Hydrolysis of ATP at Only One GyrB Subunit Is Sufficient to Promote Supercoiling by DNA Gyrase*

Mutation of Glu42 to Ala in the B subunit of DNA gyrase abolishes ATP hydrolysis but not nucleotide binding. Gyrase complexes that contain one wild-type and one Ala42 mutant B protein were formed, and the ability of such complexes to hydrolyze ATP was investigated. We found that ATP hydrolysis was able to proceed independently only in the wild-type subunit, albeit at a lower rate. With only one ATP molecule hydrolyzed at a time, gyrase could still perform supercoiling, but the limit of this reaction was lower than that observed when both subunits can hydrolyze the nucleotide.

Mutation of Glu 42 to Ala in the B subunit of DNA gyrase abolishes ATP hydrolysis but not nucleotide binding. Gyrase complexes that contain one wild-type and one Ala 42 mutant B protein were formed, and the ability of such complexes to hydrolyze ATP was investigated. We found that ATP hydrolysis was able to proceed independently only in the wild-type subunit, albeit at a lower rate. With only one ATP molecule hydrolyzed at a time, gyrase could still perform supercoiling, but the limit of this reaction was lower than that observed when both subunits can hydrolyze the nucleotide.
The three-dimensional structure of DNA plays a key role in many biological processes. Reactions such as replication, transcription, or recombination not only are regulated by but also have a profound effect on the topology of the DNA molecule. The enzymes responsible for maintaining the topological state of DNA are DNA topoisomerases. One such enzyme is DNA gyrase, a bacterial topoisomerase that introduces negative supercoils into DNA in a reaction coupled to ATP hydrolysis. The action of gyrase involves the creation of a double-stranded break in one DNA segment and the passage of another segment through this enzyme-stabilized DNA gate.
The ability of gyrase to negatively supercoil DNA is unique among topoisomerases and is based on its mode of DNA binding (1,2). Gyrase wraps DNA in a right-handed manner (1), resulting in the positioning of two segments of DNA in the right orientation for supercoiling. Binding of ATP closes a protein clamp that traps the DNA segment to be transported. The nucleotide is then hydrolyzed, and the free energy is coupled to the supercoiling reaction. After hydrolysis, the enzyme is reset for another round of supercoiling. The limit of supercoiling is believed to be thermodynamic rather than steric, because gyrase can supercoil very small DNA circles, whereas a nucleotide analog (ATP␣S) 1 with higher free energy of hydrolysis than ATP is capable of taking the limit of the supercoiling reaction to higher negative superhelical density (3,4).
Gyrase is a heterotetramer in which two A (GyrA, 97 kDa) and two B (GyrB, 90 kDa) subunits constitute an A 2 B 2 complex (5). There is one ATP-binding site per GyrB, which is situated in the 43-kDa N-terminal domain of the protein. The structure of this domain complexed with the ATP analog 5Ј-adenylyl-␤,␥-imidodiphosphate (ADPNP) was solved by x-ray crystallography and was found to be a dimer (6). Study of the ATPase reaction of this domain revealed that dimerization is an essential step for ATP hydrolysis (7). It is very likely that dimerization of the 43-kDa domain also occurs in the ATPase reaction of intact gyrase (8). The rate of ATP hydrolysis by gyrase is stimulated by the presence of DNA (9), and the kinetics of hydrolysis show positive cooperativity between the two ATPbinding sites (10 -12).
A number of issues concerning the mechanism of ATP hydrolysis are still unclear. These include the mechanism of cooperativity between the two sites and the coupling of the free energy produced by these two reactions to a strand passage event. Directly related to these issues is the question of whether ATP hydrolysis can take place only in one of the two sites and the capacity of such a reaction to support supercoiling. At the molecular level the mechanism of hydrolysis of ATP by gyrase involves nucleophilic attack by water on the ␥-phosphate of ATP with Glu 42 of GyrB acting as a general base (13). Mutation of Glu 42 to Ala in GyrB abolishes ATP hydrolysis but not nucleotide binding (13). We formed heterogeneous gyrase tetramers containing one wild-type and one mutant GyrB Ala42 subunit and used these complexes to address the above questions.

