On the Coupling between ATP Usage and DNA Transport by Yeast DNA Topoisomerase 11”

The initial rates of ATP hydrolysis and relaxation of negatively supercoiled DNA by highly purified wild- type and mutant yeast DNA topoisomerase I1 were measured under identical conditions to study the cou- pling between the ATPase activity of a type I1 DNA topoisomerase and its catalysis of the transport of one DNA segment through another. The results indicate that the binding of the enzyme to DNA stimulates its intrinsic ATPase activity by about 2O-fold, and ATP binding to the pair of ATPase sites in a DNA-bound dimeric enzyme appears to be cooperative. The cooperativity in ATP binding may be significant in the coordination of the two halves of a DNA-bound enzyme dimer. At low ATP concentrations, the rate-limiting step in ATP usage appears to be slower than that in DNA transport, and DNA transport is relatively effi- cient in terms of ATP consumption: 1.9 f 0.5 ATP molecules are hydrolyzed/DNA transport event. At a saturating ATP concentration, however, there appears to be a reversal of these rate-limiting steps, and DNA transport is less efficient: 7.4 A 1.0 ATP molecules are hydrolyzed/DNA transport event. These data are in-terpreted in terms of a model in which a DNA-bound enzyme acts as an ATP-operated clamp for the capture and transport of a second DNA segment.

Type I1 DNA topoisomerases (EC 5.99.1.3) are ubiquitous enzymes that catalyze the ATP-dependent transport of one double-stranded DNA segment through an enzyme-mediated transient break in another (reviewed in Maxwell and Gellert, 1986;Hsieh, 1990;Reece and Maxwell, 1991; see also Wang, 1985;Caron and Wang, 1993). In such a reaction, the enzymeoperated gate in the double-stranded DNA is opened through the attack of a pair of active site tyrosyl hydroxy groups, one in each half of the dyadic enzyme, on a staggered pair of DNA phosphodiester bonds. Breakage of the pair of DNA backbone bonds creates a transient break or gate in the double-stranded DNA segment, and at the same time a pair of covalent links is formed between the active site tyrosines and the 5' phosphoryl ends of the severed DNA strands. Closing of the DNA gate is achieved through nucleophilic attack of the 3' hydroxys of the severed DNA ends on the enzyme-DNA phosphotyrosine linkages. The same enzyme molecule that operates the DNA gate can transport a second double-stranded DNA segment through the DNA gate. All known type I1 DNA topoiso-*This work was supported by a United States Public Health Service Grant GM 24544 and by a postdoctoral fellowship (to J. E. L.) from the American Cancer Society. 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.
$ To whom correspondence should be addressed.
How a type I1 DNA topoisomerase couples ATP binding and hydrolysis to the transport of DNA segments through each other is a key question in mechanistic studies of this class of enzymes. We have recently used highly purified type I1 DNA topoisomerase from the budding yeast Saccharomyces cereuisiae in our mechanistic studies of this class of enzymes. S. cereuisiae DNA topoisomerase I1 is a typical eukaryotic type I1 DNA topoisomerase, and each polypeptide of the homodimeric enzyme (Goto et al., 1984) is 1,429 amino acids long . From a comparison of the amino acid sequences of the yeast enzyme and Escherichia coli DNA gyrase (DNA topoisomerase 11), it can be inferred that the ATPase domain of the yeast enzyme is contained within the amino-terminal 400 amino acids. This inference is based mainly on the sequence homology  and the known crystal structure of a 43-kDa amino-terminal fragment of E. coli gyrase B-subunit complexed with a nonhydrolyzable ATP analogue AMPPNP' (Wigley et al., 1991). The active site tyrosine involved in DNA breakage and rejoining is Tyr783 of the yeast enzyme (Worland and Wang, 1989), which corresponds to Tyr-122 of E. coli gyrase A-subunit (Horowitz and Wang, 1987). Recently, proteolysis studies of yeast DNA topoisomerase I1 by SV8 endoprotease in the presence and absence of AMPPNP or ATPyS, another nonhydrolyzable ATP analogue, have provided evidence that there are allosteric interdomainal movements in the enzyme following the binding of ATP (Lindsley and Wang, 1991). Furthermore, results on the binding of different forms of DNA to the yeast enzyme and its AMPPNP complex have led to a model in which a type I1 DNA topoisomerase, either free or bound to a DNA segment, acts as an ATP-dependent protein clamp: in the absence of ATP, the jaws of the clamp are open and a second DNA segment can enter the channel between them; the closure of the jaws upon ATP binding traps the second DNA segment and transports it through the first DNA segment (Roca and Wang, 1992).
To obtain a clearer picture of the roles of ATP in type I1 DNA topoisomerase-catalyzed reactions we have measured the number (n) of ATP molecules hydrolyzed/DNA transport event catalyzed by purified yeast DNA topoisomerase 11. There have already been several measurements of n in the literature. A priori, it is reasonable to assume that n is greater than 1 and can easily be 2, as there are two ATPase sites/ enzyme molecule. Liu et al. (1979) estimated that n is 1-2 for phage T4 DNA topoisomerase, a type I1 enzyme mechanistically very similar to eukaryotic DNA topoisomerase 11. A The abbreviations used are: AMPPNP, adenosine 5'(P,-y-im-id0)triphosphate; ATP-yS, adenosine 5'-3-0-(thio)triphosphate; kb, kilobase(s). value of n around 0.8 was arrived a t from a more extensive set of measurements with E. coli DNA gyrase (Sugino and Cozzarelli, 1980). Because bacterial gyrase, unlike eukaryotic DNA topoisomerase I1 or phage T4 DNA topoisomerase, catalyzes the endergonic negative supercoiling of DNA, the level of ATP usage can also be estimated from thermodynamic considerations of the endergonic DNA supercoiling reaction and the exergonic ATP hydrolysis reaction; at the maximal attainable level of DNA negative supercoiling by E. coli gyrase, n was estimated to be about 2 (Sugino and Cozzarelli, 1980;Tamura et al., 1992). For Drosophila DNA topoisomerase 11, which, like all other eukaryotic type I1 and T-even phage DNA topoisomerases, catalyzes the ATP-dependent relaxation of supercoiled DNA, about eight ATP molecules were found to be hydrolyzed/DNA transport event (Osheroff et al., 1983).
