DNA Topoisomerase I1 from Drosophila melanogaster RELAXATION OF SUPERCOILED DNA*

In order to study the double-strand DNA passage reaction of eukaryotic type I1 topoisomerases, a quantitative assay to monitor the enzymic conversion of supercoiled circular DNA to relaxed circular DNA was developed. Under conditions of maximal activity, relaxation catalyzed by the Drosophila melanogaster and the energy of activation was 14.3 kcal. mol”. Removal of supercoils was accompanied by the hydrolysis of either ATP or dATP to inorganic phosphate and the corresponding nucleoside diphosphate. Apparent K,,, values were 200 p~ for pBR322 plasmid DNA, 140 I.~M for SV40 viral DNA, 280 NM for ATP, and 630 PM for dATP. The turnover number for the Drosophila enzyme was at 200 supercoils of DNA relaxed/min/molecule of topoisomerase 11. The enzyme interacts preferentially with negatively supercoiled DNA over relaxed molecules, is capable of removing positive superhelical twists, and was found to be strongly inhibited by sin-gle-stranded DNA. Kinetic and inhibition studies indicated that the B and y phosphate groups, the 2’-OH of the ribose sugar, and the Ce-NH2 of the adenine ring are important for the interaction ATP the enzyme. topoisomerase I1 a DNA strand event, hydrolysis required for enzyme turnover. The ATPase activity of the topoisomerase 17-fold the presence of negatively supercoiled 4 were hydrolyzed/supercoil

In order to study the double-strand DNA passage reaction of eukaryotic type I1 topoisomerases, a quantitative assay to monitor the enzymic conversion of supercoiled circular DNA to relaxed circular DNA was developed. Under conditions of maximal activity, relaxation catalyzed by the Drosophila melanogaster topoisomerase I1 was processive and the energy of activation was 14.3 kcal. mol". Removal of supercoils was accompanied by the hydrolysis of either ATP or dATP to inorganic phosphate and the corresponding nucleoside diphosphate. Apparent K,,, values were 200 p~ for pBR322 plasmid DNA, 140 I .~M for SV40 viral DNA, 280 NM for ATP, and 630 PM for dATP. The turnover number for the Drosophila enzyme was at least 200 supercoils of DNA relaxed/min/molecule of topoisomerase 11. The enzyme interacts preferentially with negatively supercoiled DNA over relaxed molecules, is capable of removing positive superhelical twists, and was found to be strongly inhibited by single-stranded DNA. Kinetic and inhibition studies indicated that the B and y phosphate groups, the 2'-OH of the ribose sugar, and the Ce-NH2 of the adenine ring are important for the interaction of ATP with the enzyme. While the binding of ATP to Drosophila topoisomerase I1 was sufficient to induce a DNA strand passage event, hydrolysis was required for enzyme turnover. The ATPase activity of the topoisomerase was stimulated 17-fold by the presence of negatively supercoiled DNA and approximately 4 molecules of ATP were hydrolyzed/supercoil removed. Finally, a kinetic model describing the switch from a processive to a distributive relaxation reaction is presented.
Type I1 topoisomerases are enzymes which catalyze changes in the topological structure of DNA by the generation of transient double-stranded breaks in the DNA backbone (1-6). Activity has been detected in many diverse sources. Among these are bacteria (7-11), bacteriophages (12, 13), yeast (14), insects (15), amphibians (16), and mammals (17-21). Since the first type I1 enzyme, DNA gyrase, was isolated from Escherichia coli in 1976 (7), a large body of information * This work was supported by National Institutes of Health Grant GM-28079. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. concerning the properties, activities, and functions of prokaryotic type I1 topoisomerases has accumulated (1-6). Unfortunately, comparatively little is known about the eukaryotic proteins.
