Modulation of the P1 Plasmid Partition Protein ParA by ATP, ADP, and P1 ParB*

(Received for publication, December 20, 1996, and in revised form, March 28, 1997)

Megan J. Davey and Barbara E. Funnell Dagger

From the Department of Molecular and Medical Genetics, University of Toronto, Toronto, Ontario M5S 1A8, Canada

ABSTRACT
INTRODUCTION
EXPERIMENTAL PROCEDURES
RESULTS
DISCUSSION
FOOTNOTES
ACKNOWLEDGEMENTS
REFERENCES


ABSTRACT

ParA is an essential P1 plasmid partition protein. It represses transcription of the par genes (parA and parB) and is also required for a second, as yet undefined step in partition. ParA is a ParB-stimulated ATPase that binds to a specific DNA site in the par promoter region. ATP binding and hydrolysis by ParA affect ParA activities in vitro. ATP and ADP binding stimulate ParA DNA binding and dimerization; however, ATP hydrolysis has a negative effect on DNA binding. Our current experiments reveal that ATP binding and hydrolysis affect ParA conformation and ParA sensitivity to ParB. Nucleotide binding assays show that ParA binds ATP better than ADP (Kd values of 33 and 50 µM, respectively). Interaction with these nucleotides as well as ATP hydrolysis by ParA alter ParA conformation as established by CD and ParA sensitivity to heat denaturation. Finally, we show that ParB stimulates ParA DNA binding. This stimulation requires ATP hydrolysis in vitro, suggesting that one role for ATP hydrolysis in vivo is to make ParA repressor sensitive to ParB. Our observations lead to the suggestion that ATP binding and hydrolysis have separable roles in ParA repressor function and perhaps in ParA partition functions as well.


INTRODUCTION

ParA is an essential component of the P1 plasmid partition system (1). The prophage of bacteriophage P1 exists as an autonomously replicating plasmid, and its copy number is about the same as that of the bacterial chromosome (2). The P1 partition system, Par, directs proper segregation and thus stable maintenance of the plasmid. The mechanism of partition is unknown, but can be thought of as a positioning process that via interaction with the Escherichia coli host ensures proper distribution of the plasmid. Par encodes two essential trans-acting proteins, ParA and ParB, and contains a cis-acting DNA site called parS (1). In addition, the E. coli integration host factor, IHF,1 also participates, although it is not absolutely required (3). ParB and IHF bind to parS to form the partition complex (3-5). Assembly of this complex is assumed to be an early step in the partition pathway. Formation of this complex does not require the action of ParA (3-5), and we infer that ParA acts during a later step in the partition process.

ParA has at least two roles in partition. First, ParA represses transcription of its own gene and parB from a promoter upstream of parA (6). ParA repressor activity is stimulated by ParB; however, ParB has no effect on par gene expression on its own. A second role for ParA in partition is inferred from genetic data that show a requirement for ParA in partition even when its regulatory role is bypassed (6, 7). Therefore, we consider ParA's repressor activity its "regulatory" function, and this second, as yet undefined role as ParA's "partition" function because we think it reflects a direct role for ParA in the positioning reaction. The latter function requires that ParA interact, directly or indirectly, with the ParB-IHF-parS partition complex.

ParA is an ATPase and a site-specific DNA-binding protein (8). The ATPase is stimulated by ParB and nonspecific DNA (8). ParA is one of the better characterized proteins in a superfamily of ATPases defined by a modified Walker type A motif in the protein sequence (9-11). This superfamily includes other plasmid partition proteins such as F SopA, as well as plasmid and chromosomally encoded proteins from various bacteria species whose functions have not yet been determined (10-12). Many of the plasmids and bacterial chromosomes also encode a ParB homolog adjacent to the ParA homolog. The similarities of these ParA-like proteins with P1 ParA and F SopA have led to the suggestion that these homologs are also involved in plasmid or chromosome segregation (10, 11).

ParA binds to the par promoter region called parOP (8), and this binding is thought to mediate ParA repressor activity in vivo. The recognition site for ParA is likely a large inverted repeat in the par promoter region, since (i) several mutations in this repeat reduce or eliminate ParA repression in vivo (13) and (ii) in DNase I footprinting experiments, protection centers over the inverted repeat, especially at low ParA concentrations (8, 14). ParA DNA binding activity is strongly stimulated by ATP (8, 14). ATP hydrolysis is not required for this stimulation; in fact, it appears to be inhibitory since ATPgamma S, AMPPNP, and ADP stimulate ParA DNA binding about 5-fold better than ATP (14). In addition, ATP and ADP promote ParA dimerization (14), suggesting that ParA prefers to bind DNA as a dimer.

We are interested in how ParA performs both its regulatory and partition roles. Mutation of the ParA nucleotide binding motif A interferes with both ParA repressor and partition functions in vivo (7). However, the steps at which ATP binding and hydrolysis modulate ParA function are not understood. We have further probed ParA interaction with ATP, ADP, and ParB, by examining their effects on ParA structure and function in vitro, to address these questions. We show that ATP binding and hydrolysis alter ParA conformation and that these conformational changes can be correlated to changes in ParA activity. In addition, we demonstrate that ParB stimulates ParA site-specific DNA binding activity in vitro, consistent with ParB's role as a corepressor in vivo. ParB modulation of ParA DNA binding activity requires ATP hydrolysis. We conclude that ATP binding and hydrolysis have different roles in repression and perhaps partition, presumably mediated by their effects on ParA conformation.