EXPERIMENTAL PROCEDURES
Enzymes, DNA, and Assays-Wild-type GyrA and GyrB proteins were prepared as described previously (14). Mutant GyrB Ala42 or Gyr-B Gln42 proteins were a gift of Dr. A. P. Jackson and were made as described previously (13). Relaxed and supercoiled forms of pBR322 DNA were gifts of A. J. Howells (Leicester University), whereas linear pBR322 was made by digestion of the supercoiled form by EcoRI. Limited proteolysis experiments were performed as described previously (8). ATPase assays were performed as described previously (7,15), in 35 mM Tris⅐HCl (pH 7.5), 24 mM KCl, 4 mM MgCl 2 , 5 mM dithiothreitol, 6.5% w/v glycerol, 2 mM ATP at 25°C. The extent of supercoiling was determined by removing samples from the ATPase reactions and analyzing them by agarose gel electrophoresis.
Formation of Heterogeneous Tetramers-Heterogeneous tetramers were formed by mixing the indicated amounts of wild-type and mutant GyrB proteins and allowing the mixture to equilibrate at 25°C for 1 h. GyrA was then added in excess of the total B protein, and the gyrase complexes were allowed to form for 1 h at 25°C. After DNA was added (where indicated), reactions were incubated at 25°C for a further 30 min. 2 mM ATP was then added to initiate the reactions.

ATP Traps A 2 B 2
Ala42 in the Dimerized Form (Complex II)-Limited proteolysis has been used as a sensitive probe for detecting conformational changes in DNA topoisomerases (8,16). Treatment of gyrase with trypsin produces two major fragments, one of ϳ62 kDa, derived from the A protein, and another of ϳ25 kDa from the B protein (Fig. 1); this proteolytic fingerprint has been termed complex I (8). Binding of ADPNP induces dimerization of the B subunits resulting in a conformation that protects the 43-kDa domains of GyrB from tryptic digestion. The characteristic fingerprint of this conformation (termed complex II) consists, apart from the two above men-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ Supported by a BBSRC-CASE studentship funded by the BBSRC and Glaxo-Wellcome and a grant from the Alexander S. Onassis Public Benefit Foundation.
§ Lister-Institute Jenner Fellow. To whom correspondence should be addressed. Tel.: 44-116-2523464; Fax: 44-116-2523369; E-mail: ony@ le.ac.uk. 1 The abbreviations used are: ATP␣S, adenosine 5Ј-O-(1-thiotriphosphate); ADPNP, 5Ј-adenylyl-␤,␥-imidodiphosphate. tioned fragments, of another two bands, corresponding to the 43-kDa domain and a 33-kDa fragment of this domain ( Fig. 1 and Ref. 8). ATP cannot trap the wild-type enzyme in this conformation because hydrolysis resets the enzyme to its native form (Fig. 1). We probed the A 2 B 2 Ala42 complex using limited proteolysis to determine the state of the ATP-operated clamp in the presence of ATP or ADPNP. The characteristic complex II signature was apparent after treatment with trypsin in the presence of both nucleotides (Fig. 1). These results suggest that upon incubation of the A 2 B 2 Ala42 complex with ATP, the nucleotide binds to the enzyme and induces dimerization of the B subunits but due to the inability of this complex to hydrolyze the nucleotide, it becomes trapped in the dimerized conformation (complex II).
ATP Hydrolysis Can Proceed Only at One Site-We addressed the possibility of ATP hydrolysis occurring only at one of the two ATP-binding sites in the case where both sites are occupied by the nucleotide. To achieve this we formed heterotetramers of gyrase that contained, apart from the two GyrA protomers, one wild-type GyrB subunit and one mutant Gyr-B Ala42 subunit. Such complexes can be formed by reconstituting the gyrase heterotetramer in the presence of both wild-type and mutant GyrB proteins (17). Electron microscopy and crosslinking studies have failed to detect any dimeric forms of GyrB in the absence of GyrA, suggesting that a significant proportion of GyrB is monomeric in solution (18,19). Therefore, we expect that by allowing a mixture of the wild-type and mutant GyrB proteins to equilibrate and then adding GyrA, a random distri-bution of gyrase complexes containing two wild-type, two mutant, or one of each kind of GyrB proteins would be produced. Indeed, in similar experiments, O'Dea and co-workers (17) found no bias for the formation of any particular one of the above complexes. It is unlikely that heterogeneous tetramers could be studied separately due to the uncertainties concerning the stability of the gyrase tetramer (i.e. such an isolated population would probably re-equilibrate to a mixture of heterogeneous and homogeneous tetramers). Therefore, we decided to study the behavior of complexes formed at different ratios of wild-type to mutant subunit and analyze the results obtained in terms of the predicted population of each tetrameric species, assuming random association.