Although the topological state of DNA in bacteria has been shown to affect replication, repair, recombination, and transcription (4-6, 22), the role of DNA topology in eukaryotic species is not as yet understood. Clearly, before the cellular functions of eukaryotic type I1 topoisomerases can be properly determined, their in vitro activities must be well characterized. The recent purification of these enzymes from eukaryotic species (14,20,23) has made such characterizations possible. Fundamental to all the in vitro interactions which have been ascribed to type I1 topoisomerases is their ability to resolve interlocking constraints of DNA by the double-strand passage mechanism (1-6). The most straightforward reaction which can be employed to monitor this central function is the interconversion between superhelical and relaxed forms of circular DNA. This paper describes the development of a quantitative assay which has been used to follow the ATPdependent relaxation of supercoiled circular DNA by topoisomerase I1 from the fly, Drosophila rnelanogaster. Reaction parameters, kinetic constants, requirements for the interaction of DNA and ATP with the enzyme, and the relationship between ATP hydrolysis and DNA relaxation are discussed. A kinetic model describing the processive versus distributive nature of the reaction is also presented. Results are compared with those obtained for the supercoiling-relaxation and ATPase reactions of E. coli DNA gyrase (24, 25), the prototypical type I1 enzyme found in bacteria.
Relaxation of Supercoiled DNA by Topoisomerase 11 9537 NJ), imidodiphosphate was from R. G. Yount (Washington State University), and 6174 single-stranded bacteriophage DNA was from J. Kaguni (Stanford University). M13-KB H o n~ single-stranded DNA, which is a derivative of the bacteriophage M13 that contains a 740base pair insert of human mitochondrial DNA, including the heavy strand origin of replication (29), was the gift of D. P. Tapper (Stanford University). pAGl plasmid DNA, which is a deleted form of the bacterial plasmid pBR322 containing DNA from the PuuII site to the EcoRI site,' was provided by W. A. Segraves (Stanford University). All other chemicals were analytical reagent grade.
Definition of Enzymic Actiuity-One unit of topoisomerase I1 activity is defined as the minimal amount of enzyme required to fully relax 0.3 pg (0.5 nmol base pairs of DNA) of supercoiled pBR322 plasmid DNA in 15 min at 30 "C in 20 pl of 10 mM Tris-C1, pH 7.9, 50 mM NaCl, 50 mM KCl, 5 mM MgC12, 0.1 mM EDTA, 15 pg/ml of bovine serum albumin, and 1 mM ATP.
Kinetics ATP and dATP concentration were studied over a range of 0.1 to 1.5 mM at a constant concentration of 4.6 X M DNA (0.6 pg/ reaction). The data were analyzed by Eadie-Hofstee single reciprocal plots (30).
The relaxation of positively supercoiled DNA was examined by employing fully relaxed circular pBR322 plasmid DNA and adding ethidium bromide (2 pg/ml final concentration) to the reaction mixture (31). All other conditions are as described above except that prior to electrophoresis on agarose gels, the ethidium bromide was extracted with phenol.
The effects of ionic strength, magnesium concentration, temperature and pH on the kinetics of relaxation were measured under standard assay conditions changing only the indicated variable. For ionic strength measurements, the sodium:potassium ratio was always 1:l. When buffers other than Tris were employed for pH studies, their concentrations were always 10 mM. Thermal stability studies were carried out by incubating topoisomerase I1 in the standard reaction mixture minus DNA at the appropriate temperature. DNA was then added and activity was determined by assaying at 30 "C for 6 min. Inhibition studies employed at least two different concentrations of each inhibitor and results were analyzed by Eadie-Hofstee plots (30).
Relaxation of Negatively Supercoiled Circular DNA with Stoichiometric Amounts of Topoisomerase ZZ-Unless otherwise noted, the Segraves, W. A., personal communication.
standard reaction mixture contained 0.1 pmol (0.3 pg) of supercoiled circular pBR322 plasmid DNA, 0.5 to 4.0 pmol of Drosophila topoisomerase I1 (35 to 280 units), and 1 mM APP(NH)P in a total of 20 p1 of relaxation buffer. Samples were incubated at 30 "C for 6 min, followed by the addition of 3 pl of 0.77% sodium dodecyl sulfate, 77 mM EDTA. Proteinase K (2 pl of a 625 pg/ml solution in 50 mM Tris-C1, pH 7.9, 1 mM CaCI2) was added and digestion was at 37 "C for 20 min. Samples were mixed with 3 p1 of 60% sucrose, 0.05% bromphenol blue, 0.05% xylene cylanol FF and heated at 70 "C for 2 min. Reaction products were resolved by electrophoresis on agarose gels as described above. When cofactors other than APP(NH)P were employed, their concentrations were at least 2-fold and usually 10fold higher than their K, values (as reported in Table 11).
