Geometry of DNA Supercoils Modulates Topoisomerase-mediated DNA Cleavage and Enzyme Response to Anticancer Drugs^†

Enzymes and Materials

Human topoisomerase IIα and topoisomerase IIβ were expressed in Saccharomyces cerevisiae (48) and purified as described previously (49). Human topoisomerase I was a gift from Dr. Mary Ann Bjornsti (St. Jude Children’s Research Hospital). Negatively supercoiled pBR322 DNA was prepared from Eschericia coli using a Plasmid Mega Kit (Qiagen) as described by the manufacturer.

Positively supercoiled pBR322 DNA was prepared by incubating negatively supercoiled plasmids with Archaeoglobus fulgidus reverse gyrase (50, 51). The average number of superhelical twists present in DNA substrates and the resulting σ values were determined by electrophoretic band counting relative to fully relaxed molecules (51). For negatively supercoiled substrates, time courses for the relaxation of pBR322 by topoisomerase I were resolved by electrophoresis in 1% agarose gels containing 1-2 μg/ml chloroquine (Sigma) in the running buffer. The initial plasmid contained ∼15 to 17 negative superhelical twists per molecule (σ ≈ -0.035 to -0.039). This superhelical density is typical of plasmids isolated from E. coli. For positively supercoiled substrates, time courses for the generation of positive superhelical twists by reverse gyrase were resolved by electrophoresis as above in running buffer containing 5-15 μg/ml netropsin B (Boehringer Mannheim). These plasmids contained ∼15 to 17 positive superhelical twists per molecule (σ ≈ +0.035 to +0.039). The handedness of positively supercoiled DNA was confirmed by two-dimensional gel electrophoresis (51). Thus, the supercoiled substrates employed for this study contained equivalent numbers of superhelical twists, but were of opposite handedness.

It should be noted that positively supercoiled plasmids bind less ethidium bromide than negatively supercoiled molecules. To ensure that equal amounts of plasmid were used in all experiments, DNA concentration was assessed by spectrophotmetric analysis and confirmed by ethidium bromide staining of linearized plasmid substrates (see insets, Figures 1, 6, and 8).

[γ-³²P]ATP (∼5000 Ci/mmol) and genistein were obtained from ICN, etoposide and camptothecin were from Sigma, TOP-53 and TAS-103 were gifts from Taiho Pharmaceuticals, amsacrine was a gift from Bristol-Myers Squibb, and CP-115,953 was a gift from Pfizer Global Research. Etoposide, camptothecin, genistein, TOP-53, amsacrine, and CP-115,953 were stored at 4 ° C as 10 or 20 mM stock solutions in 100% DMSO. TAS-103 was stored at 4 ° C as a 10 mM stock solution in water. All other chemicals were analytical reagent grade.

Plasmid DNA Cleavage

DNA cleavage reactions were carried out using the procedure of Fortune and Osheroff (52). Unless stated otherwise, topoisomerase II DNA cleavage assays contained 220 nM human topoisomerase IIα or β and 10 nM negatively or positively supercoiled pBR322 DNA in a total of 20 μL of topoisomerase II cleavage buffer [10 mM Tris-HCl (pH 7.9), 5 mM MgCl₂, 100 mM KCl, 0.1 mM EDTA, and 2.5% glycerol (v/v)]. Assays were carried out in the absence of compound or in the presence of 0-100 μM etoposide, 50 μM genistein or TOP-53, 5 μM CP-115,953, 0-500 μM amsacrine, 0-200 μM TAS-103, or 0-25 μM ethidium bromide. Topoisomerase I DNA cleavage assays contained 11 nM human topoisomerase I and 10 nM negatively or positively supercoiled pBR322 DNA in a total of 20 μL of 50 mM Tris-HCl (pH 7.5), 10 mM MgCl₂, 50 mM KCl, 0.1 mM EDTA, 0.5 mM DTT, 30 μg/ml BSA. Assays were carried out in the presence of 0-25 μM camptothecin. In all cases, reaction mixtures were incubated at 37 °C for 6 min and enzyme-DNA cleavage complexes were trapped by the addition of 2 μL of 5% SDS followed by 1 μL of 375 mM EDTA (pH 8.0). Proteinase K (2 μL of 0.8 mg/mL) was added and samples were incubated at 45 °C for 30 min to digest the enzyme. Samples were mixed with 2 μL of agarose gel loading buffer [60% sucrose in 10 mM Tris-HCl (pH 7.9)], heated at 45 °C for 5 min, and subjected to electrophoresis in 1% agarose gels in 40 mM Tris-acetate (pH 8.3), 2 mM EDTA containing 0.5 μg/mL ethidium bromide. DNA bands were visualized with ultraviolet light and quantified using an Alpha Innotech digital imaging system. DNA cleavage was monitored by the conversion of supercoiled plasmid DNA to linear molecules.