We tested the ability of heterogeneous tetramers to hydrolyze ATP. Mixtures containing varying amounts of homogeneous and heterogeneous tetramers were formed by mixing a constant concentration of the wild-type subunit with increasing concentrations of the mutant protein (up to a 12-fold excess). GyrA was always in excess of the total B protein, and the ATPase reactions of both the DNA-free enzyme and the gyrase-DNA complex were studied. Increasing concentrations of the mutant protein resulted in inhibition of the rate of hydrolysis (Fig. 2). However, this inhibition was not full, and, in the case of the gyrase-DNA complex, the ATPase rate seemed to plateau at approximately 40% of the rate exhibited by the sample containing only wild-type enzyme-DNA complex. In the absence of DNA, increasing concentrations of the mutant resulted in a decrease in the rate of hydrolysis that levelled out at ϳ65% of the rate of the reaction containing only the DNA-free wildtype enzyme (Fig. 2). The ATPase rates observed were specific to the gyrase complexes and not due to any contamination, because they were completely inhibited by the addition of novobiocin to the reaction. We calculated the amount of tetramers with two wild-type subunits at the concentrations of wild-type and mutant protein used in these experiments and were able to estimate the theoretical rate of hydrolysis in the case where the only complexes that were able to hydrolyze the nucleotide were those that contained two wild-type B subunits (see the Appendix). The theoretical curve representing this situation is shown Ala42 in the dimerized form. Samples contained 0.3 mg/ml GyrA, 0.3 mg/ml GyrB (wild type or mutant), 0.4 mg/ml linear pBR322, and 2 mM ATP or ADPNP (where indicated) and were incubated for 1 h at 25°C to allow the nucleotide to bind. 10 g/ml trypsin was then added, and the reactions were incubated for 1 h at 37°C. The results were analyzed by SDS-polyacrylamide gel electrophoresis. On the right is a diagrammatic representation of the gyrase fragment corresponding to each band. The letter denotes the subunit (A or B), and the number indicates the approximate size of the fragment in kDa. in Fig. 2. The deviation of the experimental data from this theoretical curve is clear. These results suggest that ATP can be hydrolyzed independently only at one site but that the rate of hydrolysis is lower than when both sites are capable of undergoing this reaction. The turnover number of the wild-type enzyme, in the absence of DNA, was measured to be k BB ϪDNA ϭ 0.12 s Ϫ1 . By estimating the concentration of the different species in the above mixtures, the turnover number exhibited by the heterogeneous complexes can be determined. The equation that relates the observed rate of hydrolysis to the concentration of the wild-type and mutant proteins, in terms of the turnover numbers of the wild-type enzyme and the heterogeneous tetramer, is shown in the Appendix. By fitting the experimental data to Equation 5 (see the Appendix) the turnover number of the heterogeneous complex was determined to be k BB* ϪDNA ϭ 0.04 s Ϫ1 . In the presence of DNA, the wild-type enzyme was measured to hydrolyze the nucleotide with a turnover number of k BB ϩDNA ϭ 1.06 s Ϫ1 , compared with an estimated k BB* ϩDNA ϭ 0.18 s Ϫ1 in the case of the heterogeneous tetramer-DNA complex.
In another experiment, mixtures of tetramers were formed where the total concentration of GyrB protein (wild type and mutant) was kept constant while the ratio of the wild-type to mutant subunit was varied from 100% GyrB Ala42 to 100% wildtype GyrB. The rate of hydrolysis for both the enzyme-DNA and the DNA-free complexes was measured; the results obtained from this experiment were plotted together with two theoretical curves (Fig. 3A). One of these curves describes the situation where heterogeneous tetramers cannot hydrolyze ATP (complete inhibition), and the other shows the situation where wild-type GyrB subunits can hydrolyze the nucleotide at the same rate irrespective of them being part of heterogeneous or homogeneous tetramers (no inhibition). The experimental data fall between those two lines. The rate of the enzyme-DNA complex is the one affected more by the formation of the heterogeneous tetramer and is closer to the curve describing complete inhibition, whereas the rate of the DNA-free complex is affected less and is closer to the theoretical curve of no inhibition, consistent with the results in Fig. 2. By fitting these results to Equation 6, the turnover number of the heterogeneous complexes can be determined. The values obtained in this experiment (k BB* ϪDNA ϭ 0.03 s Ϫ1 for the DNA-free complex and k BB* ϩDNA ϭ 0.20 s Ϫ1 for the DNA-bound one) are in very good agreement with those determined in the experiment described in Fig. 2, giving an average of k BB* ϪDNA ϭ 0.035 s Ϫ1 and k BB* ϩDNA ϭ 0.19 s Ϫ1 for the two experiments. ATP Hydrolysis at One Site Is Sufficient for Strand Passage-To test the ability of heterogeneous complexes to support supercoiling, samples were removed from the above ATPase reactions at regular intervals, and the extent of supercoiling was determined (Fig. 3B; the reactions were performed in the presence of relaxed pBR322 DNA and in the absence of spermidine). The wild-type A 2 B 2 complex produced a distribution of topoisomers that had a center that differed from the center of the distribution of the relaxed substrate by a linking number (⌬Lk) of 20. The limit of supercoiling lowered gradually as the ratio of mutant to wild-type increased. At 90% GyrB Ala42 the center of the distribution had a ⌬Lk of 17 (Fig. 3B). At 100% GyrB Ala42 supercoiling was virtually abolished (the limited amount of supercoiling observed in this case is probably due to low levels of contamination of the GyrA preparation with wildtype GyrB and does not affect the results of this experiment). The levels of supercoiling observed in the case of the 90% mixture was not due to the low levels of wild-type A 2 B 2 complexes present in the reaction (1%). This was confirmed by directly comparing the level of supercoiling supported from the heterogeneous mixture to that supported by an amount of wild-type enzyme equal to the predicted concentration of wildtype complexes in this mixture. In this experiment, at the time when the heterogeneous mixture had reached the limit of the supercoiling reaction, the wild-type enzyme had been unable to support significant levels of supercoiling (data not shown). This extent of supercoiling is due to a catalytic reaction, because the levels of enzyme used are not sufficient to ascribe this result to a stoichiometric reaction. It appears from these results that heterogeneous tetramers have not lost their ability to supercoil DNA but that the limit of this reaction is lower. The experiments reported here with GyrB Ala42 were also performed with the GyrB Gln42 mutant, which also binds but does not hydrolyze ATP (13), yielding similar results (data not shown).

DISCUSSION
The A 2 B 2 Ala42 complex is unable to hydrolyze ATP (13), and this is manifested by the trapping of this protein in the dimerized conformation (complex II) in the presence of the nucleotide FIG. 3. ATP hydrolysis at one site is sufficient for strand passage. A, mixtures of heterogeneous and homogeneous tetramers were formed by keeping the total GyrB (wild type and mutant) concentration, [B] tot , constant at 40 nM, whereas the ratio of wild type to mutant was varied from 100% GyrB Ala42 to 100% wild-type GyrB. The concentration of GyrA was 60 nM, and 20 nM relaxed pBR322 was present where appropriate. The rates shown here are the averages of three determinations. Results are presented as the ratio of the measured rate to the rate of the sample containing only wild-type gyrase. B, samples were removed from the above reactions at 30-min intervals and were analyzed by agarose gel electrophoresis to measure the limit of supercoiling. The 1% agarose gel contains 1 g/ml chloroquine. At this concentration of chloroquine, the relaxed topoisomers appear positively supercoiled on the gel, whereas the negatively supercoiled products of the reaction still appear negatively supercoiled but with lower absolute superhelical density. (Fig. 1). This result also reveals that the complex II-characteristic proteolytic signature is not an attribute of the particular complex formed with ADPNP but reflects the conformation of the catalytic intermediate formed in the presence of ATP. Thus, the structure of complex II formed with ATP or ADPNP is indistinguishable, at least at the level detectable by limited proteolysis.
The mechanism of ATP hydrolysis by DNA gyrase can be summarized in the scheme shown in Fig. 4. The central point in this mechanism is the dimerization of the B subunits (formation of complex II). ATP binds to the monomerized subunits within the A 2 B 2 complex, and dimerization has to occur before hydrolysis can take place (7). Binding of ATP in only one of the two subunits is sufficient for dimerization (16,20). Although, once the clamp is in the dimerized form, binding of a second ATP molecule appears not to be possible (6), no direct experimental evidence exists on this issue. It has recently been reported that when ATP binding at one of the two sites was abolished, gyrase could not perform catalytic supercoiling (17). However, it is not clear whether this is because the complex with one ATP has poor or no ATPase activity or whether hydrolysis in this case is uncoupled from strand passage. Nevertheless, with wild-type enzyme, occupation of the two ATPbinding sites is cooperative; therefore the complex with two ATP molecules bound should be the predominant species at high ATP concentrations (10 -12). Moreover, experiments with the 43-kDa domain of gyrase showed that dimerization of two ATP-bound domains is favored over dimerization of one bound and one free domain by 2 orders of magnitude (20). ADP does not stabilize the dimerized form (7), resulting in the monomerization of the subunits after hydrolysis (Fig. 4). Hydrolysis in the doubly occupied complex comprises two hydrolysis reactions. In the experiments described in this paper, we set out to address the issue of interdependence between these two hy-drolysis events and their coupling to strand passage.
