Molecular Analysis of Yeast and Human Type II Topoisomerases

The DNA sequence selectivity of topoisomerase II (top2)-DNA cleavage complexes was examined for the human (top2α), yeast, and Escherichia coli (i.e. gyrase) enzymes in the absence or presence of anticancer or antibacterial drugs. Species-specific differences were observed for calcium-promoted DNA cleavage. Similarities and differences in DNA cleavage patterns and nucleic acid sequence preferences were also observed between the human, yeast, and E. coli top2 enzymes in the presence of the non-intercalators fluoroquinolone CP-115,953, etoposide, and azatoxin and the intercalators amsacrine and mitoxantrone. Additional base preferences were generally observed for the yeast when compared with the human top2α enzyme. Preferences in the immediate flanks of the top2-mediated DNA cleavage sites are, however, consistent with the drug stacking model for both enzymes. We also analyzed and compared homologous mutations in yeast and human top2, i.e.Ser740 → Trp and Ser763 → Trp, respectively. Both mutations decreased the reversibility of the etoposide-stabilized cleavage sites and produced consistent base sequence preference changes. These data demonstrate similarities and differences between human and yeast top2 enzymes. They also indicate that the structure of the enzyme/DNA interface plays a key role in determining the specificity of top2 poisons and cleavage sites for both the intercalating and non-intercalating drugs.

DNA topoisomerases are enzymes that catalyze changes in the topology of DNA via a mechanism involving the transient breakage and rejoining of phosphodiester bonds in the DNA backbone (1,2). Studies in both prokaryotic and eukaryotic cells have demonstrated the importance of topoisomerases in transcription, DNA replication, and chromosome segregation. The type II topoisomerases (top2) 1 make transient DNA double-strand breaks and change the linking number of DNA in steps of two. They play key roles in DNA metabolism and chromosome structure and are essential in eukaryotic cells (2,3). In order to maintain the integrity of the cleaved DNA during this process, the top2 enzymes form a proteinaceous bridge that spans the DNA break. This bridge is anchored by covalent phosphotyrosyl bonds established between each of the active site tyrosine residues of the homodimeric enzyme and the 5Ј-DNA termini of the newly created DNA double-strand break (2). Under physiological conditions, these covalent top2-DNA complexes (referred to as cleavage or cleavable complexes) are normally short lived intermediates in the catalytic cycle of the enzyme.
Beyond its vital cellular functions, top2 is the primary cytotoxic target for some of the most active drugs for the treatment of human cancers (4 -8). Top2 inhibitors can be divided into two groups, top2 catalytic inhibitors and top2 poisons (8). Top2 catalytic inhibitors do not stabilize DNA cleavage complexes. Bisdioxopiperazines (ICRF 159, 187 (dexrazoxane), and 193) belong to this category (9). Top2 poisons inhibit the enzyme by increasing the steady-state levels of DNA cleavage complexes (8,10,11). Hence they convert top2 into a physiological toxin that creates DNA double-strand breaks in the genome of treated cells (5,8,10,12). Top2 poisons can be further subdivided into two groups as follows: the DNA intercalators that include doxorubicin, mitoxantrone, amsacrine, ellipticines/olivacines, and the non-intercalators whose main representatives are the demethylepipodophyllotoxins etoposide (VP-16) and teniposide (VM-26), the quinolones among which CP-115,953 acts as a dual eukaryotic and prokaryotic top2 poison (13,14), and some azatoxin derivatives (15).
Although top2 cleaves DNA at preferred sequences, little is known regarding the mechanism by which the enzyme selects its sites of action. Recent studies with etoposide suggested that etoposide interacts with top2 rather than with the DNA (7). On the other hand, studies with a photoactivated amsacrine derivative and with bisantrene/amsacrine congeners indicated that for these agents, drug-DNA interactions are critical for the formation of top2-DNA cleavage complexes (16,17). Analyses of drug-induced top2 cleavages revealed drug-specific base preferences in the immediate vicinity of the cleavage sites. In the case of amsacrine, A at position ϩ1 was preferred, whereas in the case of etoposide, teniposide, mitoxantrone, and quinolones the highest preference is for C at position Ϫ1 (see diagram in Fig. 3). From these results, it has been proposed that drugs bind at the enzyme-DNA interface and form a ternary complex with top2 and the DNA. This model has been referred to as the drug stacking model (8,18) or position poison model (19,20).