Site-specific DNA Cleavage by Human Topoisomerase IIα

DNA sites cleaved by human topoisomerase IIα in negatively and positively supercoiled DNA were mapped as described previously (51). DNA cleavage mixtures contained 2.2 μM human topoisomerase IIα, 10 nM negatively or positively supercoiled pBR322 DNA in a total of 160 μL of topoisomerase II cleavage buffer. In some cases, the MgCl₂ was replaced with CaCl₂ or reaction mixtures included 100 μM etoposide. Samples were incubated at 37 °C for 6 min and enzyme-DNA cleavage complexes were trapped by the addition of 16 μL of 1% SDS followed by 8 μL of 375 mM EDTA (pH 8.0). Proteinase K (16 μL of 0.8 mg/mL) was added and mixtures were incubated at 45 °C for 30 min to digest the type II enzyme. Reaction products were purified by passage through Qiaquick Spin Columns (Qiagen) as described by the manufacturer. DNA cleavage complexes were linearized by treatment with HindIII. Terminal 5′-phosphates were removed by treatment with calf intestinal alkaline phosphatase and replaced with [³²P]phosphate using T4 polynucleotide kinase and [γ-³²P]ATP. Samples were treated with EcoRI and the singly-end labeled DNA products were purified by passage through a CHROMA SPIN+TE-100 column (Clontech). Loading buffer (40% formamide (w/v), 8.4 mM EDTA, 0.02% bromophenol blue, and 0.02% xylene cyanole FF) was added and samples were subjected to electrophoresis in 6% polyacrylamide sequencing gels. Gels were fixed in 10% methanol/10% acetic acid (v/v) and dried in vacuo. DNA cleavage products were visualized with a BioRad Molecular Imager FX.

The effects of ethidium bromide (0-10 μM) on the cleavage of linear DNA by human topoisomerase IIα was monitored using the above protocol with the exception that a singly-end labeled linear pBR322 fragment (53, 54) was used as the initial substrate. DNA cleavage was monitored by the loss of full-length linear molecules.

DNA Ligation by Human Topoisomerase IIα

DNA ligation assays were carried out by a modification of the procedure of Kingma et al. (55). DNA cleavage/ligation equilibria were established in topoisomerase II cleavage buffer as described above except that MgCl₂ in the reaction buffer was replaced by 5 mM CaCl₂. Assays were carried out in the absence of compound or in the presence of 100 μM etoposide. Topoisomerase II-DNA cleavage complexes were trapped by the addition of EDTA (pH 8.0) to a 6 mM final concentration. NaCl was added to a 500 mM final concentration in order to prevent re-cleavage of the DNA substrate. Ligation was initiated by the addition of MgCl₂ at a 0.25 mM final concentration and terminated at times up to 60 s by the addition of 2 μL of 5% SDS. Samples were processed and analyzed as described for plasmid DNA cleavage. The percent DNA cleavage at time 0 was set to 100% and the rate of ligation was determined by quantifying the loss of cleaved DNA over time.

DNA Intercalation

Intercalation reaction mixtures contained 220 nM topoisomerase IIα and 5 nM pBR322 DNA in a total of 20 μL of 10 mM Tris-HCl (pH 7.9), 0.1 mM EDTA, 175 mM KCl, 5 mM MgCl₂, 1 mM ATP, and 2.5% glycerol (v/v) that contained 0-25 μM ethidium bromide, 0-500 μM amsacrine, or 0-200 μM TAS-103. Mixtures were incubated at 37°C for 6 min, extracted with phenol:chloroform:isoamyl alcohol (25:24:1), and added to 3 μl of 0.77% SDS, 77 mM EDTA (pH 8.0). Samples were mixed with 2 μL of agarose gel loading buffer, heated at 45 °C for 5 min, and subjected to electrophoresis in a 1% agarose gel in 100 mM Tris-borate (pH 8.3), 2 mM EDTA. Gels were stained with 1 μg/mL ethidium bromide, and DNA bands were visualized as described for plasmid DNA cleavage.