Whereas these earlier measurements and estimates were in general agreement within a factor of 10, there was a fundamental problem in the calculation of n from these measurements. Because of the relatively high turnover number for type I1 DNA topoisomerase-catalyzed DNA transport, of the order of 1 s-l (Higgins et al., 1978;Sugino and Cozzarelli, 1980;Osheroff et al., 1983;Maxwell and Gellert, 1984;Baker et al., 1987), all experiments cited above were carried out with a low molar ratio of enzyme to DNA molecules so that the net reaction would continue for at least several minutes to permit manual measurements. Under these conditions, however, the processive nature of the DNA-bound enzyme predicts that the linking number of a DNA with a bound enzyme would reach that of the final product in seconds; thus the measured total linking number change for the entire population of DNA molecules may depend strongly on how fast a DNA-bound enzyme can dissociate and reassociate with a different DNA molecule. The rate of linking number change measured under these conditions thus may have little to do with the enzyme-catalyzed rate of DNA transport. Theoretical estimates based on free energy considerations are not subject to this complication; these estimates can only set a minimal value for n for the bacterial gyrase-catalyzed DNA supercoiling reaction, however, and even this minimal estimate is valid only at the maximal level of DNA negative supercoiling attainable.
In the measurements reported here, we have employed conditions that avoid the complication described above. Our results indicate that when the ATP concentration is in the PM range, the slow step in the yeast DNA topoisomerase IIcatalyzed reaction pathway is probably ATP binding or hydrolysis, and n is 1.9 0.5; when ATP is in excess, the rate of DNA transport is considerably slower than the rate of ATP hydrolysis, and n is 7.4 f 1.0. Through the use of ATPase mutants of yeast DNA topoisomerase 11, in which Gly-144 of the wild-type enzyme is replaced by an isoleucine, valine, or proline, we show also that mutational inactivation of the ATPase, similar to the omission of ATP in the reaction with the wild-type enzyme, completely blocks the DNA transport activity. In terms of the ATP-modulated protein clamp model (Roca and Wang, 1992), these results suggest that conformational changes in the protein, which are associated with the closure of the clamp triggered by ATP binding, are responsible for the capture and transport of a DNA segment through the DNA gate operated by the same enzyme in another DNA segment; the rates of opening and closing of the protein clamp are likely to increase with increasing ATP concentration, but at a high ATP concentration a protein clamp may often close without trapping a DNA segment for transport.
Overexpression of Wild-type and Mutant S. cerevisiae DNA Topoisomerase ZZ-Wild-type yeast DNA topoisomerase I1 was overexpressed in yeast from the inducible promoter PGALl in a multicopy expression plasmid YEpTOPP-PGALI, as described (Worland and Wang, 1989). The BamHI to KpnI fragment from YEpTOP2-PGAL1, encoding approximately the amino-terminal quarter of the enzyme, was first subcloned into pBluescript KS'. Site-directed mutagenesis was carried out with this subclone using an oligonucleotide-directed in vitro mutagenesis kit (Amersham version 2). Each of the mutant BamHI to KpnI regions was sequenced to confirm the intended nucleotide changes and cloned back into YEpTOP2-PGAL1 for overexpression. The ability of mutant top2 to complement a conditional lethal top2 allele top2-4 (Holm et al., 1985) was tested by transforming a top2-4 strain CH1106 cells (Gartenberg and Wang, 1992) with the mutant versions of YEpTOP2-PGAL1. Colonies of transformants were tested for growth at 35 "C, a nonpermissive temperature for untransformed CH1106 cells, on minimal agar plates with 2% glucose and no uracil. Even in the presence of 2% glucose, there is sufficient expression of the PGAL1-linked wild-type TOP2 gene to complement the thermal sensitive product of the top2-4 mutant allele.
Expression vectors were transformed into the yeast strain JELl (a leu2 trpl urd-52 prbl-1122 pep4 Ahis3::PGALI-GAL4). In this strain, the expression of GAL4 protein is also induced by galactose; the higher level of the GAL4 protein upon induction with galactose in turn improves the expression of genes linked to GAL4-activated promoters such as PGALl (Schultz et al., 1987). Strain JELl was derived from the protease deficient strain NKY879 ( a leu2 trpl urd-52 prbl-I122 pep4) by the use of an integrating plasmid pKHint-C, in which a GAL4 gene linked to a GAL10 promoter is flanked by the HIS3 sequences (a kind gift of Dr. James E. Hopper, Hershey Medical School). Integration of the plasmid into the chromosomal HIS3 locus inactivates the HIS3 gene and introduces a URA3 marker, which is present on the integration plasmid the URA3 marker was subsequently disrupted by targeted gene transplacement (Rothstein, 1983), using a AURA3 gene missing 250 base pairs in the middle section of the marker gene, to restore JELl to ura-for transformation with the URA3-marked overexpression plasmids. The levels of expression of DNA topoisomerase I1 from YEpTOP2-PGAL1 are typically 5-10 times higher in the JELl strain than that in the NKY879 parent (data not shown). Cell growth and induction with galactose were done as described previously (Worland and Wang, 1989).
Protein purification was done essentially as described (Worland and Wang, 1989), except that two additional protease inhibitors, leupeptin (0.5 pg/ml) and pepstatin (0.7 pglml), were included in all but the final dialysis buffer. For experiments measuring the number of ATP molecules hydrolyzed/DNA transport event, the protein was further purified on an high pressure liquid chromatography MATQ anion exchange column (Bio-Rad). Yeast DNA topoisomerase I1 came off at 0.38-0.4 M KC1 upon eluting the column with a 0.15-0.5 M KC1 linear gradient. All proteins were stored at -70 'C and protein concentrations of 1 mg/ml or higher, in 20% glycerol (v/v), 50 mM Tris-HCl (pH 7.51, 150 mM KCl, and 5 mM 2-mercaptoethanol. All experiments were done with preparations stored for no more than 3 months.