ATPase Activity of Topoisomerase ZZ-Prior to assay, topoisomerase I1 was freshly diluted to a concentration of 5 units/pl in the dilution buffer described above. The standard reaction mixture contained 5 units of enzyme, 15 pg of DNA (1.15 X M base pairs of double-stranded or 2.30 X M bases of single-stranded molecules), and 1 mM [Y-~'P]ATP (0.02 Ci/mmol) in a total of 20 pl of relaxation buffer. Reaction was at 30 "C. Samples (2.5 p l ) were removed at 2-or 4-min intervals up to 16 min, spotted onto thin layer cellulose plates impregnated with polyethyleneimine (Polygram CEL 300 PEI, Brinkmann) and chromatographed in freshly made 0.4 M NH,HC03. ATP, ADP, and AMP standards were visualized by fluorescence, while 32Pi standards and reaction products were visualized by autoradiography with Kodak XAR film. Radioactive areas corresponding to reaction products were cut out of the chromatograms and the rate of ATP hydrolysis was quantitated employing a Beckman LS-230 liquid scintillation counter and toluene-based scintillation fluid. When necessary, the corresponding rate of relaxation of supercoiled DNA was determined as described above.

RESULTS
The time course for the relaxation of 0.3 pg of negatively supercoiled circular pBR322 plasmid DNA (2.3 X M base pairs) by 1 unit of D. rnelanogaster topoisomerase I1 is shown in Fig. 1. The velocity is linear for the first 8 min of the reaction, and for at least that long at higher DNA concentrations. All subsequent initial velocities were calculated for time points up to 6 min, well within the linear range for every condition examined. The fact that as much as 75% of the supercoiled DNA is relaxed during the linear period indicates t h a t the enzyme interacts more strongly with its substrate, supercoiled DNA, than with relaxed DNA, which is the product of the reaction. Under optimal conditions, the relaxation reaction is extremely processive. As can be seen in the agarose gel presented in Fig. 1, substrate and final product distributions predominate and virtually no reaction intermediates are visible. In contrast, distributive reactions where DNA intermediates are prevalent were observed only under suboptimal conditions (see below). Thus, under the assay conditions employed, one measurable event is the complete relaxation of the bacterial plasmid, pBR322. This corresonds to the removal of approximately 30 supercoils/molecule of DNA (12).

Requirements for Relaxation
Ionic Strength Dependence-The ionic strength dependence of the topoisomerase I1 reaction is shown in Fig. 2  the ATP should be complexed (32), but the enzymic activity is only 50% that of the optimal rate. Over the concentration range examined (2.5 to 25 mM), manganese and calcium showed no activity.
pH Dependence-No relaxation of DNA was observed below pH 6.5 in acetate, Pipes, or Tris buffers. In Tris, activity increased rapidly between pH 6.5 and 7.5, and plateaued up to pH 10, the highest pH examined (Fig. 2C). No distributive reaction was observed. The activity profile in glycine buffer was much different, showing no activity below pH 8 or above pH 10.5, with a sharp maximum at pH 9.5. Relaxation became distributive at pH 10.