The DNA intercalation assay is based on the fact that intercalative agents induce constrained negative supercoils and compensatory unconstrained positive superhelical twists in covalently closed circular DNA. Therefore, as the concentration of an intercalative compound increases, a plasmid that is negatively supercoiled or relaxed (i.e., contains no superhelical twists) appears to become positively supercoiled. Treatment of an intercalated plasmid with topoisomerase IIα removes the unconstrained positive DNA superhelical twists. Subsequent extraction of the compound allows the local drug-induced unwinding to redistribute in a global manner and manifest itself as a net negative supercoiling of the plasmid. Thus, in the presence of an intercalative agent, topoisomerase treatment converts relaxed plasmids to negatively supercoiled molecules (see insets, Figures 4 and 5).

Effects of DNA Supercoil Geometry on Drug-induced DNA Cleavage Mediated by Human Topoisomerase IIα

Collisions with DNA tracking systems are critical for the conversion of transient topoisomerase-DNA cleavage complexes to permanent strand breaks (4, 7, 10, 32, 37). Since the double helix is acutely overwound immediately ahead of tracking systems (3, 6, 41, 43-47), cleavage complexes most likely to produce permanent strand breaks should be formed between topoisomerases and positively supercoiled DNA. Unfortunately, relatively little is known about the interaction between these enzymes and positively supercoiled substrates.

A recent study demonstrated that human topoisomerase IIα, which is involved in replicative processes in vivo, relaxes positive superhelical twists >10-fold faster than it does negative supercoils (51). Furthermore, the enzyme maintains a lower concentration of DNA cleavage complexes with these overwound substrates (51). Levels of DNA cleavage mediated by topoisomerase IIα in the presence of either Mg²⁺ or Ca²⁺ are ∼3- to 4-fold lower with positively supercoiled DNA (Figure 1). The decreased baseline levels of DNA cleavage observed with overwound DNA may reflect, at least in part, somewhat lower binding affinity (∼2-fold) that topoisomerase IIα displays for positively over negatively supercoiled molecules (51). The lower level of cleavage notwithstanding, the site specificity of DNA scission is not affected by the handedness of the substrate (Figure 1, right panel) (51).

Many anticancer drugs kill cells by increasing levels of double-stranded DNA breaks generated by topoisomerase II (4, 7, 10, 32, 37). As a first step towards characterizing the effects of these drugs on enzyme-DNA cleavage complexes formed with positively supercoiled substrates, the ability of etoposide to induce DNA cleavage by human topoisomerase IIα was assessed. Etoposide was chosen for initial experiments because its mechanism of action is well defined (4, 7, 15, 56-59). In addition, the drug is not intercalative in nature and displays little, if any, affinity for DNA (4, 7, 15, 56, 60). Over a range of drug concentrations, DNA cleavage induced by etoposide was ∼3- to 4-fold lower with positively as compared to negatively supercoiled molecules (Figure 1). Moreover, as observed in the absence of drug, the geometry of DNA supercoils did not affect the site specificity of etoposide-induced scission (Figure 1, right panel).

Etoposide increases levels of topoisomerase II-DNA cleavage complexes primarily by inhibiting the ligation of cleaved strands (4, 7, 15, 57, 58). Therefore, the effects of etoposide on the ability of topoisomerase IIα to ligate positively or negatively supercoiled molecules were determined. As seen in Figure 2, the handedness of superhelical twists had little impact on rates of DNA ligation in either the presence or absence of etoposide. Thus, the decreased levels of DNA strand breaks induced by etoposide in positively supercoiled substrates do not result from a decreased inhibition of ligation.