EXPERIMENTAL PROCEDURES

Reagents and Buffers

Sources for reagents were as follows: Mono Q and Mono S columns, GTP, Sephadex G-25 M, Pharmacia Biotech Inc.; bovine serum albumin (FrV), ATP, and ADP, Sigma; ATPgamma S and bovine pancreatic DNase I, Boehringer Mannheim; heparin-agarose, Bio-Gel P-6, dithiothreitol (DTT), and Bradford protein assay dye, Bio-Rad; and [alpha -32P]dATP, [gamma -32P]ATP, and [2,8-3H]ADP, DuPont NEN. Restriction enzymes and E. coli DNA polymerase I large fragment were purchased from New England Biolabs.

Buffer A contained 10 mM Tris acetate, pH 7.5, 30 mM NaCl, and 5 mM MgCl2. Buffer B was 50 mM HEPES, pH 7.5, 0.1 mM Na2EDTA, and 20% glycerol. Dilution buffer was 50 mM Tris-HCl, pH 7.5, 100 mM NaCl, and 10% glycerol.

Proteins

ParA was prepared as described (14) except that the glycerol concentration of FrIV was adjusted to 50% (v/v) and the protein was stored at -20 °C. We consistently maintain ParA in buffers containing 10 mM DTT (14), because we have observed that at lower DTT concentrations ParA oxidizes and forms disulfide cross-linked multimers, particularly after long term storage (data not shown).

ParB (FrIV) was prepared as described (3, 4) except that ParB FrIII was concentrated using a 1-ml Mono S column prior to gel filtration. ParB FrIII was diluted to 50 mM KCl in buffer B containing 2 mM DTT and then loaded on a Mono S column equilibrated in the same buffer. ParB was eluted in one step with 600 mM KCl in buffer B containing 2 mM DTT. Fractions containing ParB were then purified by gel filtration (4).

ParB (FrIV) occasionally contained a contaminating ATPase activity that could potentially interfere with the ParA assays. This ATPase activity was removed by chromatography through a heparin-agarose column. ParB (FrIV) in buffer B containing 50 mM KCl and 2 mM DTT was applied to a 5-ml column equilibrated in the same buffer and eluted with a 100-ml, 50 mM to 1 M KCl gradient in buffer B. ParB (FrV) was stored at -80 °C.

For DNA binding assays, ParB and/or ParA were equilibrated in dilution buffer by gel filtration through either 1-ml Sephadex G-25M or Bio-Gel P-6 columns.

Nucleotide Binding

Hummell and Dreyer equilibrium gel filtration experiments were used to measure ATP and ADP binding by ParA (15, 16). Binding was determined as the amount of ATP and/or ADP that co-eluted with ParA through a gel filtration column. The column was equilibrated in the same buffer and [32P]ATP and/or [3H]ADP concentration as the ParA sample. A 5-ml Bio-Gel P-6 column was equilibrated in buffer A containing [gamma -32P]ATP (10-20 µCi/µmol) and/or [2,8-3H]ADP (10 µCi/µmol). Two hundred microliters of ParA in the same buffer with the same concentration of nucleotide was applied to the column. Four drop fractions were collected (150-180 µl), and the radioactivity in 100 µl of each fraction was measured by liquid scintillation counting. The amount of ParA in the peak fractions was determined by Bradford assay, using a ParA standard whose concentration was determined by amino acid analysis (see "Protein Concentration Determination"). The amount of ATP bound by ParA was determined as the concentration of ATP in the peak fractions minus the background concentration of ATP (in the buffer). Experiments were performed with buffers containing 0.5, 2, 5, 10, 15, 20, 30, or 50 µM ATP using either 31 µM or 62 µM ParA in the sample loaded on the column. At least two experiments were performed at each concentration of ATP. Concentrations used in the ADP binding assays are in Table I.

Table I. Relative binding of ATP and ADP by ParA as measured using equilibrium gel filtration

ParA concentration was 60 µM.

Nucleotide concentrationa
Nucleotide per ParA monomer
ATP ADP ATP:ParA ADP:ParA

µM pmol/pmol
20 20 0.26  ± 0.01 0.16  ± 0.02
15 23 0.16  ± 0.04 0.17  ± 0.04
20  --- 0.34  ± 0.03  ---

a These assays measured binding of [32P]ATP and [3H]ADP (alone or together) to ParA as described under "Experimental Procedures."

Protein Concentration Determination

Protein concentration was routinely measured by Bradford assay (17). The accuracy of this assay for ParA was determined by comparing the measurement of protein concentration by amino acid analysis to the measurement by Bradford assay. For amino acid analysis, ParA was equilibrated to 20 mM NH4HCO3 by gel filtration through a 1-ml Bio-Gel P-6 column. ParA hydrolysis and amino acid analysis were performed at the HSC Biotech Service Center (University of Toronto). ParA molar concentration is expressed as a monomer and ParB as a dimer.

DNA Binding Assay

ParA DNA binding to parOP was measured using DNase I protection assays (14). The substrate was the 388-base pair HindIII-XhoI fragment from pMD6 (14) and was labeled at HindIII on the 3' end with [alpha -32P]dATP using DNA polymerase I large fragment (18). A typical binding assay contained between 0.1 and 30 pmol of ParA (monomer), 10 fmol of pMD6 HindIII-XhoI fragment, 50 nmol of CaCl2, and 1 µg of sonicated salmon sperm DNA in 15 µl of buffer A. Nucleotide, when added, was present at 1 mM. The mixtures were incubated for 10 min at 30 °C, 1.25 µg of bovine pancreatic DNase I was then added, and the mixtures were incubated for an additional 60 s at 30 °C. DNase I digestion was stopped by the addition of 5 µl of "stop" solution (98% formamide, 10 mM Na2EDTA, pH 8, 0.025% xylene cyanol, 0.025% bromphenol blue (18)). The samples were heated to 90 °C for 3 min and loaded on a 6% acrylamide urea gel. After electrophoresis, the gel was dried on Whatman DE81 paper and exposed to a PhosphorImager screen (Molecular Dynamics) and/or to Kodak XRP film.