Enzyme Assays-The standard reaction buffer consisted of 50 mM Tris acetate (pH 7.81, 150 mM potassium acetate, 6 mM magnesium acetate, 5 mM 2-mercaptoethanol, and 250 pg/ml bovine serum albumin (Sigma, fraction V). In the determination of the number of ATP hydrolyzed/DNA transport event, the salts in the buffer were changed slightly to 120 mM potassium acetate and 8 mM magnesium acetate to improve the processivity of the DNA-bound enzyme; this change did not significantly alter the enzymatic activity of the toPoisomerase. The DNA used in the assays was either a 2-kb plasmid pHC624 (Worland and Wang, 1989) or a 2.95-kb plasmid pBluescript KS'. The concentrations of topoisomerase 11, DNA, and ATP are indicated in the appropriate figure legends, and the reaction temperature was 30 "C unless otherwise indicated.
ATPase Assays-The ATPase activity of yeast DNA topoisomerase I1 was measured by two methods. In the spectrophotometric assay, rapid conversion of ADP back to ATP by pyruvate kinase and phosphoenolpyruvate, a reaction coupled to NADH oxidation, was used to measure the rate of ATP hydrolysis; the procedure described in Morrical et al. (1986) was followed (see also Tamura and Gellert, 1990). In this assay, the concentration of ATP remains at its initial concentration. Each reaction mixture (1 ml) contained 2 mM phosphoenolpyruvate, 5 units of pyruvate kinase, 8 units of lactate dehydrogenase, and 0.16 mM NADH in addition to yeast DNA topoisomerase 11, DNA, ATP, and other reagents in the standard assay buffer. The decrease in NADH concentration as a function of time was monitored by absorbance at 340 nm using a Hewlett-Packard 8452A diode array spectrophotometer. The temperature inside the thermostatted sample cuvette was checked immediately before and after each reaction using a thermocouple (Omega Engineering). This coupled ATP hydrolysis assay was fully functional under all reaction conditions employed; doubling any component of the ATP regenerating/ coupling system had no effect on the measured rate of ATP hydrolysis, whereas doubling the topoisomerase concentration doubled the measured rate of ATP hydrolysis, as expected.
In the TLC assay, the standard reaction buffer was used. ATP (concentration as indicated), magnesium acetate (8 mM final), and [w3*P]ATP (1 pCi, Amersham) were added to initiate the reaction. Aliquots were removed at the indicated time points and quenched by the addition of an equal volume of 50 mM EDTA and 1% sodium dodecyl sulfate to each. The quenched reactions were chromatographed in triplicate (1 pl each) on polyethyleneimine sheets (Polygram Cel 300 PEI, Sybron/Brinkmann), as described in Shibata et al. (1981). The dried sheets were exposed to imaging plates (Fuji, type BAS-111) for 1-10 min, and the exposed screens were scanned using a Bio-Image analyzer (FUJIX BAS2000). The concentration of ATP hydrolyzed was determined by dividing the ADP counts by the total number of counts/lane and multiplying that fraction by the starting ATP concentration.
Determination of the Number of ATP Molecules HydrolyzedlDNA Transport Euent-Purified, supercoiled pBluescript KS' DNA (50 pg) was nicked by pancreatic DNase I in the presence of excess ethidium bromide (Barzilai, 1973;Hsieh and Wang, 1975). The reaction was stopped by the addition of EDTA to 50 mM, and the DNA was purified by two phenol-chloroform extractions followed by ethanol precipitation. A coupled nick translation/ligation reaction of the nicked DNA was carried out by adding E. coli DNA polymerase I (5 units), dATP, dTTP, dGTP (0.8 mM each), and [cx-~'P]~CTP (50 pCi) in a medium containing 50 mM Tris-HC1 (pH 81, 10 mM MgC12, and 50 pg/ml bovine serum albumin. After 5 min at 15 "C, NAD (26 p~) and E. coli DNA ligase (20 units) were added, and the incubation was continued for 30 min. The reaction was stopped with excess EDTA. Following two phenol-chloroform extractions, the DNA was dialyzed against two changes of 500 ml of 2 M NaCl to remove the nucleotide cofactors and one change of 500 ml of 10 mM Tris-HC1 (pH 8), 0.1 mM EDTA. The labeled, relaxed DNA was then combined with nonlabeled pBluescript DNA (500 pg), ethidium bromide (31 pg), and vaccinia virus DNA topoisomerase I (2 pg, a generous gift from Dr. Ryo Hanai) in a buffer containing 10 mM Tris acetate (pH 7.4), and 100 mM potassium acetate for 2 h at 37 "C. The reaction was stopped with two phenol extractions, and negatively supercoiled DNA with approximately 22 negative supercoils/2.95 kb of monomeric pBluescript DNA was purified from nicked DNA by CsClethidium bromide density gradient centrifugation (Radloff et al., 1967). Less than 10% of the total DNA used in these reactions was nicked, and therefore no correction was made for this contaminant in the quantitation of the rate of relaxation of the DNA.
Two series of reactions were carried out. When the rate of ATP hydrolysis was followed spectrophotometrically, ATP hydrolysis and DNA linking number measurements were performed separately but always on the same day. The DNA used in the ATPase assays contained no radiolabeled tracer, and the DNA relaxation experiment was carried out in a smaller volume; otherwise the two sets of measurements were carried out under closely matched conditions. When the ATPase rate was followed by TLC, the same reaction mixture was used in both the ATPase assays and the DNA relaxation measurements; reactions were performed in the standard buffer except that the potassium acetate concentration was 120 mM and that magnesium acetate was added (to 8 mM final concentration) along with ATP to initiate the reactions. Topoisomerase and DNA concentrations used are indicated in the table and figure legends. Prior to these experiments, the fraction of active topoisomerase in each protein preparation was determined by titrating 20 nM supercoiled DNA with topoisomerase under highly processive reaction conditions (50 instead of 120 mM potassium acetate). Assuming a Poisson distribution of protein dimers/plasmid, the concentration of protein that relaxes 64% of the plasmids was taken to correspond to one active protein/DNA molecule. In most of the enzyme preparations used, the active enzyme concentration was found to be about one-third to onehalf of the total protein concentration. In most of the measurements of n, the number of ATP hydrolyzed/DNA transport event, the molar DNA to active dimeric enzyme ratio in each reaction mixture was adjusted to be 1; changing this ratio by a few fold is not expected to affect the measured n. Unless otherwise noted, the topoisomerase concentration refers to the total rather than active protein concentration. To ensure a random distribution of topoisomerase on DNA, the reaction buffer without DNA, enzyme, Mg(I1) and ATP were first divided into two equal portions; topoisomerase was added to one and DNA to the other. The two portions were rapidly mixed, preincubated for 10 min at the desired temperature, and magnesium acetate and ATP were then added to initiate the reactions.