Despite the lower absolute percent DNA cleaved with positively supercoiled substrates, the relative enhancement of scission by etoposide with overwound and underwound substrates was similar. For example, at 50 μM drug, this enhancement was ∼4.5-fold and ∼3.6-fold for positively and negatively supercoiled DNA, respectively (Figures 1 and 3, Table 1). These data yield a “superhelical specificity” for etoposide (i.e., relative cleavage enhancement observed with positively supercoiled DNA divided by the relative cleavage enhancement with negatively supercoiled DNA) of ∼1.3 (Table 1). It should be noted that the superhelical specificity for etoposide varied little (1.0-1.3) over the entire range of drug concentration examined. These correlations, together with the ligation and site-specificity data, imply that interactions between etoposide and topoisomerase IIα are not altered significantly by the handedness of DNA. Rather, the decrease in drug-induced cleavage of positively supercoiled DNA reflects the reduced baseline (i.e., non-drug) level of enzyme-mediated scission with this substrate.

In light of this finding, the influence of DNA geometry on the actions of other topoisomerase II-targeted drugs was characterized. The agents employed represent a structurally diverse group of natural and synthetic compounds. TOP-53 is a non-intercalative drug that is a more potent derivative of etoposide (61, 62). Genistein and CP-115,953 are non-intercalative compounds that share a similar core ring structure. Genistein is a naturally occurring bioflavanoid that is believed to possess chemopreventative properties (63-65). CP-115,953 belongs to the quinolone family (4, 66-69), a drug class that includes a number of widely used antibacterials that target prokaryotic type II topoisomerases. TAS-103 displays strong interactions with DNA and is both an outside binder and an intercalating agent (70-72). Finally, amsacrine is an intercalative compound that is in clinical use (4, 73, 74).

All of the above compounds increased DNA cleavage mediated by human topoisomerase IIα. Data for 50 μM drugs are shown in Figure 3 and Table 1. The only exception is CP-115,953, for which data are shown for 5 μM drug (higher drug concentrations were not used because they induced multiple DNA cleavage events per plasmid). As found for etoposide, most induced higher absolute levels of scission with negatively supercoiled substrates, but displayed greater relative cleavage enhancement with positively supercoiled DNA. The non-intercalative drugs displayed superhelical specificities between 1.3 and 1.8. These values remained constant from the lowest drug concentrations examined (5 or 10 μM) up to concentrations that induced multiple DNA cleavage events per plasmid (50 or 100 μM). It was not possible to determine superhelical specificities for CP-115,953 above 5 μM due to the reason discussed above. Taken together, these results suggest that the superhelical specificity for DNA cleavage with non-intercalative topoisomerase II poisons is independent of drug concentration.

The superhelical specificities of the two intercalative drugs (TAS-103 and amsacrine) at 50 μM, 7.1 and 3.0, respectively, were greater than those seen for the non-intercalative compounds. These higher values suggest that DNA geometry may influence the actions of intercalative compounds towards topoisomerase IIα beyond their effects on baseline DNA cleavage mediated by the enzyme.

Drugs that target topoisomerases are believed to work at the enzyme-nucleic acid interface (4, 7, 59, 75-79). However, intercalative agents have two additional effects on DNA that could impact levels of topoisomerase-mediated scission in a geometry-specific manner. First, since these compounds locally underwind DNA, they induce compensatory unconstrained positive superhelical twists in distal regions of covalently closed circular molecules (80, 81). Thus, as the concentration of an intercalating agent increases, a plasmid that is topologically negatively supercoiled would appear to contain positive superhelical twists. As discussed above, baseline levels of DNA cleavage mediated by topoisomerase IIα are lower with positively supercoiled substrates. Consequently, the apparent change in DNA topology induced by intercalation could diminish the ability of a compound to enhance cleavage with underwound substrates. In contrast, the apparent geometry of a positively supercoiled plasmid (which already is overwound) would not change substantially upon addition of an intercalative drug.

Second, the accumulation of drugs in the double helix has the potential to inhibit enzyme binding or activity. Because the generation of positive superhelical twists by DNA intercalation induces torsional stress in the double helix, the ability of covalently closed molecules to absorb these compounds is limited. Since overwound plasmids are under positive torsional stress even in the absence of drugs, they cannot bind as many intercalative molecules as underwound DNA. Therefore, enzyme activity on positively supercoiled substrates is less likely to be inhibited by the accumulation of bound drug.