Quantification of DNase I Protection Assays

DNase I protection assays were quantified using a PhosphorImager and ImageQuant software (Molecular Dynamics). We defined binding as protection from DNase I digestion of the inverted repeat in parOP by ParA. To quantify this binding in each lane, we drew a box in the area that corresponded to the inverted repeat sequences and another box around a set of bands outside of the protected region but in the same lane. The volume ("counts") of the box corresponding to the inverted repeat was divided by the volume of the box outside of the footprint to yield the relative DNase I sensitivity of the inverted repeat. The sensitivity values obtained were normalized so that the assays that did not contain ParA had a DNase I sensitivity of 1. As the sensitivity to DNase I decreases, the protection or binding by ParA increases.


RESULTS

ATP and ADP Binding by ParA

Since ATP affects ParA activity, we decided to directly measure ATP binding by ParA. We used equilibrium gel filtration (15, 16), which measures ATP binding as the amount of ATP that co-elutes with protein over a gel filtration column. Since this is an equilibrium technique, one can deduce ATP stoichiometry as well as binding affinity. ParA was loaded on a Bio-Gel P-6 column that had been equilibrated in the same [32P]ATP concentration as the ParA sample. ATP that bound to ParA co-eluted with the ParA peak, and there was a corresponding trough of ATP concentration after ParA eluted (Fig. 1A) (15). ParA concentration was determined by amino acid analysis (see "Experimental Procedures"). A typical gel filtration experiment is shown in Fig. 1A. The experiment was repeated multiple times, varying ATP concentration from 0.5 to 50 µM and using two different concentrations of ParA (31 or 62 µM). The ratio of ATP bound per monomer of ParA was calculated for each experiment, and each experiment represents one point on a binding curve (Fig. 1B) or a Scatchard curve (Fig. 1C) (19). We were unable to measure ATP binding by ParA to saturation (i.e. at higher nucleotide and protein concentrations) because of limitations of ParA solubility at high concentration. Regression analysis on the Scatchard curve results in a Kd of 33 µM ATP and 0.8 ATP binding sites (n) in a ParA monomer. This Kd is close to the Km for ParA ATPase (50 µM; Ref. 8).


Fig. 1. Equilibrium gel filtration analysis of ATP binding by ParA. A, the profile from a typical equilibrium gel filtration experiment. In this experiment a 200-µl sample containing 50 µM ATP and 62 µM ParA was loaded on a 5-ml Bio-Gel P-6 column and eluted as described under "Experimental Procedures." The ATP concentration in each fraction (bullet ) and the ParA concentration in peak fractions (open circle ) were plotted against the fraction number. Note that the fractions containing protein had a slightly larger volume than the fractions without protein; thus, the peak appears smaller than the trough. B, binding curve of the results from many equilibrium gel filtration experiments performed at 31 µM (bullet ) and 62 µM (black-square) ParA (in the sample load) at various ATP concentrations in the column buffers (see "Experimental Procedures") is shown. All data points (i.e. both data sets) were used to derive the curve. C, the data from B are plotted on a Scatchard curve (19).
[View Larger Version of this Image (17K GIF file)]

Mixing experiments indicate that ParA has a slightly lower affinity for ADP than for ATP. When [gamma -32P]ATP and [2,8-3H]ADP were present at equal concentrations in an equilibrium gel filtration experiment, ParA bound approximately 1.5 times less ADP than ATP (Table I), suggesting that the ParA affinity for ADP is about 1.5 times weaker than ParA affinity for ATP. When the ADP concentration was 1.5 times that of ATP, approximately equal amounts of each nucleotide were bound by ParA (Table I). In the presence of ADP, less ATP was bound by ParA than when the same concentration of ATP alone was used, suggesting that ATP and ADP are binding to the same site. We estimated a Kd of 50 µM for ADP, approximately 1.5 times lower than the Kd for ATP (33 µM). The different affinities of ParA for ATP and ADP indicate that ParA can distinguish between adenine nucleoside diphosphates and triphosphates.

Nucleotide Effects on ParA Stability

Adenine nucleoside di- and triphosphates strongly stimulate ParA DNA binding (8, 14); however, ADP and ATPgamma S are better cofactors than ATP (14). To test whether the ability of adenine nucleotides to modulate ParA activity was a consequence of their effects on ParA conformation, we examined ParA stability with and without nucleotide. We found that although ParA was quite stable at 0 °C (for at least 4 days), it was rapidly inactivated by short heat treatments at 52 °C (Fig. 2), and we used this treatment to detect nucleotide-dependent changes in ParA conformation.


Fig. 2.