Yeast is a powerful model system to study topoisomerase inhibitors (3,21,22). However, no detailed comparison has been reported for DNA cleavage complexes formed by the yeast and the human top2 enzymes. Furthermore, since detailed fundamental information is available for the yeast enzyme (2, 23), but not for the human enzymes, direct comparison of the human and yeast proteins is useful for a structural under-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ Supported by NCI Grants CA52814 and CA21765 from the National Institutes of Health and the American Lebanese Syrian Associated Charities.
standing of the human enzymes as a drug target. Yeast top2 is also a potential target for antifungal treatment, and structural differences between the yeast and human top2 may allow selective targeting of the yeast top2 over its human counterpart. In this way, a clear and detailed comparison between yeast and human top2 is warranted and necessary. Since mutation of a conserved serine residue (Ser 740 3 Trp) in yeast top2 was recently reported to alter both enzyme-DNA and drug interactions (24), the homologous mutation (Ser 763 3 Trp) in human top2␣ was analyzed in this study.

EXPERIMENTAL PROCEDURES
Materials, Chemicals, and Enzymes-Etoposide (VP16) was obtained from Bristol-Myers Squibb Co. Amsacrine and mitoxantrone were from the Drug Synthesis and Chemistry Branch (NCI, Bethesda, MD). Azatoxin and its derivatives were provided by Dr. T. Macdonald, Department of Chemistry of Virginia, Charlottesville, VA (15 Preparation of End-labeled DNA Fragments by PCR-Three sets of labeled DNA fragments were prepared from the human c-MYC gene by PCR. A 254-base pair DNA fragment from the first intron was prepared between positions 3035 and 3288, with numbers referring to Gen-Bank TM genomic positions using oligonucleotides 5Ј-GTAATCCAGAA-CTGGATCGG-3Ј for the upper strand and 5Ј-ATGCGGTCCCTACTCC-AAGG-3Ј for the lower strand (annealing temperature 56°C). A 401base pair DNA fragment from the junction between the first intron and first exon was prepared between positions 2671 and 3072 using oligonucleotides 5Ј-TGCCGCATCCACGAAACTTT-3Ј for the upper strand and 5Ј-TTGACAAGTCACTTTACCCC-3Ј for the lower strand (annealing temperature 60°C). A 480-base pair fragment from the first exon containing promoters P 1 and P 2 was prepared between positions 2265 and 2745 using the oligonucleotides 5Ј-GATCCTCTCTCGCTAATCTC-CGCCC-3Ј for the upper strand and 5Ј-TCCTTGCTCGGGTGTTGTAA-GTTCC-3Ј for the lower strand (annealing temperature 70°C). A 213base pair fragment from the human c-JUN gene was prepared between positions 5Ј-TGTTGACAGCGGCGGAAAGCAGS-3Ј for the upper strand and 5Ј-CGTCCTTCTTCTCTTGCGTGGCTCT-3Ј for the lower strand (annealing temperature 64°C). Single end labeling of these DNA fragments was obtained by 5Ј-end labeling of the specific primer oligonucleotide. Ten picomoles of DNA was incubated for 60 min at 37°C with 10 units of T4 polynucleotide kinase and 10 pM [␥-32 P]ATP (100 Ci) in kinase buffer (70 mM Tris-HCl, pH 7.6, 0, 1 M KCl, 10 mM MgCl 2 , 5 mM dithiothreitol, and 0.5 mg/ml bovine serum albumin). Reactions were stopped by heat denaturation at 70°C for 15 min. After purification using Sephadex G-25 columns (Roche Molecular Biochemicals), the labeled oligonucleotides were used for PCR. Approximately 0.1 g of the c-MYC DNA that had been restricted by SmaI and PvuII (fragment 2265-2745) and XhoI and XbaI (fragment 2671-3072 and fragment 3035-3288) was used as template for the PCR. Ten picomoles of each oligonucleotide primer, one of them being 5Ј-labeled, was used in 22 temperature cycle reactions (each cycle with 94°C for 1 min, annealing for 1 min, and 72°C for 2 min). The last extension was for 10 min. DNA was purified using PCR Select-II columns (5 Prime 3 3 Prime, Inc., Boulder, CO).