Aliquots of the DNA relaxation reactions were removed from the reaction tube at the desired time points (generally every 5-10 s) and quenched with an equal volume of 50 mM EDTA, 1% sodium dodecyl sulfate. If the rate of ATP hydrolysis was to be determined by TLC, 1-pl aliquots were directly spotted onto the TLC plates. To each of the remaining quenched reaction, an equal volume of 2 X gel loading buffer (20% glycerol, 0.02% bromphenol blue, 0.02% xylene cyanol, and 12 mM EDTA) was added. A set of four samples (10 pl each) was loaded in four equally spaced sample wells of each 20 X 20 X 0.5 cm 1% agarose gel slab containing TBE (0.1 M Tris borate, 2 mM EDTA (pH 8.3)) plus 1 pg/ml chloroquine diphosphate. Following electrophoresis in the first dimension (35 V, 22 h), the gel slab was equilibrated in TBE plus 6 pg/ml chloroquine diphosphate, and electrophoresis was performed in the second dimension in the same buffer (35 V, 19 h). Gels were dried onto filter paper (Whatman 3 " ) and exposed to imaging plates for quantitation of the topoisomers and subsequently to x-ray films for graphical presentations. Each imaging plate was scanned, and the relative amount of DNA in each topoisomer spot was determined. The total increase in linking number for the entire population of topoisomers over that of the same sample at time zero was obtained for each time point, and the number of the DNA transport event was calculated from this linking number change divided by 2, as each DNA transport event in the relaxation of a negatively supercoiled DNA increases the linking number by 2 (see the reviews cited in the Introduction). A plot of the number of DNA transport events uersus time was matched by a polynomial fit, and the rate at time zero was determined from the first derivative of this curve at time zero.
DNA Cleauuge Experiments-The cleavage of DNA by yeast DNA topoisomerase I1 in the presence of VP-16 was performed as described (Worland and Wang, 1989), and the enzyme and DNA concentrations are given in the legend to Fig. 6. In the cleavage experiments in media containing Ca(II), the DNA used was a dumbbell-shaped oligonucleotide with short hairpin turns at the two ends of a 40-base pair duplex segment (see "Materials"). The 5' ends of the two initially singlestranded oligonucleotides were first phosphorylated with [r3'P1ATP and T4 polynucleotide kinase. The oligonucleotides were then annealed and ligated to form a labeled, covalently closed 88-nucleotidelong continuous strand with a 40-base pair duplex region in the middle. Reaction conditions for topoisomerase-induced DNA cleavage were as described in the figure legend. The final reactions were mixed 1:1 with the 2 X gel loading buffer and loaded in the sample wells of a 15% nondenaturing polyacrylamide gel. Electrophoresis was performed in TBE buffer at 250 V for 3 h.

RESULTS
DNA-independent ATPase Actiuity of Yeast DNA Topoisomerase 11-Purified yeast DNA topoisomerase I1 exhibits a weak but readily detectable ATPase activity in the absence of DNA. Hydrolysis of ATP by this activity is adequately represented by Michaelis-Menten kinetics, as shown by the linear u/[ATP] uersus u plot depicted in the inset of Fig. 1 ( u being the rate of ATP hydrolysis). From the data shown, which were obtained from four independent sets of measurements using the same enzyme preparation, values for the maximal velocity Vmax and the Michaelis constant K,,, are calculated to be 1.1 pM/min and 0.3 mM, respectively. From the total amount of yeast enzyme in the reaction mixture, the apparent turnover number kc,, can be estimated to be 0.4/s/ dimeric enzyme at 30 "C (in a medium containing 50 mM Tris acetate (pH 7.8), 150 mM potassium acetate, 6 mM magnesium acetate, 5 mM 2-mercaptoethanol, and 250 pg/ml bovine serum albumin). Comparable VmBx and K, values were obtained with a different preparation of the same enzyme. The absolute kcat value is likely to be two to three times higher or around l/s/dimeric enzyme, as estimated from the fraction of active enzyme in a typical preparation (see "Experimental Procedures"). As will be discussed in a later section, this DNA-independent ATPase activity is not caused by a contaminating ATPase in the enzyme preparations but is largely intrinsic to the topoisomerase.
Strong Dependence of the ATPase Activity of Yeast DNA Topoisomerase 1 1 on DNA and Cooperatiuity between the two ATPase Sites in a DNA-bound Enzyme-The addition of DNA to yeast DNA topoisomerase I1 greatly stimulates the ATPase activity, as expected from previous studies with other type I1 DNA topoisomerases (see for example Mizuuchi et al., 1978;Liu et al., 1979;Osheroff et al., 1983). At a total enzyme dimer concentration of 50 nM, half-maximal stimulation of ATPase by DNA was reached at 5 p~ base pairs or an enzyme dimer to DNA base pair ratio of 1:lOO (data not shown). No difference in the rate of ATP hydrolysis was observed in the presence of 30 p~ (in base pairs) of negatively supercoiled, linear, or relaxed covalently closed DNA; this lack of dependence on the topological form of the input DNA effector is expected, as a supercoiled DNA would be completely relaxed in the first few seconds under these conditions (see below).