ATP, ATPgamma S, and ADP stabilize ParA. A, activity of heat-treated ParA, assessed by DNase I footprinting at the parOP site. ParA was heated for 3 min at 52 °C in buffer A containing 100 mM NaCl, 10% (v/v) glycerol, 100 ng of bovine serum albumin/µl, and 1 mM of the indicated nucleotide (ATP, ATPgamma S, ADP, or GTP; above each set of lanes). Following the heat treatment, ParA was diluted and titrated into DNA binding assays containing 1 mM ATP at 30 °C, as described under "Experimental Procedures." Each lane is labeled with the amount of ParA in the DNA binding mixture. Note that this ParA scale varies; ParA concentration was lower in assays containing ATP or ATPgamma S than ones containing ADP or GTP. The labels and boxes on the left indicate the -10 and -35 transcriptional signals, the ribosome binding site (RBS), and the parA start codon (ATG). The imperfect inverted repeat is delineated by the inverted arrows. B, ParA DNA binding activity was quantified from the data in A as described under "Experimental Procedures." The relative sensitivity to DNase I cleavage of the inverted repeat sequences in each lane was plotted against the amount of ParA (pmol), heated in buffer containing 1 mM ATP (bullet ), 1 mM ATPgamma S (open circle ), 1 mM ADP (black-square), or 1 mM GTP (square ). C, DNase I footprints of ParA following a heat treatment with (+) or without (-) 1 mM GTP. ParA was heated for 1 min at 52 °C in buffer A containing 100 mM NaCl, 100 ng of bovine serum albumin/µl, 10% (v/v) glycerol with or without 1 mM GTP before measuring DNA binding. ParA DNA binding was assayed in the presence of 1 mM ATP at 30 °C as described under "Experimental Procedures." The heating step was shorter in this experiment than in A because ParA is less stable in these conditions. The positions of the DNA sequence elements are represented as described in A.


[View Larger Version of this Image (52K GIF file)]

ParA stability at 52 °C was measured in the presence of various nucleotides (Fig. 2). When ParA was heated to 52 °C for 3 min in buffer containing 1 mM ATP or ATPgamma S, the protein retained significantly more DNA binding activity than when it was heated in buffer with 1 mM GTP (Fig. 2, A and B). ADP (1 mM) also stabilized ParA, although not to the same extent as ATP and ATPgamma S stabilized ParA (Fig. 2, A and B). After this heat treatment, ParA with GTP was essentially inactivated (Fig. 2, A and B). After a shorter heat treatment (1 min), ParA with GTP retained the same activity as ParA that was heated without any nucleotide (Fig. 2C), consistent with other assays that indicate that ParA does not bind GTP (8, 14). Therefore, ATP, ADP, and ATPgamma S stabilize ParA, suggesting that ParA interaction with these nucleotides alters ParA conformation. However, ADP is less effective than ATP or ATPgamma S.

In these experiments, ParA was heated with different nucleotides and then assayed for DNA binding activity in 1 mM ATP. Although different adenine nucleotides stimulate ParA DNA binding to different extents (14), the differences in DNA binding activity observed here were not due to carryover of nucleotide from the heating step into the DNA binding step for the following reasons. First, the nucleotide specificity for stability (Fig. 2) differed from the nucleotide specificity for DNA binding (14). Second, when ParA was incubated at 0 °C rather than 52 °C in buffer containing 1 mM ATP, ADP, or ATPgamma S and then assayed for DNA binding activity in 1 mM ATP, there was no difference in activity compared with each other or to untreated ParA (data not shown). Finally, when ATP was added to ParA after, rather than before, incubation at 52 °C, there was no increase in DNA binding activity compared with no nucleotide. Therefore, we have measured the effect of nucleotide on protein stability, not on DNA binding activity, in these assays (Fig. 2).

ATPgamma S and ATP stabilized ParA to the same extent in this assay, suggesting that ATP hydrolysis was not required for maximum ParA stability under these conditions. Because some ATPases can hydrolyze ATPgamma S, we tested whether ParA hydrolyzes ATPgamma S at 52 °C. We did not detect any ParA-dependent hydrolysis of ATPgamma S at 52 °C under conditions where we could detect 1% or more hydrolysis (data not shown). Under the same conditions, 2-5% of ATP (at 1 mM) was hydrolyzed, which is approximately 2-5 fold more than the rate of hydrolysis at 30 °C (data not shown). The similarity between ATP and ATPgamma S was also not due to ATP contamination of ATPgamma S. By Mono Q chromatography, no ATP was detected in the ATPgamma S sample (as little as 1% could be detected; data not shown).

These experiments show that adenine nucleotides stabilize ParA; however, the triphosphate-bound forms are more stable than the diphosphate-bound form. This difference may reflect the difference in affinity for adenine nucleoside diphosphates and triphosphates, but may also reflect a difference in conformation. These observations suggest that ParA interaction with adenine nucleotides alter ParA conformation, resulting in increased stability.

ATP Binding and Hydrolysis Alter ParA Conformation

We tested the conclusion that adenine nucleotides alter ParA conformation using circular dichroism (CD). Addition of 1 mM ATP elicited a decrease in ParA molar ellipticity (theta ) indicating that ATP alters ParA conformation (Fig. 3). ADP or ATPgamma S (1 mM) each had a smaller, but measurable effect in this assay (Fig. 3). AMP, which does not stimulate ParA DNA binding activity (14), had no effect on ParA conformation (Fig. 3). The high absorbance of light in the samples containing nucleotide prevented accurate measurement of ellipticity below 219 nm. In this assay we used AMP rather than GTP as a negative control so that the absorption spectra of all nucleotides in the buffers were identical. The decrease in molar ellipticity at 222 nm indicated an increase in helicity upon ParA interaction with nucleotides (20) (Table II). This increase in helicity corresponded to approximately 16 peptide bonds changing conformation in a ParA monomer (398 amino acids) with ATP and approximately 8 peptide bonds changing conformation with ADP or ATPgamma S. ParA, therefore, assumes three different conformations as detected by CD: ParA without nucleotide, ParA bound to nonhydrolyzable nucleotide (ADP or ATPgamma S), and ParA bound to hydrolyzable nucleotide (ATP). We conclude therefore that ParA interaction with nucleotide alters ParA structure and that hydrolysis further alters that structure.