Temperature Dependence and Thermal Stability-Maximal rates of activity were obtained a t 30 "C (Fig. 2 0 ) and the reaction remained processive over the entire active temperature range. The subsequent Arrhenius plot (30) (not shown) was linear from 15 to 30 "C and indicated an energy of activation for the topoisomerase I1 relaxation reaction of 14.3 kcal. mol-'. The thermal stability of the Drosophila enzyme in the absence of DNA was studied at temperatures ranging from 15 to 45 "C (Fig. 3). The energy for the thermal inactivation of the topoisomerase I1 is 13.4 kcal.mo1-I. Over the course of a 6-min incubation a t 30 "C (standard assay conditions), less than 4% of enzymic activity was lost. The low rates of relaxation obtained a t elevated temperatures cannot be entirely explained by thermal inactivation. Indeed, a t 35 and 40 "C, enzyme activity dropped by 30 and 80%, respectively, (Fig.  2D), as compared to the maximal reaction rate obtained at 30 "C, while the thermal stability study (Fig. 3) indicated that a t these two temperatures, only 6 and 12% of the topoisomerase I1 should have been inactivated over the course of a 6min assay. Therefore, either the lowered rates of relaxation a t high temperatures are due to a change in the catalytic mechanism or thermal inactivation is greatly exacerbated by the enzymic reaction. Effect of DNA Concentration-The kinetics of relaxation of negatively supercoiled circular DNA by the Drosophila topoisomerase I1 are shown in Fig. 4. The apparent K,,, for the bacterial plasmid pBR322 was 2.0 X M and the V,,, was approximately 2 x M base pairs of supercoiled DNA relaxed/min/unit of enzyme. This corresponds to a turnover number of about 200 supercoils relaxed/min/molecule of topoisomerase 11. As this number is based on the assumption that every molecule of topoisomerase I1 is active, it must be considered a lower limit for the activity of the enzyme.
Under standard assay conditions, with 4.6 x M base pairs of pBR322 plasmid DNA and I unit of topoisomerase 11, the enzyme:plasmid ratio is a t least 1:15, well within the limits of the steady state assumption (i.e. enzyme <<substrate) (30). In addition, Michaelis-Menten kinetics were observed at ratios as low as 1:7. However, at an enzyme:plasmid proportion of about 1:4, the kinetics of relaxation diverged from linearity and exhibited initial velocities approximately 2-fold higher than predicted.
The relaxation of negatively supercoiled circular DNA from the eukaryotic virus SV40 by the D. melanogaster type I1 topoisomerase was also examined (Fig. 4). The enzyme showed a somewhat higher affinity for the viral DNA (apparent K,,, = 1.4 X M) as compared to that for the prokaryotic pBR322 plasmid DNA. Whether this results from an interaction between the enzyme and a specific site on the eukaryotic DNA, or is simply an effect of general base composition is unknown. However, in regard to the latter, SV40 DNA has a much higher A-T content (59.2%) (33, 34) than pBR322 DNA (46.3%) (35).
Effect of Nucleoside Triphosphate Concentration-The ATP and dATP dependence of the kinetics of relaxation of negatively supercoiled DNA is presented in Fig. 5. A low DNA concentration was employed for this study in order to overcome the difficulties of quantitating the small percentage of total DNA which is relaxed under conditions of high DNA and low ATP concentrations. This accounts for the relatively low maximal velocities which are observed. Although the Drosophila enzyme is able to utilize either ATP or dATP as substrate (15) value of 2.8 X M as compared to 6.3 X M. Thus, the 2'-OH of the ribose ring imparts some specificity to the interaction of the high energy cofactor with the enzyme. Clearly, the absence of the 2'-OH does not affect the rate of hydrolysis of the bound nucleotide, as the kinetic plots for both ATP and dATP extrapolate to the same V,,,,, value. Substrate inhibition was observed for both the ribo-and deoxyribonucleotides above 1.5 mM. In the absence of ATP or dATP, no relaxation was observed even with a 300-fold increase in enzyme concentration. The purified topoisomerase could not be activated with any other nucleoside triphosphate tested, including GTP, CTP, UTP, ITP, XTP, dGTP, dCTP, and dTTP. Moreover, no catalytic activity was found with analogues of ATP that have nonhydrolyzable 0-7 phosphonate bonds or improper conformations about their @ and/or y phosphate groups, such as APP(NH)P, ATPyS, APP(CH,)P, and AP(CH,)PP. This is despite the fact that many of these nucleotides interact strongly with the ATP binding site of the Drosophila enzyme (see Table 11).
Effect of DNA Length-To ascertain the effect of DNA length on the relaxation reaction, two deleted forms of the bacterial plasmid pBR322 (35), pBR327 (36), and pAG1,'were employed. It should be noted that at any given DNA concentration, irrespective of plasmid size, the total number of supercoils and DNA base pairs present in the assay mixture was always constant. However, decreases in plasmid length resulted in corresponding increases in the number of molecules present.
