fractions were pooled and applied to a HiLoad Superdex 75 gel filtration column pre-equilibrated with Spo0J gel filtration buffer (30 mM HEPES pH 7.6, 300 mM NaCl, 1 mM DTT and 10% glycerol) and run overnight at 0.3 ml min−1. Fractions were then analysed by SDS-PAGE and relevant fractions pooled. The protein solution was then concentrated using an Amicon Ultra-15 centrifugal filter spin column.

Size exclusion chromatography

A Superdex 75 HR 10/30 (GE Healthcare) column was pre-equilibrated with Soj gel filtration buffer (50 mM Tris pH 7.6, 150 mM potassium glutamate, 5 mM MgCl2 and 1 mM DTT) supplemented with either 2 mM ATP or ADP. The target protein was pre-incubated with or without nucleotide for 20 min at 37°C. One hundred microlitres of the protein solution (2 mg ml−1) was injected onto the column at a flow rate of 0.4 ml min−1. The A280 was monitored for the entire column volume (24 ml).

Circular dichroism spectroscopy

Fav-UV circular dichroism spectra (190–250 nm) were collected using a JASCO 810 (JASCO, UK) in a 0.2 mm pathlength cuvette. Prior to analysis Soj-His6 was buffer exchanged into 200 mM Sodium Phosphate (pH 7.0) using a PD-10 column (GE Healthcare) and concentrated using an Amicon Ultra-15 centrifugal filter spin column to 0.5 mg ml−1. A buffer only spectra was collected as a reference and subtracted from the sample spectra.

UV cross-linking

[α-32P]-ATP was cross-linked to Soj following exposure to UV light. Soj (5 µM) was added to cross-linking buffer [50 µl; 25 mM HEPES pH 7.6, 50 mM potassium glutamate, 1 mM DTT, 10 mM MgCl2, 0.2 mg ml−1 bovine IgG and 0.5 µM [α-32P]-ATP (400 Ci mmol−1)] in PCR tubes chilled in a metal rack on ice. The reactions were irradiated at 254 nm for 45 min in a Spectrolinker (XL-1500 UV Crosslinker, Spectronics Corp.). The samples were analysed by SDS-PAGE (Invitrogen) and protein bands visualized by staining with Coomassie Brilliant Blue. The gel was exposed to a phosphor storage screen overnight and the 32P-labelled bands visualized using a Typhoon imager (GE Healthcare).

ATPase assay

The malachite green assay solution was created by mixing 0.0812% (w/v) malachite green, 2.23% (w/v) polyvinyl alcohol, 5.72% (w/v) ammonium molybdate in HCl (6 M) and H2O in a ratio of 2:1:1:2. The solution was incubated for 2 h with gentle agitation (during this time the solution turned from a muddy brown colour to a golden yellow). Various combinations of Soj, Spo0J and DNA (all at a final concentration of 2 µM) were mixed and diluted into ATPase buffer (50 mM HEPES pH 7.6, 100 mM potassium glutamate and 10 mM MgCl2) in a final volume of 400 µl. The DNA substrate is a fragment of the spo0J gene containing a parS site and was constructed by annealing complimentary oligonucleotides (5′-CTGAATCAGCAGTTGAATCAGAATGTTCCACGTGAAACAAAGAAAAAAGAACCTG-3′). The protein mixtures were incubated for 5 min at 37°C, followed by addition of ATP (to a final concentration of 2 mM). At various time points 50 µl of the reaction solution was removed and immediately mixed with 800 µl of malachite green assay solution, followed by addition of 100 µl of sodium citrate (34%). The colour was allowed to stabilize for 10 min before the absorbance was detected at 620 nm. A standard curve was created using a serial dilution of sodium phosphate (3–500 µM). The A620 was converted into mole Pi produced per mole of protein using a phosphate standard.

Surface plasmon resonance

Surface plasmon resonance was performed using a NeutrAvidin coated (NLC) sensor chip in a Proteon XPR36 system. A 148 bp PCR product carrying a 5′ Biotin-tag and harbouring a parS site in the middle was amplified from the 3′ region of the spo0J gene. A PCR product carrying a mutated parS site was also created and used as a negative control (Lin and Grossman, 1998). Approximately 500 RU of specific and non-specific DNA substrates were immobilized in separate ligand channels on the NLC chip in running buffer at 25°C (25 mM HEPES pH 7.6, 200 mM NaCl, 10 mM MgCl2, 1 mM DTT and 0.005% Tween-20). Spo0J was exchanged (PD-10 column, GE Healthcare) into running buffer and injected over both channels at a flow rate of 50 µl min−1. Soj was exchanged into running buffer and pre-incubated with 2 mM of the relevant nucleotide for 20 min at 37°C. To investigate the Spo0J–Soj interaction, Soj was injected during the Spo0J dissociation phase at a flow rate of 50 µl min−1. The reported data represent the response observed using the DNA substrate carrying the specific parS site following subtraction of the response observed using the DNA substrate carrying the mutated parS site. Affinity analysis was performed using the Proteon Manager software.

Acknowledgements

We thank the Lakey Lab (Newcastle University) for advice on SPR and circular dichroism analysis, the Lutkenhaus Lab (Kansas University) for providing the sojR189A allele, David Adams for providing purified FtsZ and Ian Selmes for technical assistance. This work was supported by fellowships from the European Molecular Biology Organization, the Human Frontier Science Program and the Royal Society to H.M., and by Grant No. 43/G18654 from the BBSRC to J.E.

Ancillary