Two independent experiments were carried out to determine whether the above effects contribute to the higher superhelical specificity of intercalative agents. The first utilized ethidium bromide, a classical intercalating agent that does not enhance topoisomerase II-mediated DNA scission. When ethidium bromide was included in reaction mixtures, there was a precipitous drop in the ability of human topoisomerase IIα to cleave negatively supercoiled pBR322 (Figure 4). Levels of cleavage decreased ∼6-fold at 10 μM ethidium bromide, which corresponds to the concentration at which “full” intercalation was observed (see inset). Beyond 10 μM, little additional inhibition was observed. Thus, the decrease in DNA cleavage induced by ethidium bromide correlates with the change in the apparent supercoiled state of the plasmid substrate. Consistent with this conclusion, ethidium bromide had a much smaller effect on DNA cleavage when the initial substrate was positively supercoiled (Figure 4). Furthermore, once the concentration of ethidium bromide exceeded 10 μM, DNA cleavage levels for positively and negatively substrates were virtually identical.

To determine whether ethidium bromide accumulation on the double helix affects the DNA cleavage activity of topoisomerase IIα independent of changes in DNA geometry, a linear substrate was employed (Figure 4, right panel). Since linear molecules are not topologically constrained, DNA intercalation does not induce torsional stress or positive supercoiling in these substrates. The addition of 10 μM ethidium bromide only had a minor effect on levels of DNA cleavage with linear pBR322. These results suggest that ethidium bromide inhibits DNA cleavage mediated by human topoisomerase IIα primarily by altering the apparent topology of the DNA substrate and to a lesser extent by accumulation within the double helix.

The second experiment examined the effects of a broad concentration range of amsacrine or TAS-103 on DNA scission mediated by topoisomerase IIα. In contrast to ethidium bromide, these two intercalating drugs are potent topoisomerase II poisons that enhance DNA cleavage (4, 70-74). The initial study examined amsacrine. When negatively supercoiled plasmids were employed as the substrate, peak levels of cleavage were observed between 25 and 50 μM drug (Figures 3 and 5, left panel). This is the amsacrine concentration range in which changes in DNA topology begin to appear (Figure 5, left panel inset). Cleavage levels dropped ∼50% between 25 and 250 μM drug and plateaued thereafter (Figure 5, left panel). It is notable that “full” intercalation is observed at an amsacrine concentration of ∼200 μM (see inset). As discussed above, these findings are consistent with the suggestion that the rise in drug-induced DNA cleavage by amsacrine is attenuated by the concomitant fall in baseline scission caused, at least in part, by the apparent change in the geometry of the plasmid substrate.

A different pattern was seen with positively supercoiled substrates. Levels of DNA scission peaked between 100 and 200 μM amsacrine and remained constant at higher drug concentrations (Figure 5, left panel). Because of the changes in cleavage enhancement seen with negatively and positively supercoiled substrates, the superhelical specificity of amsacrine-induced DNA cleavage was concentration-dependent. Values were in the range of 1.0-1.5 at drug concentrations that did not substantially alter the apparent topology of negatively supercoiled substrates (<25 μM), increased to 3.0-3.5 at concentrations that began to affect apparent topology (25-150 μM), and plateaued at ∼5.1 at concentrations that induced full intercalation.

It is notable that the plateau level of cleavage observed with positively supercoiled DNA was higher than that observed with negatively supercoiled molecules. It is unlikely that amsacrine intercalation could make negatively supercoiled plasmids appear to be more overwound than positively supercoiled molecules at a comparable drug concentration (see Figure 4). Therefore, it is concluded that the preferential accumulation of amsacrine in underwound substrates also contributes to the decrease in DNA cleavage observed with negatively supercoiled molecules at high drug concentrations.

To determine whether the concentration dependence of superhelical specificity is a general feature of intercalative drugs, the ability of TAS-103 to induce DNA cleavage by topoisomerase IIα was examined (Figure 5, right panel). Results were similar to those seen with amsacrine. Scission of negatively supercoiled plasmid rose initially, peaked at drug concentrations that began to alter the apparent topology of the substrate, and fell to approximately baseline thereafter. Conversely, scission of positively supercoiled plasmids increased and remained high over the entire range of TAS-103 examined. As a result, the superhelical specificity of TAS-103, like amsacrine, was concentration dependent. Values were 1.2-1.6 at drug concentrations that did not substantially alter the apparent topology of negatively supercoiled substrates (<5 μM), increased to 2.6-3.7 at concentrations that begin to affect apparent topology (5-10 μM), and plateaued at 7.1-9.3 at concentrations that induced full intercalation. This concentration dependence for superhelical specificity is in sharp contrast to those calculated for non-intercalative compounds and may be a defining distinction between intercalative and non-intercalative topoisomerase II poisons.