Three major observations resulted from this study ( Fig. 6 and Table I). First, within the homologous series employed, plasmid size had no effect on the apparent K,,, of the relaxation reaction. Second, the maximal velocity of the reaction was found to be directly proportional to the length of the DNA being assayed. Finally, the turnover number of the D. melanogaster topoisomerase 11, expressed as the number of plasmid molecules relaxed/min/molecule of enzyme, remained constant, despite the fact that the smaller DNAs have correspondingly fewer supercoils/molecule. Therefore, when the enzyme operates processively, the velocity of relaxation must  "Expressed as the number of plasmid molecules relaxed/min/ molecule of topoisomerase 11. Turnover number = Vma,/(plasmid length in base pairs. enzyme concentration). be related to the rate of dissociation of the topoisomerase 11. relaxed DNA complex. Since smaller plasmids obviously contain fewer base pairs of DNA/molecule, this accounts for the observed proportional decrease in V,,,,, values.
Relaxation of Positively Supercoiled DNA-In the presence of ethidium bromide, fully relaxed circular DNA can be induced to form positive superhelical turns (31). If this positively supercoiled DNA is subsequently relaxed with a topoisomerase and the ethidium bromide is removed, the resulting structure will show a net gain of negative supercoils. This reaction is reflected by an increased electrophoretic mobility of the DNA when applied to agarose gels. By employing the strategy described above, the Drosophila topoisomerase I1 was found to relax positive supercoils (not shown). Moreover, at low concentrations of DNA, the initial velocities of relaxation were comparable to those found for negatively supercoiled molecules. Unfortunately, the highest DNA concentration which could be employed for this study was 6.0 X M, as increased levels required the inclusion of prohibitively high amounts of ethidium bromide. Table I1 lists the K, values for a number of inhibitors of the Drosophila type I1 topoisomerase. The site of inhibition was determined by the ability of the compound to compete with either ATP or DNA as judged by Eadie-Hofstee analysis (30).

Inhibition of Relaxation
Coumermycin A, and novobicin, both potent inhibitors of the ATPase reaction of the prokaryotic type I1 topoisomerase, DNA gyrase (24, 37), inhibited the Drosophila topoisomerase 11, but at concentrations 3 to 4 orders of magnitude higher (Table 11). Oxolinic acid and nalidixic acid, which inhibit the interaction of DNA with gyrase (38-40), also affect the eukaryotic enzyme, but once again much higher levels are required. It was originally reported (15) that the Drosophila enzyme was insensitive to nalidixic acid. However, in that study the maximal concentration of inhibitor employed was less than one-third of the determined Ki value, making inhibition difficult to detect.
Relaxation of DNA was strongly inhibited by micromolar levels of the ATP analogues ATPyS and APP(NH)P (Table  11), both of which have conformations about their P-7 phosphonate bonds which are nearly identical to that of ATP (41, 42). Inhibition by ADP was approximately one order of magnitude less efficient and no inhibition was observed with AMP, adenosine, or adenine. These results indicate that the majority of binding interactions between ATP and the enzyme involve the @ and y phosphate groups of the nucleotide, with little or no contribution from the a phosphate. This was supported by the finding that PPPi had a Ki value which was comparable to those of PPi and imidodiphosphate. Steric requirements for the interaction of phosphate groups with the nucleoside triphosphate binding site of Drosophila topoisomerase I1 were analyzed by using the P-y and a-/3 methylene analogues of ATP as inhibitors of the relaxation reaction (Table 11). The spatial arrangements of the carbonphosphate bonds in these derivatives differ from those of the corresponding phosphonate linkages in ATP, since the bond angle of P-C-P (117") is more acute than that of P-0-P (129") (42). Thus, the K, of APP(CH,)P, which has an altered conformation about the 6-7 bond, was similar to that of ADP and the Kt for AP(CH,)PP, in which the spatial orientations of both the P and y phosphates are affected by the a-p methylene bond, was an order of magnitude higher.