Fig. 3. ParA conformation with and without nucleotide. A, CD spectra of ParA (20 µM) in buffer A containing 100 mM NaCl without nucleotide (bullet ) or with 1 mM AMP (open circle ), 1 mM ADP (black-triangle), 1 mM ATPgamma S (triangle ), or 1 mM ATP (black-square) are shown. Spectra were measured on an Aviv 62A DS circular dichroism spectrometer. Measurements were read from 300 to 219 nm in 1-nm intervals with a 1-s averaging time. The spectrum of a buffer blank with or without 1 mM nucleotide was subtracted from the ParA spectrum with or without the corresponding nucleotide. Each experiment is the average of five scans, and each experiment was performed at least three times. B, the data between 219 and 225 nm from A replotted on an expanded scale.
[View Larger Version of this Image (22K GIF file)]

Table II. Effects of adenine nucleotides on ParA helicity, measured by CD


Nucleotide Helicitya

%
None 42
AMP 42
ADP 44
ATPgamma S 44
ATP 46

a Calculated from molar ellipticity at 222 nm as described by Chen et al. (20). The data are averages from at least three different experiments.

We also used circular dichroism to examine the effects of adenine nucleotides on ParA stability and confirmed our initial results (Fig. 2). We measured temperature-dependent changes in ParA conformation by monitoring ParA molar ellipticity at 220 nm between 25 °C and 65 °C (Fig. 4). Since ParA precipitated when it was heated, denaturation was not reversible and we cannot make any conclusions about the thermodynamics of ParA conformation in response to heat, such as the free energy change. However, we can estimate the temperatures at which ParA denatured with and without nucleotide (Table III). As in the initial stability assay, adenine nucleotides increased ParA stability (Fig. 4, Table III). However, in this assay ATPgamma S did not stabilize ParA as well as ATP stabilized ParA (Table III). This result suggests that ATP hydrolysis modifies ParA conformation, as was also suggested by the ATP hydrolysis-dependent increase of ParA helical content (Fig. 3). However, the ADP-bound form of ParA was still less stable than both adenine nucleoside triphosphate-bound forms.


Fig. 4. The effects of adenine nucleotides on ParA stability. ParA conformation was monitored by CD at 220 nm (averaging time 15 s) from 25 °C to 65 °C. The temperature was increased in 2 °C increments, and the sample was equilibrated to each temperature for 1 min before measurement of the signal. ParA (20 µM) was in buffer A containing 100 mM NaCl and 1 mM ATP (black-triangle), 1 mM ADP (black-square), or no nucleotide (bullet ). The experiment shown is one of three experiments performed.
[View Larger Version of this Image (19K GIF file)]

Table III. Effect of adenine nucleotides on heat denaturation of ParA


Nucleotidea Denaturation temperatureb

None 49.4  ± 0.1
ADP 52.6  ± 0.4
ATPgamma S 55.9  ± 0.8
ATP 57.7  ± 0.5

a Concentration of nucleotide, when present, was 1 mM.
b Temperature at which signal at 220 nm had changed by 50% of total change. The data are the averages from three experiments.

ParB Stimulates ParA DNA Binding

ParB stimulates ParA repressor activity in vivo; however, ParB has no repressor activity on its own (6). Simplistically one might expect that ParB stimulates ParA repressor activity in vivo via a direct interaction of ParB with ParA at parOP. We predicted that such an interaction would influence ParA DNA binding to the operator, either by increasing ParA affinity for parOP or by altering ParA activity, making it a more effective block to RNA polymerase (resulting, for example, in a change in the DNase I footprint). We show here that a large molar excess (10-fold as ParB dimers to ParA monomers) of ParB over ParA stimulated ParA DNA binding activity up to 5-fold in the presence of ATP (Fig. 5A). ParB did not protect parOP from DNase I cleavage on its own (Fig. 5B), nor did ParB alter the pattern of DNase I cleavage in the presence of ParA (e.g. compare the last lane (1.8 pmol of ParA) without ParB to the fifth lane (.37 pmol of ParA) with ParB in Fig. 5A). This implies that ParB does not directly contact the promoter and that ParB probably acts by increasing ParA affinity for parOP in the presence of ATP. At low ParB to ParA ratios, ParB stimulation of ParA DNA binding was only observed when ParA and ParB were preincubated together before assaying for DNA binding. Less stimulation of ParA DNA binding by ParB was observed under these conditions (about 2-fold; Fig. 5B).