Topoisomerase II 8099
In reactions used to determine the steady-state kinetic constants of DNA-dependent hydrolysis of ATP by yeast DNA topoisomerase 11, the dimeric enzyme to DNA base pair ratio was set at 1:600 to ensure DNA excess and a maximal level of ATPase stimulation. In Fig. 1, the rate u of ATP hydrolysis is plotted as a function of [ATP] for four sets of experiments carried out at 30 "C and in the 150 mM potassium acetate medium described above. The maximal velocity, V,,,.,, at high ATP concentrations is approximately 21 phi min", which is 20 times higher than the value 1.1 p~ min" in the absence of DNA. Significantly, the u uersus [ATP] plot is sigmoidal in shape in the presence of DNA, suggesting cooperativity in the ATPase kinetics (Fig. 1). Deviation from Michaelis-Menten kinetics is also evident when u/[ATP] is plotted uersus [ATP], which shows distinct curvature at ATP concentrations below 125 p~ (plot not shown). Digestion of the DNA with AluI and H h I restriction endonucleases, so as to generate DNA fragments shorter than 400 base pairs, had no effect on the sigmoidal shape of the curve shown in Fig. 1. Thus the cooperativity is probably a result of interaction between the two ATPase sites in each DNA-bound dimeric enzyme rather than interaction between two dimeric enzyme molecules bound to the same DNA fragment.
The experimental ATPase data in the presence of excess DNA can be fitted by a simple model in which the rate of hydrolysis of ATP occupying an ATPase catalytic pocket is the same whether one or both pockets of the enzyme dimer are occupied, but the dissociation constant is K, for E. ATP and rK, for E . (ATP), (see Segel, 1975). The curve drawn in Fig. 1 is the one calculated for this model with a value of r = 0.01. However, because u is sensitive to the value of r only at low ATP concentrations at which a significant fraction of the enzyme is in the E. ATP form, the relatively large experimental error in u when [ATP] is low makes it difficult to evaluate the cooperativity parameter accurately; the value r = 0.01 should therefore be viewed as an order of magnitude estimate. It is straightforward to show that K, = [ATP]o.5/r1/2, where is the ATP concentration at half-maximal velocity; from the data shown in Fig. 1 dependent ATPase activity of the enzyme does not require the transport of DNA through the DNA gate, the value of n, the number of ATP hydrolyzed/DNA transport event catalyzed by a type I1 DNA topoisomerase, is mechanistically meaningful only under conditions when DNA transport readily accompanies ATP hydrolysis; otherwise n would approach infinity. Furthermore, as described in the Introduction, n must be measured under conditions such that the rate-limiting step is not the dissociation of a DNA-bound enzyme or its reassociation with a different DNA substrate. At a turnover rate of DNA transport around l/s/enzyme molecule, a typical 5-kb supercoiled DNA with a specific linking difference of -0.06 would be relaxed by a single bound enzyme in about 15 s. T o slow down this rate without resorting to the use of a low enzyme to DNA molar ratio, which may make the dissociation of a DNA-bound enzyme or its reassociation with a different DNA the rate-limiting step, we either lowered the temperature or the ATP concentration.
There is a technical problem in measuring the DNA linking number change by a type I1 DNA topoisomerase under conditions such that the enzyme is processive and that the enzyme to DNA molar ratio is around 1. Because the number of enzyme molecules/DNA ring follows a Poisson distribution, under these conditions different DNA molecules are relaxed a t different rates, and the linking number distribution of the DNA substrate becomes very broad shortly after its incubation with the enzyme, which makes the quantitation of the linking number changes more difficult. T o overcome this problem, we used two-dimensional agarose-gel electrophoresis to resolve all "P-labeled DNA topoisomers of different linking numbers; the overall linking number change was then calculated by quantitating the relative amounts of all DNA topoisomers using a Phospho-Imager (see "Experimental Procedures"). Fig. 2 depicts a typical experiment on the rate of relaxation of a negatively supercoiled DNA by yeast DNA topoisomerase I1 at 7 "C and a saturating level of ATP (1 mM). In this experiment a negatively supercoiled DNA was incubated with the yeast enzyme, and aliquots of the reaction mixture were sampled a t various times for the analysis of DNA linking number distributions by two-dimensional agarose-gel electrophoresis. The rate of ATP hydrolysis was followed spectrophotometrically through an NADH-coupled reaction (see "Experimental Procedures") in a parallel experiment carried out under conditions identical to those of the DNA relaxation experiment. The rate of ATP hydrolysis remained constant during the entire time course, and no significant change in the rate was detectable as the DNA became progressively less negatively supercoiled. It was reported previously that ATP hydrolysis by Drosophila DNA topoisomerase I1 was four times faster in the presence of a negatively supercoiled rather than a linear or relaxed DNA, under otherwise identical conditions (Osheroff et al., 1983). The half-time for the removal of negative supercoils in the experiment shown in Fig.  2 can be estimated to be about 1 min (see below), and a 4-fold drop in the rate of ATP hydrolysis in the first couple of minutes would have been detectable.