The specificity for ATP binding must reside in its nucleoside moiety, since APP(NH)P and ATPyS inhibit relaxation two orders of magnitude more effectively than PPPi (Table   11) and only ATP and dATP can be functionally employed by the enzyme (Fig. 5 ) . By examining the inhibitory properties of a number of purine and pyrimidine ribonucleoside triphosphates (Table 11), it was determined that the Cs amino group based on the following two findings. 1) ITP, which is identical to ATP except for the repalacement of the c6 amino on the purine ring with a ketone, had a Ki value of 360 p~, approximately 40-fold higher than those observed for APP(NH)P and ATPyS. 2) CTP, which has an amino group at the C4 position of tbe pyrimidine ring (corresponding to the c6 position of purine rings), had a Ki value of 100 pm, 15 times lower than that for UTP, which carries a C, ketone. The relaxation of supercoiled molecules 'is strongly inhibited by single-stranded DNA (Table 11). Nucleic acids from bacteriophage 6x174 and M13-KB HoriH (29) (a derivative of bacteriophage M13 which contains a cloned 740-base pair fragment of human mitochondrial DNA including the heavy strand origin of replication) were utilized for this study. These single-stranded DNAs both competed with supercoiled double-stranded molecules for the same site on the Drosophila enzyme. Moreover, they were the most potent inhibitors found, exhibiting Ki values of 0.8 and 0.9 p~, respectively.
Poly-L-glutamic acid did not inhibit the relaxation reaction a t concentrations as high as 3 mM. Thus, the DNA binding site on topoisomerase I1 requires more than a polyanion for proper interaction.

Stoichiometric Relaration of Supercoiled DNA
As demonstrated above and in the following section, both the binding and hydrolysis of ATP by the D. melanogaster topoisomerase I1 are necessary for the catalytic conversion of supercoiled DNA to relaxed structures (Figs. 5 and 8, Tables I1 and 111). It has been shown for E. coli DNA gyrase (24) and bacteriophage T 4 topoisomerase I1 (12) that ATP binding alone is sufficient to induce a strand passage event, while hydrolysis of the cofactor is required for enzyme turnover. T o see if this was also the case for the Drosophila enzyme, the relaxation of negatively supercoiled pBR322 plasmid DNA by stoichiometric amounts of topoisomerase I1 was examined.
Since pBR322 DNA contains approximately 30 supercoils/ molecule (12) and type I1 topoisomersases remove 2 supercoils/strand passage, if every DNA-bound enzyme catalyzes a single event, approximately 15 molecules of enzyme would be required to completely relax 1 molecule of DNA. It should be noted that native Drosphila topoisomerase I1 is a homodimer (23).
As can be seen in Fig. 7, in the absence of a nucleoside triphosphate cofactor, no relaxation was observed, even at an enzyme:plasmid ratio of 201. Moreover, this ratio could be increased to 40:l (not shown) with no effect on the topological state of the DNA. However, when APP(NH)P was present, relaxation was evident a t ratios as low as 7.5:l (Fig. 7). Similar   results were found with ATPyS. Thus, even in the absence of hydrolysis, the binding of ATP to the Drosophila enzyme is sufficient to induce a DNA strand passage event. Moreover, the presence of the adenine ring, the y-phosphate group, and the correct spatial arrangement about the P-y phosphonate bond are all required for strand passage, as APP(CH,)P, AP(CH,)PP, ITP, CTP, ADP, and PPPi were not functional as cofactors.
ATPase Activity of Topoisomerase I1 Topoisomerase I1 hydrolyzes ATP to ADP and Pi as determined from reactions containing ATP labeled with '*P a t either the P or y pbosphate. No other reaction products could be detected, even during prolonged incubation times (up to 1 h). As can be seen in Table 111, the ATPase activity.of the Drosophila enzyme was DNA-dependent and was stimulated 17-fold by the presence of negatively supercoiled plasmid DNA. Moreover, the topoisomerase was sensitive to the topological state of the DNA substrate, as supercoiled DNA yielded a reaction velocity which was 4-fold higher than generated by either relaxed circular or linear molecules (Table  111). Together with the kinetic data discussed earlier (Fig. l ) , this strongly suggests that the affinity of the enzyme for its supercoiled substrate is higher than for its relaxed product. Although single-stranded DNA interacts very tightly with the DNA site on topoisomerase I1 (Table II), it serves as a poor

9542
Relaxation of Supercoiled DNA by Topoisomerase I1 substrate for the ATPase reaction (Table  111). Therefore, either double-stranded DNA is necessary to promote hydrolysis of ATP, or the single-stranded substrate binds strongly but incorrectly to the enzyme. Despite the fact that CTP binds tightly to the ATP site on the topoisomerase, it is not hydrolyzed by the enzyme. This was demonstrated by substituting [~u-~'P]CTP for ATP in the ATPase assay.