Fig. 5. ParB stimulates ParA DNA binding in the presence of ATP. A, DNase I footprints of ParA binding to parOP in the absence of ParB (left) or in the presence of a 10-fold molar excess of ParB (right) were performed as described under "Experimental Procedures." Each assay contained the indicated amount of ParA and 1 mM ATP. The amount of ParB, when present, was 10 pmol of ParB (dimer) for every 1 pmol of ParA (monomer). In the absence of ParB, the same molar concentration of bovine serum albumin in ParB buffer was added to the assays. The parA start codon (ATG), the ribosome binding site (RBS), and the -10 and -35 transcriptional signals are indicated by the boxes to the left. The position of the inverted repeat sequence is indicated by the inverted arrows. B, DNA binding was measured in the presence or absence of ParA and a 2-fold molar excess of ParB (ParB dimers to ParA monomers). The proteins, together or separately, were preincubated on ice in dilution buffer for 1 h before assaying for DNA binding at 30 °C. DNase I footprints were performed in the presence of 1 mM ATP as described under "Experimental Procedures." The positions of the DNA sequence elements are labeled as in A.
[View Larger Version of this Image (44K GIF file)]

In previous experiments (8, 14) no ParB effect on ParA DNA binding was detected. We (14) had used the lower levels of ParB with no preincubation step. Under these conditions, we saw identical results in our current study: Without preincubation, lower ParB levels did not stimulate ParA DNA binding (data not shown). Since ParB stimulated ParA DNA binding at lower ParB to ParA ratios after preincubation, we think it reasonable that the initial requirement for a large excess of ParB over ParA is the result of a weak ParB-ParA interaction rather than a high stoichiometric ratio.

Nucleotide Specificity of ParB Stimulation of ParA DNA Binding Activity

ParB stimulated ParA DNA binding activity only in the presence of ATP. We did not detect any stimulation of ParA DNA binding by ParB in the presence of ADP (Fig. 6A) or in the absence of nucleotide (Fig. 6B). In the absence of nucleotide, ParA binds parOP weakly, about 10-fold less well than ParA-ATP binds parOP (14). Because a large amount of protein was required in these assays and because nonspecific DNA binding interfered, we could not add a 10-fold molar excess of ParB to ParA DNA binding assays without nucleotide. Instead, less ParB was used and the two proteins were preincubated on ice before assaying for ParA DNA binding activity (as in Fig. 5B). Under these conditions, ParB did not affect ParA DNA binding activity (Fig. 6B). In the presence of ADP or ATPgamma S (but without ParB), ParA binds to parOP very well, about 5-fold better than in the presence of ATP (14). When a 10-fold molar excess of ParB was added to ParA DNA binding assays in the presence of 1 mM ADP, no effect of ParB was detected (Fig. 6A). ParB also did not stimulate ParA DNA binding in the presence of ATPgamma S (data not shown).


Fig. 6. ParB does not stimulate ParA DNA binding in the presence of ADP or in the absence of nucleotide. A, DNase I footprints of ParA binding to parOP in 1 mM ADP with (+) or without (-) a 10-fold molar excess of ParB (dimer) are shown. Assays without ParB contained a 10-fold molar excess of bovine serum albumin in dilution buffer. The positions of the ParA start codon (ATG), the ribosome binding site (RBS), and the -10 and -35 transcriptional signals are indicated by the boxes on the left. The inverted arrows mark the position of the inverted repeat sequences. B, ParA binding to parOP in the absence of nucleotide, with and without ParB, was measured using DNase I footprints as described under "Experimental Procedures." ParA was mixed in dilution buffer with a 2-fold molar excess of either bovine serum albumin (-) or ParB (+) and preincubated on ice for 1 h before assaying for ParA DNA binding activity at 30 °C. The positions of the various DNA sequence elements are marked as described in A.
[View Larger Version of this Image (89K GIF file)]

Since ParA-ADP binds DNA better than ParA-ATP binds DNA (14) and ParB stimulates ParA ATPase (8), one explanation of our current results is that ParB stimulates ParA DNA binding via ParB stimulation of ParA ATPase and hence production of ADP. This seemed a remote possibility since approximately 0.1 mM ADP is required to stimulate ParA DNA binding (14), ParA has a lower affinity for ADP than for ATP (Table I), and previous measurements of ParA ATPase with a large molar excess of ParB (although at a lower ATP concentration than used in our assays) indicate that insufficient ADP was produced to stimulate ParA DNA binding (8). Nevertheless, we measured ATP hydrolysis by ParA under DNA binding conditions with a 10-fold molar excess of ParB. In agreement with previous measurements of ParA ATPase (8), insufficient ADP was produced (final concentration, less than 1 µM; data not shown) to account for ParB stimulation of ParA DNA binding activity.

We suggest therefore that ParB stimulates ParA DNA binding via a direct interaction with ParA. Interestingly, ParB stimulated ParA DNA binding (with ATP) to the same extent that ADP does over ATP without ParB (up to 5-fold; Ref. 14) (Fig. 5A). Since ParB stimulates ParA ATPase (8) and ATP hydrolysis by ParA is inhibitory to ParA DNA binding (14), we propose that ParB stimulates ParA DNA binding by circumventing the negative action of ATP hydrolysis on ParA DNA binding activity.


DISCUSSION

P1 ParA binds (Fig. 1) and hydrolyzes ATP (8), and one or both of these activities are essential for both its regulation and partition activities (7). Nucleotide binding experiments (Fig. 1) suggest one ATP binding site per ParA monomer, which agrees well with the presence of one putative ATP binding site defined by sequence analysis (10), and the observation that deletion of this region of the protein destroys ATPase activity (7). We have used different adenine nucleotide cofactors to probe the structure and function of ParA. Our results suggest that ATP binding and ATP hydrolysis by ParA induce different conformations of ParA that in turn have different roles in ParA activity. In addition, we suggest that ATP binding and hydrolysis play separable roles in ParA activities in vitro and consequently in partition and par gene expression in vivo.