We found that complexes containing only one ATPase-proficient GyrB subunit hydrolyze the nucleotide at a lower rate than the wild-type enzyme. The pathway that describes reactivity at one of the two sites is shown in the upper shaded part of Fig. 4. The spectrophotometric assay used in these experiments for determining the rate of hydrolysis measures changes in the concentration of free ADP; therefore the rate of the actual hydrolysis reaction cannot be measured separately from that of the monomerization and product release steps. However, experiments with the 43-kDa domain of GyrB suggested that the rate-limiting step in the mechanism of ATP hydrolysis by gyrase is either a conformational change associated with monomerization or product release (7). Thus, due to the two ATP molecules hydrolyzed per round, the apparent rate constant of the monomerization/product release step in the wildtype complex can be determined to be k wt app ϭ k BB ϩDNA /2 ϭ 0.53 s Ϫ1 . In the case of the heterogeneous tetramer, the mutant site would contain the unhydrolyzed nucleotide after hydrolysis on the other site had occurred. This would likely result in stabilization of the dimerized form, inhibiting the rate of the monomerization/product release step. Therefore, in the heterogeneous tetramer-DNA complex, the rate-limiting step slows down to k h app ϭ k BB* ϩDNA ϭ 0.19 s Ϫ1 (note that only one ATP is hydrolyzed per heterogeneous tetramer). In the absence of DNA, gyrase exhibits a much lower turnover number, k BB ϪDNA ϭ 0.12 s Ϫ1 in our experiments. This is because binding of gyrase to DNA stimulates the ATPase activity of the enzyme (9). It is not yet clear what is the rate-limiting step in the reaction of the DNA-free complex. If monomerization or product release is again the rate-limiting step, then the results obtained here with the heterogeneous tetramer could be explained in terms of the inhibition of the monomerization/product release step by the bound nucleotide. If this is the case, DNA binding should stimulate the ATPase activity by accelerating the monomerization/product release part of the mechanism. However, it is possible that in the absence of DNA, the rate-limiting step is the actual hydrolysis reaction. DNA binding would stimulate the rate of this step, thus making monomerization/product release rate-limiting in the enzyme-DNA complex.
The limit of the supercoiling reaction in the presence of the heterogeneous tetramers was ⌬Lk 17 in contrast with ⌬Lk 20 observed in the case of wild-type gyrase. The value of ⌬Lk 20 is in good agreement with that determined previously for the limit of supercoiling in the absence of spermidine (21). Assuming that the limit of the supercoiling reaction is directly related to the free energy of ATP hydrolysis (4), the lower absolute superhelical density reached in the case of the heterogeneous complex should reflect the reduction in the free energy released when only one ATP is hydrolyzed. However, the free energy released by the extra ATP molecule does not reflect the difference in the free energy of the products, if this is determined according to the relationship described by Cullis et al. (4). In other work, the dependence of the limit of supercoiling on the free energy of ATP hydrolysis was studied by varying the ratio of [ATP]/[ADP] present in the reaction while keeping the total concentration of nucleotides constant (21). These data suggested that, especially in the absence of spermidine, the relationship between linking number change and phosphate potential is not proportional, due to significant ATPase slip and ATP-independent DNA relaxation (21). We believe that the results obtained here reflect a complicated equilibrium established between the ATP-dependent supercoiling and ATP-independent relaxation reactions of the wild-type and heterogeneous complexes. Therefore, making any correlation between the FIG. 4. The mechanism of ATP hydrolysis by DNA gyrase. This schematic diagram shows the steps involved in the mechanism of hydrolysis by DNA gyrase. The upper shaded part encloses the pathways undergone by the heterogeneous tetramers. E-E is used here to represent the dyadic symmetry of the gyrase molecule. EE indicates the conformation of gyrase where the B subunits are in the dimerized (clamp-closed) form (complex II). The rate constants are discussed in the text; the equilibrium dissociation constants K d1 and K d2 are according to Ref. 12. limit of supercoiling and the free energy of ATP hydrolysis is beyond the capacity of these experiments.
In conclusion, we found that gyrase can hydrolyze ATP at one active site and can supercoil DNA even with only one ATP hydrolyzed at a time. Although this reaction is unlikely to occur in vivo, due to the physiological ATP concentration being relatively high, the results obtained here provide us with useful information on the mechanism of energy coupling in DNA gyrase.