The effects of coumermycin A, and oxolinic acid on the ATPase reaction were examined. A concentration of 20 pm coumermycin A, was found to inhibit the reaction by 50%. This value is similar to the K, (10 pm) determined for the drug-induced inhibition of the relaxation reaction (Table 11). Although oxolinic acid inhibits relaxation with a K, of 1.3 mM ( Table 11), levels of 2 mM had almost no effect on the ATPase activity of the Drosophila enzyme. A similar resistance was previously found with DNA gyrase (25). Thus, oxolinic acid appears to be a preferential inhibitor of relaxation over ATPase activity.
Since the catalytic relaxation of supercoiled DNA by Drosophila topoisomerase I1 is coupled to the hydrolysis of ATP (Fig. 5, Tables I1 and 111) (15), the ratio of ATP molecules hydrolyzed for every supercoil removed was determined (Fig.  8). It was found that approximately 4.1 molecules of ATP are hydrolyzed/supercoil relaxed. A different preparation of enzyme yielded 3.8 ATP molecules/supercoil. These values are higher than those obtained for the supercoiling reaction of DNA gyrase (-0.4 ATP hydrolyzed/supercoil induced) (25) and the relaxation reaction of bacteriophage T4 topoisomerase I1 (-1 ATP hydrolyzed/supercoil removed) (12). However, since the Drosophila enzyme acts processively and can turn over ATP molecules even on fully relaxed DNA substrates, the calculated ATP/supercoil ratio may be inflated by hydrolysis occurring in the complex between the topoisomerase and relaxed DNA.

Processive versus Distributive Nature of the Relaxation
Reaction Under conditions which yield optimal rates (Fig. a), the enzymic conversion of negatively supercoiled circular DNA to fully relaxed circular DNA by D. melanogaster topoisomerase I1 proceeds in a processive manner. Thus, the enzyme forms a complex with the supercoiled DNA, catalyzes several successive rounds of relaxation, and does not dissociate until the DNA has been completely relaxed. The kinetic scheme depicting this reaction is shown in Fig. 9. Therefore, under processive conditions, for any enzyme-bound DNA intermediate, the rate of relaxation (krel) must be considerably faster than the rate of dissociation of the complex ( k d . This leads to two conclusions. First, since the topoisomerase is present in catalytic amounts and must dissociate from fully relaxed molecules in order to interact with other supercoiled molecules, the velocity of the relaxation reaction must be related to the rate of dissociation of the E. R complex (see Fig. 6 and Table I). Second, the processive versus distributive nature of the relaxation reaction is controlled by the term, krel,koff, which is the ratio of the rate of relaxation to the rate of dissociation for any given enzyme-bound DNA intermediate. When this ratio is high, relaxation takes precedence over dissociation and the reaction is processive. However, conditions which decrease krel and/or increase kOff such that their ratio approaches unity, lead to a distributive reaction in which the enzyme catalyzes only a few or as little as one round of relaxation before dissociating from the DNA.