ParA Conformation with and without Nucleotide

Upon interaction with ATP, ADP, and ATPgamma S, ParA changes conformation (Fig. 3) and ParA dimerization and DNA binding are stimulated (8, 14). The nucleotide specificities of the changes in ParA helical content detected by CD and of the stimulation of ParA DNA binding are the same, implying that the effects of ATP binding and hydrolysis on ParA DNA binding activity are mediated by their effects on ParA conformation. However, the nucleotide specificities for dimerization and helical content are different, suggesting that dimerization is not the only factor determining the change in ParA conformation upon interaction with nucleotides. The concentration of intracellular nucleotide pools is sufficiently high (e.g. ATP is in the millimolar range; Ref. 21) that ParA would rarely be free of nucleotide in vivo. However, differences between ParA-ATP and ParA-ADP or the perhaps the interchange between these conformations are likely to be important for ParA function in partition.

Differentiation between Adenine Nucleoside Diphosphates and Triphosphates

We found it intriguing that ADP and ATPgamma S behaved similarly in our assays, such as site-specific DNA binding (14) and conformation (Fig. 3). This raised the formal possibility that ParA did not discriminate between nucleoside di- and triphosphates, except via the act of hydrolysis. Here, measurements of nucleotide binding indicate a slightly better affinity for ATP than for ADP, and a large difference in the ability of adenine nucleoside tri- and diphosphates to stabilize ParA. The differences in stability may be a function of the differences in affinity for nucleotide, or due to differences in ParA conformation that we cannot detect by CD. Differences in ParA stability mediated by nucleotide could also potentially play roles in partition or regulation.

Effects of ATP Hydrolysis on ParA Structure and Function

ParA consistently behaves differently when bound to a hydrolyzable nucleotide than when bound to a nonhydrolyzable nucleotide. The ATP-bound form of ParA has a lower affinity for DNA (14), it has a different conformation than the other forms (Figs. 3 and 4), and it is the only form detectably modulated by ParB (Figs. 5 and 6). Our current results provide further support that ATP hydrolysis plays roles in ParA function that are distinct and separable from ATP binding.

ParA-ParB Interactions

ParB stimulates ParA binding to parOP, but only in the presence of ATP. ParB did not stimulate ParA DNA binding in the absence of nucleotide (Fig. 6B), a form that binds parOP weakly (14). We detected no ParB effect on ParA-ADP or ParA-ATPgamma S DNA binding (Fig. 6A; data not shown), forms that bind to DNA about 5-fold better than ParA-ATP binds to DNA (14). ParB stimulates ParA-ATP DNA binding activity to the same level as ParA-ADP DNA binding activity with or without ParB. Since ATP hydrolysis by ParA in the absence of ParB is inhibitory to DNA binding (14), we think that the simplest explanation is that association with ParB prevents the inhibitory action of ATP hydrolysis on ParA DNA binding. This model also explains the observation that ParB can stimulate both ParA ATPase and ParA DNA binding, since one might otherwise expect increased ATP hydrolysis by ParA to further interfere with ParA DNA binding. Mutations have been isolated in other ATPases that can bind but not hydrolyze ATP (22, 23). Our hypothesis predicts that a similar ParA mutant would repress transcription from parOP as well as wild-type ParA represses in the presence of ParB (6), and that this repression would be independent of ParB.

Why is ATP hydrolysis required for ParB stimulation of ParA DNA binding? One possible explanation is that ATP hydrolysis is required for physical association of ParA and ParB. Preliminary co-immunoprecipitation experiments suggest that ParA and ParB associate in the absence of ATP,2 so we favor other interpretations of these results. Perhaps a ParA-ParB complex forms regardless of the nucleotide present but is detectable by our DNA binding assay only in the presence of ATP hydrolysis. ParA has no detectable kinase activity (8),2 so we think it unlikely that phosphorylation of either protein by ParA is required for activity.

Roles of ATP in par Gene Regulation and Partition

The ATP site in ParA is essential for both ParA's regulation and partition activities in vivo (7). One role of ATP binding is to promote ParA binding to parOP for ParA repressor activity (8, 14). A role for ATP hydrolysis in repression is suggested by its ability to make ParA DNA binding, and thus ParA repressor, sensitive to ParB levels. Alternatively, one might argue that the role of ParB in repression is to make ParA DNA binding insensitive to ATP hydrolysis. We do not yet know what other roles ATP binding and hydrolysis have for ParA function in vivo. It is possible, for example that ATP hydrolysis is necessary only for partition and not regulation. Intuitively, we think it likely that ATP hydrolysis is required for the positioning reaction, in particular to fulfil the energy requirement inherent in the partition process. For example, conformational changes induced by ATP binding and hydrolysis in some proteins are thought to mediate the displacement of other structures in the cell, such as transport across membranes (MalK; Ref. 24), passage of one DNA strand through another (topoisomerase II ; Ref. 25), or movement along microtubules (kinesin; Ref. 26). Perhaps the conformational changes induced in ParA by ATP binding and hydrolysis result in the movement or orientation of the partition apparatus.

ParA-ParB interactions are likely important for ParA's activity in the positioning reaction. Host factors might interact with ParA, ParB, or both proteins. These host factors are still undefined, although they are not DNA sites on the host chromosome (27). We assume that they are proteins or membrane components that position partition complexes and/or time positioning events with respect to the cell division cycle. A careful analysis of ParA-ParB interactions at parS, including the roles of ATP binding and hydrolysis in these interactions, as well as identification of host factors, are crucial to the dissection of the P1 plasmid partition pathway.