From this pair of experiments the rate of ATP hydrolysis was calculated to be 43 nM s-I, and the rate of DNA transport a t time zero was calculated to be 5 nM s"; the ratio of the rates indicates that 8.6 ATP are hydrolyzed/DNA transport event under the experimental conditions employed. From the concentration of active enzyme and the concentration of DNA rings in the reaction mixture, the measured rates corresponds to 1.3 ATP hydrolyzed/s/dimeric enzyme and 0.15 DNA transport event/s/dimeric enzyme. These results and those from similar experiments, measured a t a saturating ATP concentration of 1 mM and temperatures from 7 to 17.5 "c, are tabulated in Table I. The average number of ATP hydrolyzed/DNA transport event, n, is 7.4 f 1, where the error indicated is for a particular measurement of n. The Arrhenius activation energy is estimated to be 20 and 24 kcal mol", respectively, for DNA transport and ATP hydrolysis from these data; the difference between these values is insignificant Reaction time

Rates of yeast DNA topoisomerase II-catalyzed ATP hydrolysis and DNA transport at low temperatures
Reactions were performed at the indicated temperatures in the standard reaction buffer, as described under "Experimental Procedures." Each reaction contained 100 nM enzyme dimers (one-third of which was active), 33 nM pBluescript monomeric DNA, and 1 mM ATP. ATP hydrolysis was monitored spectrophotometrically. The experimentally measured rates of linking number changes were divided by 2 and the active enzyme concentration to give the rates of DNA transport/enzyme; similarly, the rates of ATP hydrolysis were expressed in units of ATP hydrolyzed/s/active enzyme. The quantity n is the number of ATP hydrolyzed/DNA transport event and is obtained directly from the ratio of the rates; this quantity is independent of the fraction of active enzyme molecules in the preparation, as the same preparation was used in both rate measurements. . For the zero time point, an aliquot of the reaction without Mg(I1) and ATP was simultaneously mixed with magnesium acetate and ATP and a stop solution (to give 0.5% sodium dodecyl sulfate and 25 mM EDTA final concentration). All other points were taken by removing aliquots from the original reaction and quenching with the stop solution at the indicated times. The concentration of ATP hydrolyzed at each time point was measured by the TLC method as described under "Experimental Procedures." Three two-dimensional agarose gels were used to separate the topoisomers of all quenched samples. Electrophoresis and quantitation of topoisomers were performed as described under "Experimental Procedures." relative to the uncertainties in the experimental data.

Reaction
In a second set of measurements, low ATP concentrations in the range of 5-25 PM were employed, and the yeast DNA topoisomerase 11-catalyzed reduction in ATP concentration was measured at 30 "C either spectrophotometrically as described above or by thin layer chromotographic analysis of [w3'P]ATP in the reaction mixture. The results of a typical experiment by TLC analysis of radiolabeled ATP are shown in Fig. 3. In such an experiment, ATP was not regenerated, and its depletion was responsible for the nonlinearity of the plot of [ATP] hydrolyzed uers'sus time. From the ratio of the Topoisomerase II 8101 slopes of the two curves shown in Fig. 3 as time approaches zero, n is calculated to be 2.6. Data from the low [ATP] set of measurements are listed in Table 11; the average value of n is 1.9 f 0.5 from this data set.
Mutations in the ATP Binding Domain of Yeast DNA Topoisomerase II and the Absolute Dependence of DNA Passage on ATP-Several mutations were introduced into a putative ATP binding region of yeast DNA topoisomerase 11. Table I11 summarizes the results obtained for five mutants with amino acid substitutions within a short stretch of yeast DNA topoisomerase 11, from Asn-130 to Gly-146; the counterpart of this stretch in E. coli GyrB protein, from Lys-103 to Gly-119, appears to be involved in ATP binding (Tamura and Gellert, 1990;Wigley et al., 1991). Expression of a plasmid-borne K137A mutant top2, in which Lys-137 in the wildtype codon is replaced by an alanine codon, or the DJ32-134A3 mutant top2, in which the three aspartic acid codons 132-134 are replaced by alanines, was found to complement a temperature-sensitive yeast top2 strain carrying a top2-4 mutation in the chromosomal copy of the gene. The lethal phenotype of the top2-4 strain at 35 "C was not rescued, however, by the expression of the G1441, G144V, or G144P

TABLE I1
Rates of topoisomerase II-catalyzed ATP hydrolysis and DNA transport at low concentrations of ATP Reactions were performed at 30 "C in the standard reaction buffer, as described under "Experimental Procedures." The molar ratio of active enzyme dimer to pBluescript monomeric DNA was approximately 1 in all reactions except reaction 6, for which the ratio was about 0.5; the pBluescript DNA concentration was 150 nM (monomer rings) for reaction 5 and 66 nM for all others. The asterisk denotes reactions for which ATP hydrolysis was measured by TLC; otherwise the rates were followed spectrophotometrically. See the Table I legend for the conversion of the experimentally measured rates to those in units indicated in the table and

TABLE I11
Mutations in a predicted ATP binding region of yeast DNA topoisomerase II Amino acids 130-146 of the yeast enzyme are shown. Boldface letters represent amino acids that are conserved among all or most of the type I1 DNA topoisomerases of known sequences; in the regions of other type I1 DNA topoisomerases corresponding to the DDD triplet shown in the yeast enzyme sequence, there is at least one aspartic acid. Bold underlined letters represent the amino acids that have been altered by site-directed mutagenesis. The ability of the enzyme to complement a temperature-sensitive top2-4 mutant allele at 35 "C when expressed from a multicopy plasmid is indicated. The activity of the enzyme in crude lysates of yeast cells overexpressing it was determined by the DNA unknotting assay (Liu et al., 1981).

Enzyme Amino acid sequence Complemen-tation Activity
Wild-type NYDDDEKKVTGGRNGYG

D3132-134A3 NYeEKKVTGGRNCYG
+ + --mutant tope, in which Gly-144 is replaced by isoleucine, valine, or proline, respectively. Assays of the ATP-dependent unknotting of knotted phage P4 DNA by yeast DNA topoisomerase I1 in extracts of cells overexpressing the plasmidborne top2 showed that the K137A and D3132-134A3 mutant enzymes were nearly as active as the wild-type enzyme, whereas the G1441, G144V, and G144P enzymes were inactive. Lys-137 of the yeast enzyme corresponds to Lys-110 of E. coli GyrB, which was shown to link covalently to an ATP analogue (Tamura and Gellert, 1990); the results with K137A indicate that the presence of a lysine at this position is functionally unimportant. Similarly, Lys-103 of E. coli GyrB was shown to form a covalent adduct with the ATP analogue in the same study, but the lack of a lysine or arginine at this position in the yeast enzyme suggests that the presence of Lys-103 in E. coli GyrB is also coincidental rather than obligatory. The amino acid triplets corresponding to D3132-134 of the yeast enzyme contain 1-3 acidic residues in all other type I1 DNA topoisomerases of known sequences (Caron and Wang, 1993). Replacing the triplet by 3 alanines has little effect on the yeast enzyme, however; thus the presence of an acidic residue in the triplet is apparently nonessential.