Overexpression and Purification of Yeast and Human Topoisomerase II-Wild-type yeast and human top2, Ser 740 3 Trp, and Ser 763 3 Trp proteins were overexpressed using YEpTOP2-PGAL1 or YEptop2-S*W-PGAL1 using yeast strain JEL1t1 Ϫ (25) and purified to homogeneity as described previously (26). The detailed procedure has been described elsewhere (27).
Electrophoresis and Base Preference Analysis-For DNA sequence analysis, samples were precipitated with ethanol and resuspended in 5 l of loading buffer (80% formamide, 10 mM NaOH, 1 mM EDTA, 0, 1% xylene cyanol, and 0.1% bromphenol blue). Samples were heated to 95°C for 5 min and thereafter loaded onto DNA sequencing gels (7% polyacrylamide, 19:1 acrylamide/bisacrylamide) containing 7 M urea in 1ϫ Tris borate/EDTA buffer. Electrophoresis was performed at 2500 V (60 watts) for 2-3 h. The gels were dried on Whatman No. 3MM paper sheets and visualized using a PhosphorImager (Molecular Dynamics, Sunnyvale, CA) and ImageQuant software. The determination of preferred bases around top2 cleavage sites was done as described previously (28 -30).

RESULTS
Calcium-promoted DNA Cleavages Differ between Human, Yeast, and Bacteria Top2 Enzymes-The calcium-promoted, drug-independent DNA cleavage sites induced by yeast and human wild-type top2 and by E. coli gyrase (i.e. in bacterial type II topoisomerase) (31, 32) were mapped on the upper strand of the c-MYC first intron fragment (Fig. 1). Even in the presence of magnesium, differences in the cleavage sites could Top2 reactions were performed at 37°C for 30 min in the presence of 5 mM MgCl 2 or 5 mM CaCl 2 as indicated and stopped by adding EDTA and SDS (25 mM and 1% final concentrations, respectively). Purine ladder was obtained after formic acid reaction. h WT, human wild-type top2␣; y WT, yeast wild-type top2; Gyrase, wild-type gyrase from E. coli. Numbers correspond to genomic positions of the nucleotide covalently linked to top2. be observed. When magnesium was replaced by calcium, higher levels of DNA cleavage were seen in the yeast protein. DNA cleavage sites common to both proteins were seen in the pres-ence of Ca 2ϩ . However, there were also major differences in the intensity of cleavage at other sites. Although a number of DNA cleavage sites in yeast were also found in gyrase, e.g. at posi- The panels present the probability of the observed base frequency deviations from expectation for the indicated enzyme. In the y axis, P is the probability of observing that deviation or more, either as excess (above base line) or deficiency (negative values below base line) relative to the expected frequency of each individual base (29). Cleavage sites for the human (panel A) and the yeast (panel B) wild-type enzymes were analyzed. Drug concentration was 100 M. A schematic representation of a top2 cleavage complex is shown between panels A and B. The top2 covalent linkage to the 5Ј-DNA termini is shown as a circle at the ϩ1 position.
tions 3227, 3221, 3064, the majority of cleavage sites were specific for either protein. These results suggest that human, yeast, and bacterial top2 are different regarding Ca 2ϩ -promoted DNA cleavage in the absence of a top2 drug.