FOOTNOTES

*   This work was supported by an Ontario Graduate Scholarship (to M. J. D.) and a grant from the Medical Research Council of Canada (to B. E. F.).The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Dagger    To whom correspondence should be addressed. Tel.: 416-978-1665; Fax: 416-978-6885; E-mail: b.funnell{at}utoronto.ca.
1   The abbreviations used are: IHF, integration host factor; parOP, the P1 par operon operator/promoter sequences; ATPgamma S, adenosine 5'-O-(3-thiotriphosphate); AMPPNP, 5'-adenylyl beta ,gamma -imidodiphosphate; DTT, 1,4-dithiothreitol.
2   M. J. Davey and B. E. Funnell, unpublished results.

ACKNOWLEDGEMENTS

We are very grateful to Alan Davidson (A. D.) for use of his CD spectrometer and his and Ariel Di Nardo's assistance in performing the CD experiments. In addition, we thank A. D. for very helpful discussions and advice, and A. D. and Marc Perry for critical reading of the manuscript.


REFERENCES

  1. Abeles, A. L., Friedman, S. A., and Austin, S. J. (1985) J. Mol. Biol. 185, 261-272 [Medline] [Order article via Infotrieve]
  2. Ikeda, H., and Tomizawa, J.-I. (1968) Cold Spring Harbor Symp. Quant. Biol. 33, 791-798 [Medline] [Order article via Infotrieve]
  3. Funnell, B. E. (1988) Proc. Natl. Acad. Sci. U. S. A. 85, 6657-6661 [Abstract]
  4. Funnell, B. E. (1991) J. Biol. Chem. 266, 14328-14337 [Abstract/Free Full Text]
  5. Davis, M. A., and Austin, S. J. (1988) EMBO J. 7, 1881-1888 [Abstract]
  6. Friedman, S. A., and Austin, S. J. (1988) Plasmid 19, 103-112 [Medline] [Order article via Infotrieve]
  7. Davis, M. A., Radnedge, L., Martin, K. A., Hayes, F., Youngren, B., and Austin, S. J. (1996) Mol. Microbiol. 21, 1029-1036 [Medline] [Order article via Infotrieve]
  8. Davis, M. A., Martin, K. A., and Austin, S. J. (1992) Mol. Microbiol. 6, 1141-1147 [Medline] [Order article via Infotrieve]
  9. Walker, J. E., Saraste, M., Runswick, M. J., and Gay, N. J. (1982) EMBO J. 1, 945-951 [Medline] [Order article via Infotrieve]
  10. Motallebi-Veshareh, M., Rouch, D. A., and Thomas, C. M. (1990) Mol. Microbiol. 4, 1455-1463 [Medline] [Order article via Infotrieve]
  11. Koonin, E. (1993) J. Mol. Biol. 229, 1165-1174 [CrossRef][Medline] [Order article via Infotrieve]
  12. Funnell, B. E. (1996) The Role of the Bacterial Membrane in Chromosome Replication and Partition, pp. 123-148, R. G. Landes, Austin, TX
  13. Hayes, F., Radnedge, L., Davis, M. A., and Austin, S. J. (1994) Mol. Microbiol. 11, 249-260 [Medline] [Order article via Infotrieve]
  14. Davey, M. J., and Funnell, B. E. (1994) J. Biol. Chem. 269, 29908-29913 [Abstract/Free Full Text]
  15. Hummell, J. P., and Dreyer, W. J. (1962) Biochim. Biophys. Acta 63, 530-532 [CrossRef]
  16. Myles, G. M., Hearst, J. E., and Sancar, A. (1991) Biochemistry 30, 3824-3834 [Medline] [Order article via Infotrieve]
  17. Bradford, M. M. (1976) Anal. Biochem. 72, 248-254 [CrossRef][Medline] [Order article via Infotrieve]
  18. Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) Molecular Cloning: A Laboratory Manual, 2nd Ed., Cold Spring Harbor Laboratory, Cold Spring Harbor, NY
  19. Scatchard, G. (1949) Ann. N. Y. Acad. Sci. 51, 660-672
  20. Chen, Y. H., Yang, J. T., and Martinez, H. M. (1972) Biochemistry 11, 4120-4131 [Medline] [Order article via Infotrieve]
  21. Franzen, J. S., and Binkley, S. B. (1961) J. Biol. Chem. 236, 515-519 [Medline] [Order article via Infotrieve]
  22. Saraste, M., Sibbald, P. R., and Wittinghofer, A. (1990) Trends Biochem. Sci. 15, 430-434 [CrossRef][Medline] [Order article via Infotrieve]
  23. van der Wolk, J., Klose, M., Breukink, E., Demel, R. A., de Kruijff, B., Freudl, R., and Driessen, A. J. M. (1993) Mol. Microbiol. 8, 31-42 [Medline] [Order article via Infotrieve]
  24. Schneider, E., Wilken, S., and Schmid, R. (1994) J. Biol. Chem. 269, 20456-20461 [Abstract/Free Full Text]
  25. Berger, J. M., Gamblin, S. J., Harrison, S. C., and Wang, J. C. (1996) Nature 379, 225-232 [CrossRef][Medline] [Order article via Infotrieve]
  26. Hirose, K., Lockhart, A., Cross, R. A., and Amos, L. A. (1995) Nature 376, 277-279 [CrossRef][Medline] [Order article via Infotrieve]
  27. Funnell, B. E., and Gagnier, L. (1995) J. Bacteriol. 177, 2381-2386 [Abstract]

©1997 by The American Society for Biochemistry and Molecular Biology, Inc.