Effects of DNA Supercoil Geometry on Drug-induced DNA Cleavage Mediated by Human Topoisomerase IIβ

In contrast to human topoisomerase IIα, topoisomerase IIβ is not required for DNA replication and does not preferentially remove positive superhelical twists (2, 26, 28-30, 51). However, the effects of nucleic acid geometry on DNA cleavage mediated by the β isoform have never been examined. Since topoisomerase IIβ is an important target for anticancer drugs (35) and also appears to mediate some of the toxic effects of these agents in differentiated tissues (33-35), the ability of the enzyme to cleave positively vs. negatively supercoiled substrates was assessed.

Similar to topoisomerase IIα, topoisomerase IIβ maintained lower (∼3- to 4-fold) levels of cleavage complexes with positively supercoiled DNA in the absence of drug (Figure 6, left panel). The enzyme also displayed lower (∼2-fold) cleavage activity with positively supercoiled substrates over a range of etoposide concentrations (Figure 6, right panel). However, the effects of DNA geometry on drug-induced scission were not as pronounced with topoisomerase IIβ as they were with the α isoform. As seen in Figure 7, the percentage of positively supercoiled molecules cleaved in the presence of several drugs approached or actually exceeded that observed with negatively supercoiled substrates. Consequently, the calculated superhelical specificities for most drugs were somewhat higher for topoisomerase IIβ than they were for topoisomerase IIα (Table 2). Once again, the superhelical specificity of the non-intercalative compounds was concentration-independent and remained constant at all drug concentrations examined. Furthermore, the highest superhelical specificities were observed for the two intercalative drugs, TAS-103 and amsacrine.

To further analyze this finding, the effects of amsacrine and TAS-103 on DNA scission mediated by topoisomerase IIβ was examined over a broad concentration range (data not shown). As found with the α isoform, levels of negatively supercoiled DNA cleaved by topoisomerase IIβ peaked at drug concentrations at which changes in apparent topology were obvious, dropped, and plateaued at concentrations that induced full intercalation. In contrast, levels of cleavage with positively supercoiled substrates increased and remained high over the entire drug ranges examined. Thus, as discussed above for topoisomerase IIα, the superhelical specificities for amsacrine and TAS-103 were concentration-dependent with topoisomerase IIβ, and rose from initial values of 1.2 and 1.7, respectively, at low concentrations, to 8.3 and 9.1, respectively, at high concentrations. These results further suggest that drug-induced cleavage of underwound substrates by topoisomerase IIβ is attenuated by the ability of intercalative agents to change the apparent geometry of DNA and by increased drug accumulation on negatively supercoiled molecules.

Effects of DNA Supercoil Geometry on Drug-induced DNA Cleavage Mediated by Human Topoisomerase I

Topoisomerase I is an important target for several new anticancer drugs that are based on camptothecin, a naturally occurring non-intercalative compound found in the Chinese tree Camptotheca acuminata (8, 10, 20, 21, 82). These drugs kill cells by increasing levels of single-stranded DNA breaks (i.e., nicks) that are generated by the type I enzyme (8, 10, 37). Since one of the major functions of topoisomerase I is to alleviate torsional stress that builds up ahead of replication forks and transcription complexes (2, 5, 10), the effects of nucleic acid geometry on enzyme-mediated DNA cleavage was examined in the absence or presence of camptothecin (Figure 8).

In sharp contrast to the type II enzymes, human topoisomerase I maintained a higher level of cleavage intermediates with overwound DNA. Three times more nicked molecules were generated with positively supercoiled plasmids than were observed with negatively supercoiled substrates (see bar graph inset). A far more striking effect of DNA geometry on topoisomerase I was seen in the presence of camptothecin (Figure 8). Over a concentration range of 1-25 μM drug, the enzyme generated dramatically higher levels of nicked DNA with positively supercoiled substrates. For example, at 1 μM camptothecin, ∼16 times more DNA cleavage was observed with positively as compared to negatively supercoiled pBR322, yielding a superhelical specificity of 4.7. Together with the findings for the type II enzymes, these results demonstrate that the handedness of DNA supercoils has a profound influence on DNA cleavage reactions mediated by human topoisomerases.