To characterize the Gly-144 mutant enzymes further, the mutant proteins were purified in parallel with the wild-type enzyme from a protease-deficient strain of yeast expressing a chromosomal copy of wild-type TOP2 gene and overexpressing at a much higher level of a plasmid-borne mutant top2 or wild-type TOP2 gene. As depicted in Fig. 4, in the presence of saturating amounts of DNA the mutant enzymes have no more than 1% of the ATPase activity of the wild-type enzyme. Because the purified mutant enzymes were contaminated by a low level of the wild-type enzyme expressed from the chromosomal TOP2 gene, which is estimated to be less than 0.5% by mass, the residual ATPase activity of the mutant enzymes could not be accurately determined data shown in Fig. 4 show, however, that the ATPase activity in the Gly-144 mutant enzymes must be at least 2 orders of magnitude lower than that of the wild-type enzyme. In the absence of DNA, the ATPase activity was reduced to an almost undectable level for the purified Gly-144 mutant enzymes. This result in turn indicates that the DNA-independent ATPase activity of the wild-type enzyme ( Fig.  1) is most likely an intrinsic activity of yeast DNA topoisomerase I1 rather than an unrelated contaminating ATPase. The purified Gly-144 series of mutant enzymes showed no ATP-independent or ATP-dependent activity in the relaxation of supercoiled DNA. A comparison of the rates of relaxation of negatively supercoiled DNA by wild-type and G144I mutant enzyme is shown in Fig. 5. In the presence of ATP, the wild-type topoisomerase relaxed fully half of the supercoiled plasmids within 30 s. There was no detectable relaxation by the mutant protein, however, after 60 min of incubation. Similar results were obtained for G144V and G144P mutants (data not shown). In all cases, the relaxation activity of the wild-type topoisomerase was more than 100-fold higher than that of mutant enzymes. The same result was obtained by measuring the topoisomerase 11-catalyzed unknotting of knotted phage P4 DNA rings (data not shown). Both wildtype and mutant enzyme preparations contained a low level of a contaminating endonuclease, and after prolonged incubation there was an increase of nicked DNA (Fig. 5).
In contrast to the elimination of the DNA transport activity, replacing Gly-144 by isoleucine, valine, or proline does not appear to abolish the DNA breakage and rejoining activity of the type I1 enzyme. Fig. 6 illustrates DNA cleavage by wildtype and mutant yeast DNA topoisomerase I1 in the presence of the eukaryotic type 11 DNA topoisomerase drug etoposide . In the absence of the enzyme, little cleavage of the DNA was observable (lane 1 ). When 500 nM wild-type topoisomerase I1 was included in a reaction mixture containing VP-16, approximately 50% of the plasmid DNA was nicked or linearized (lane 4 ) . The same level of DNA cleavage was found for each of the Gly-144 mutant enzymes (lanes 7, 10,  and 13).
We have also examined the effects of ATP, ADP, and AMPPNP on the cleavage of a DNA oligomer by wild-type and mutant yeast DNA topoisomerase I1 in media containing Ca(I1). It has been reported previously that Ca(I1) favors the trapping of covalent DNA-type I1 topoisomerase covalent complex (Osheroff and Zechiedrich, 1987) and that ATP and AMPPNP stimulate DNA cleavage by eukaryotic DNA topoisomerase I1 (Sander and Hsieh, 1983;Robinson and Osheroff, 1991). As shown in Fig. 7, for either wild-type or G144I mutant yeast DNA topoisomerase 11, replacing Mg(I1) by Ca(I1) stimulates the cleavage of the DNA by several orders of magnitude (compare the patterns in lanes 1 and 2 for the wild-type enzyme and lanes 7 and 8 for the G144I mutant enzyme). In a medium containing a 4 mM concentration each of Ca(I1) and Mg(II), cleavage of the DNA oligomer by either the wild-type or the G144I mutant enzyme was readily observable (lanes 3 and 9). As expected, however, stimulation of yeast DNA topoisomerase 11-mediated cleavage of DNA by ATP, ADP, or AMPPNP was only observed with the wildtype enzyme (lanes 4-6) and not with the G144I mutant enzyme (lanes 10-12). Measurements of protein-mediated retention of DNA to nitrocellulose membranes (Riggs et al., 1970) indicate that the presence of ATP, ADP, or AMPPNP did not significantly alter the fraction of yeast DNA topoisomerase I1 bound to the DNA oligomer under the reaction conditions of the samples shown in Fig. 7 (data not shown).

DISCUSSION
The experiments described above indicate that in the absence of DNA highly purified yeast DNA topoisomerase 11 has a weak but readily detectable ATPase activity. This activity exhibits the classical Michaelis-Menten kinetics, with K,,, and kcat values around 0.3 mM and l/s/dimeric enzyme, respectively, at 30 "C in a pH 7.8 medium containing 150 mM potassium and 6 mM magnesium acetate. Binding the enzyme to DNA greatly enhances its ATPase activity; in the presence FIG. 6. Etoposide (VP-16) enhanced DNA cleavage by wild-type and Gly-144 mutant yeast DNA topoisomerase 11. Electrophoresis in a 1% agarose gel was used to separate the supercoiled DNA substrate from the nicked and linear products. Supercoiled pHC624 (100 nM plasmid), VP-16 (500 pg/ml where indicated), and the indicated concentration of topoisomerase I1 (dimer) were incubated a t 30 "C for 10 min in the standard reaction buffer without bovine serum albumin. Sodium dodecyl sulfate was added to a linal concentration of 0.5%. and the reaction was incubated for an additional 10 min. Proteinase K (200 pg/ml final concentration) and EDTA (10 mM final concentration) were subsequently added, and the reactions were incubated a t 55 "C for 1 h; deproteination was then carried out as described under "Experimental Procedures" prior to loading of the samples for gel electrophoresis.