Three different conditions have been found to promote a E + S-2 E+ R+2 FIG. 9. Kinetic scheme for the enzymic conversion of negatively supercoiled circular DNA ( S ) to fully relaxed circular DNA (R) by topoisomerase I1 (E). As one relaxation event results in the removal of 2 supercoils, DNA intermediates are depicted as S-2 (fully supercoiled DNA minus 2 supercoils) to R+, (fully relaxed DNA plus 2 supercoils). The kinetic constant k ,~ represents the rate of relaxation for any given intermediate. K represents the dissociation constant for any given enzyme. DNA complex and is equivalent to the term koff/ko.. distributive relaxation reaction. They are high ionic strength (2175 mM), high magnesium concentration (215 mM), and high pH (210) in glycine buffer. Although it has not been directly demonstrated, two lines of evidence imply that increased ionic strength acts by increasing kOff. I) When the ionic strength dependence of the relaxation reaction was analyzed by a modified form of the Brensted equation (43) (not shown), it was found that electrostatic interactions are important for the binding of D. melanogaster topoisomerase I1 to DNA. Therefore, increased levels of salt should decrease binding between the two (i.e. decrease k,, and increase koff) by shielding charged groups. 2) Direct measurements have shown that increased levels of salt decrease the binding between DNA and the type I topoisomerase from rat liver (44), an interaction which can also be shown to depend on the involvement of electrostatically charged groups (43). Whether the effects of magnesium concentration and pH in glycine buffer manifest themselves through changes in k,.~ or koff remains to be determined.  (24). Second, gyrase has a far more stringent requirement for ATP over dATP. While the absence of the 2'-OH on the ribose ring results in a 2-fold increase in the apparent K, for the interaction of the nucleotide with the eukaryotic enzyme (Fig. 5), it raises the apparent K,,, for the prokaryotic enzyme by a factor of at least 30 (7, 24,25). This demarcation between eukaryotic and prokaryotic type I1 topoisomerases also extends to other species. Whereas dATP is reported to be fully capable of substituting for ATP with the protein from human HeLa cell nuclei (20), it cannot function as a cofactor for the bacteriophage T4 enzyme (12). Third, the nucleoside triphosphate site of DNA gyrase has a much stricter requirement for the adenine base than does the Drosophila enzyme. Although the ATPase reaction of gyrase is unaffected by the addition of CTP, UTP, or G T P (25), the relaxation reaction of Drosophila topoisomerase I1 is effectively inhibited by CTP (Table 11). Therefore, while the Cs amino group of adenine seems to be responsible for a large degree of the specificity for the nucleotide site of D. melanogaster topoisomerase 11, additional or different interactions are necessary to confer specificity in gyrase. Fourth, whereas the a, / 3, and y phosphates of ATP all contribute to binding with gyrase, only the fi and y phosphates appear to interact with the eukaryotic enzyme. This follows from the demonstration that ATP analogues, ADP, and AMP all inhibit reactions of DNA gyrase (24, 25) while only ATP analogues and ADP affect the D. melanogaster topoisomerase I1 (Table 11).
Interaction with DNA-A major difference between the DNA sites of E. coli gyrase and Drosophila topoisomerase I1 is reflected by their sensitivity toward single-stranded molecules. Whereas single-stranded DNA is a potent inhibitor of the topoisomerase I1 relaxation reaction (Table 11), it has no effect on the supercoiling reaction of gyrase (25, 39). Despite the fact that the Drosophila enzyme is able to accomodate single-stranded molecules, such structures do not appear to be intermediates in the relaxation process. This was concluded from experiments demonstrating that: 1) the rate of relaxation of positively supercoiled DNA, which is overwound, is comparable to that of undenvound, negatively supercoiled DNA and 2) single-stranded DNA is a poor activator of ATPase activity (Table 111), even at concentrations 2500 times higher than its K, value (Table 11).
Finally, both DNA gyrase and the Drosophila topoisomerase I1 are able to distinguish between different topological isomers of DNA. Moreover, both appear to interact preferentially with the form of DNA which serves as substrate for their supercoiling-relaxation reactions. Thus, DNA gyrase, which induces supercoils in relaxed DNA, has a 4-fold higher kinetic and binding affinity for relaxed, nicked, or linear DNAs over supercoiled forms (25). Conversely, the Drosophila enzyme, which relaxes supercoiled DNA, shows an increased specificity for supercoiled DNA over relaxed or linear molecules. This was deduced from the unusually long linear phase of the time course for relaxation ( Fig. 1) and the increased velocity of the ATPase reaction observed with supercoiled DNA (Table 111).
Since the type I enzyme from chicken erythrocytes appears to show the same affinity for both supercoiled and relaxed molecules (45), the ability to discern the topological state of DNA may be unique to type I1 topoisomerases.