Cleavage Sites Induced by Yeast and Human Top2 in the Presence of Drugs-To investigate further possible differences between yeast and human top2 DNA cleavage patterns, we compared the drug-induced cleavage sites (Fig. 2). Several cleavage sites induced by etoposide were much stronger with the yeast enzyme, for example in Fig In the case of CP-115,953, human top2␣ caused greater cleavage than the yeast enzyme at several sites (in Fig. 2, panel A, at positions 2842, 2880, 2901, and 2996, and in Fig. 2, panel B, at positions 3187, 3149, and 3084). In the case of the gyrasespecific quinolone ciprofloxacin, yeast top2 showed minimal cleavage induction at position 3144 (Fig. 2, panel B).
With amsacrine, human top2␣ cleaved more extensively than yeast top2 (in Fig. 2, panel A, at positions 2842, 2912, 2962, and 3008 and in Fig. 2, panel B, at positions 3064, 3081, 3084, 3091, and 3118). At some sites (e.g. positions 3144, 3121, and 2974), however, cleavage was stronger with the yeast enzyme. Similarly, several changes in cleavage sites induced by mitoxantrone were seen. The azatoxin-derivative 11␤(4Љ-nitroanilino)azatoxin (15) (Fig. 2, panel A) was markedly more active against the human top2␣ than the yeast top2 in the DNA fragment examined. Taken together, these results show different DNA cleavage patterns for yeast and human top2 in the presence of both intercalating and non-intercalating drugs.

Different Base Preferences of Amsacrine-and Mitoxantronestabilized Cleavage Complexes for the Yeast and Human
Top2-Because the yeast and human top2 enzymes presented different cleavage activity in the presence of drugs, we compared the DNA base preferences for both proteins in the presence of etoposide, amsacrine, mitoxantrone, and CP-115,953 (Figs. 3-6 and Tables I and II). Cleavage sites for the three c-MYC DNA fragments and the c-JUN fragment (see "Experimental Procedures") were analyzed for both DNA strands.
For yeast and human proteins, etoposide preferentially stabilized sites with C at position Ϫ1 (C Ϫ1 ) ( Fig. 3 and Table I). This result agrees well with previous analyses (18,29). Preference on the opposite strand showed a complementary (although slightly weaker) preference for G at position ϩ5. Thus, the different cleavage patterns for yeast and human top2 in the presence of etoposide were not associated with detectably altered base preferences.
Different Base Preferences for Yeast, Human, and E. coli Top2 in the Presence of CP-115,953-In the presence of the fluoroquinolone CP-115,953 ( Fig. 6 and Table II), the human protein preferred cleavage sites with C Ϫ1 (44 of 79 sites), A ϩ1 (38 of 79 sites), and (more weakly) A Ϫ2 (29 of 79 sites). These results are consistent with other reports (33,36). Complementary preferences for T ϩ4 and G ϩ5 were also observed. The yeast top2 showed additional preferences for T Ϫ1 (C Ϫ1 and T Ϫ1 , 49 of 65 sites) and for G ϩ1 (A ϩ1 and G ϩ1 , 54 of 65 sites). There was no clear preference at position Ϫ2 for the yeast protein. Inter-estingly, gyrase showed the T Ϫ1 and G ϩ1 preferences observed for yeast top2 (41 and 50 of 107 sites, respectively). These preferences are in agreement with previous reports obtained with a different fluoroquinolone (37). Thus, it appears that the base preferences for the CP-115,953-induced sites in gyrase are more similar to the yeast than to the human top2␣.