Wild-type GI 441 1 2 3 4 5 6 7 8 9 1 0 1 1 1 2 FIG. 7. ATP, ADP, and AMPPNP stimulation of DNA cleavage by wild-type but not Gly-144 mutant yeast DNA topoisomerase 11. A 15% nondenaturing polyacrylamide gel was used to separate the substrate DNA oligomer from its faster migrating cleavage products. Only double-stranded DNA cleavage events were detected. Reactions contained 2 p~ DNA (oligonucleotide), 2 p~ topoisomerase (dimer), and either no added nucleotide (lanes 1-3 and 7-9) or 1 mM ATP (lanes 4 and IO), ADP (lanes 5 and II), or AMPPNP (lanes 6 and 12). The reactions were in the standard buffer with either 8 mM calcium acetate (lanes I and 71, 8 mM magnesium acetate (lanes 2 and 8 ) , or 4 mM calcium acetate and 4 mM magnesium acetate (lanes 3-6 and 9-12). Reactions were incubated for 30 min a t 30 "C before stopping with EDTA (25 mM final concentration) and sodium dodecyl sulfate (0.5% final concentration) and digestion with 200 pg/ml of proteinase K a t 55 "C for 1 h. of excess DNA, kc,,, is increased 20-fold to about 20/s/dimeric enzyme. The rate of ATP hydrolysis by yeast DNA topoisomerase I1 in the presence of DNA is no longer represented by the Michaelis-Menten equation; instead, ATP binding to the two ATPase sites in each DNA-bound dimeric enzyme appears to be cooperative.
This homotropic cooperativity ensures the coordination between the two halves of a DNA-bound dimeric enzyme: when ATP concentration is above a couple of tenth mM, essentially all enzyme molecules have either both of the ATPase sites occupied or unoccupied. Because a single mutation changing Gly-144 of the yeast enzyme to isoleucine, valine, or proline greatly diminishes both the ATPase activity in the presence of DNA and the ATPase activity in the absence of DNA, both activities are most likely manifestations of the same ATPase catalytic pockets. The cooperativity between the pair of ATPase sites in a DNA-bound dimeric enzyme, but not in a free enzyme, is probably because of ATPmediated conformational changes in the DNA-enzyme complex. Previously, a similar suggestion was made based on the stimulation of AMPPNP binding by ATP to DNA-bound but not to free E. coli gyrase (Tamura et al., 1992).
The above interpretation poses an apparent dilemma. Using mixed dimeric yeast DNA topoisomerase I1 consisting of one wild-type and one epitopically tagged GI441 mutant polypeptide, we have shown that the binding of AMPPNP to the wild-type polypeptide can induce the same concerted confor-mational change in the entire molecule, whether the enzyme is DNA-bound or not Wang, 1991,1993). This AMPPNP-induced concerted conformational change in the DNA-bound or free enzyme seems to contradict the idea of an ATP-mediated conformational change only in the DNAbound enzyme. It is plausible, however, that this apparent dilemma reflects differences in the rates of the various ATPdependent steps in the presence and absence of DNA: in the absence of DNA, for example, conformational change in the enzyme induced by ATP binding might be too slow relative to ATP-hydrolysis, in which case no homotropic cooperativity would be observable; on the other hand, with a nonhydrolyzable ATP analogue, even a very slow conformational change can be easily detected.
Not only is the ATPase of yeast DNA topoisomerase I1 inactivated by mutations changing Gly-144 to isoleucine, valine, or proline, the DNA transport activity of the enzyme is also abolished. These results demonstrate a striking dependence of DNA transport by the eukaryotic type I1 enzyme on ATP, in contrast to the ATPase activity of the enzyme, which, although enhanced by the binding of the enzyme to DNA, does not appear to require the transport of DNA through the enzyme-operated DNA gates. According to the protein clamp model described in the Introduction, a key step in a type I1 DNA topoisomerase-catalyzed DNA transport is the ATP binding-triggered closure of the DNA-bound clamp for the capture of a second DNA segment, termed the T-segment; conformational changes in the protein clamp in its closed state promote the transport of the T-segment through the DNA gate operated by the same enzyme molecule in the other DNA segment, termed the G-segment (Roca and Wang, 1992).
Without the conformational changes triggered by ATP binding, enzyme-catalyzed transport of the T-segment through the G-segment is too slow to be detectable, as indicated by our data for the three Gly-144 mutants. Further insights on the mechanism of coupling DNA transport to ATP hydrolysis by yeast DNA topoisomerase I1 are gained from the ratio of the rates of these processes measured under identical conditions. At 30 "C and low ATP concentrations, 1.9 f 0.5 ATP are hydrolyzed/DNA transport event. It is likely that under these conditions, the rate-limiting step in ATP usage is slower than the rate-limiting step in the capture and transport of the T-segment through the DNA gate; thus the efficiency of coupling is relatively high in terms of ATP consumption. Based on our kinetic data for the DNA-dependent ATPase activity, at the low ATP concentrations of 5-25 p~, a significant fraction of the dimeric enzyme has only one bound ATP at any given time. The observation that there is a concerted conformational change of the entire enzyme as a result of the binding of one nonhydrolyzable ATP analogue suggests that the binding of a single ATP to a dimeric enzyme might be sufficient for DNA transport Wang, 1991, 1993). Thus when ATP usage and DNA transport are very efficiently coupled, at a low ATP concentration the expected value of n should be between 1 and 2, which is to be compared with the experimental finding of n = 1.9 & 0.5. At a high ATP concentration condition, homotropic cooperativity in the dimeric enzyme assures a coordinated action of the two halves of the enzyme. Nevertheless, at a saturating ATP concentration and low temperature the measured n is Topoisomerase II 7.4 +. 1.0. We attribute this much higher value of n to a reversal of the relative magnitude of the two rate-limiting steps under the new set of conditions; namely, in the presence of a saturating amount of ATP the rate-limiting step in ATP usage is faster than that in the capture and transport of DNA. In terms of the ATP-modulated protein clamp model, there are two plausible scenarios for the high value of n. In one, the clamp opens and closes rapidly but only one out of several times is a T-segment captured; once captured, the T-segment is transported efficiently through the DNA gate in the enzyme-bound G-segment. In the other scenario, the clamp is very efficient in trapping the T-segment at high as well as low ATP concentrations, but the catch does not get transported efficiently at a high ATP concentration. We favor the first scenario because of its conceptual simplicity and because it can encompass the low as well as the high ATP concentration results.