Homologous Mutations of Conserved Serine Residues Alter the Enzyme-DNA and Drug Interactions for Both Yeast and
Human Top2-We recently reported that mutation of Ser 740 3 Trp in yeast top2 affects both DNA and drug interactions (24). To analyze the effect of the homologous mutation in human top2␣ (Ser 763 3 Trp), we compared the calcium-promoted DNA cleavages for both mutant proteins (Fig. 7). Even in the absence of drug (in the presence of Mg 2ϩ ), both mutants presented different cleavage patterns compared with the corresponding wild-type proteins. When magnesium was replaced by calcium, higher levels of DNA cleavage were only seen in the yeast proteins, i.e. in the wild-type enzyme and in top2 S740W . New DNA cleavage sites common to both of the mutant proteins were seen in the presence of Mg 2ϩ and Ca 2ϩ , although there were considerable differences in cleavage intensity. Most of the DNA cleavage sites in the upper and lower strands were staggered by 4 base pairs with a 5Ј-overhang, as expected for concerted top2-induced double-strand cleavage (see Fig. 3) (2,8,11).
Since we demonstrated that the Ser 740 3 Trp in yeast and the Ser 763 3 Trp mutation in human top2␣ increased sensitivity to etoposide and changed the base preferences in the same way, we tested whether human top2␣ S763W and yeast top2 S740W enhanced DNA cleavage by etoposide at the same positions. Fig. 10 shows that a number of cleavage sites were enhanced for both mutant proteins (at positions 3252, 3091, 2996, 2959 and to lesser extent at positions 3141 and 3073). In addition, reduced cleavage for both mutants was observed at positions 2974 and 3121. Several sites, however, showed differences between human top2␣ S763W and yeast top2 S740W , e.g. at positions 3026, 3020, 2901, and 2816. In particular, cleavage at position 3175 was enhanced for yeast top2 S740W but markedly reduced for human top2␣ S763W . Thus, human top2␣ S763W and yeast top2 S740W preserve, at least partially, the differences described above between human and yeast protein-DNA interactions.
Base Preference of Etoposide-induced, Heat-stable Cleavage Complexes Induced by Human Top2␣ S763W -We recently reported that cleavage complexes mediated by yeast top2 S740W in the presence of etoposide have enhanced stability (24,38). The effect of the Ser 763 3 Trp mutation on the stability of human top2␣-DNA cleavage complexes was determined by examining the heat reversibility of the ternary complexes formed with drug, protein, and DNA. Cleavage reactions were carried out with the human top2␣ or top2␣ S763W for 30 min at 37°C, after which reaction mixtures were heated to 65°C for various times prior to the addition of SDS. Fig. 11 shows that most of the etoposide-stabilized cleavage sites were readily reversible for the wild-type protein. In contrast, a number of cleavage sites induced by the human top2␣ S763W showed slow reversal (positions 3091, 3207, 3223, 3238, 3124, 3183, etc.) or no detectable reversal after incubation at 65°C for 20 min (positions 3167, 3171, 3252, 3170, 3174, 3210, etc.). Enhanced heat stability of the DNA cleavage sites induced by human top2␣ S763W was also observed in other c-MYC DNA fragments (data not shown). Enhanced heat stability was also observed with the human wild-type top2␣ at certain sites (positions 3252, 3175, 3194, 3178, etc.). However, the stability was considerably less than for the human top2␣ S763W protein. As already shown for yeast top2 S740W (24), cleavage sites with slow reversibility exhibited highly significant preferences for C Ϫ1 in combination with less strong C Ϫ2 preference in human top2␣ S763W , whereas rapidly reversible cleavage sites did not show any preferences at positions Ϫ1 and Ϫ2 (data not shown). Hence, both mutations Ser 740 3 Trp in yeast and Ser 763 3 Trp in human top2␣ similarly alter the DNA recognition of the corresponding enzyme, markedly affect the interaction with inhibitors, and enhance the stability of the top2 cleavage complexes in the presence of etoposide. DISCUSSION The DNA sequence preference of drugs that target DNA top2 has been widely investigated (34). Early studies showed that topoisomerase-targeting drugs influence the sequence specificity of DNA cleavage by top2 compared with sites of DNA cleavage in the absence of drugs (28,29,39). Not surprisingly, drugs that bind DNA in the absence of enzyme most commonly resulted in cleavage specificities that differed from that seen with the enzyme in the absence of drug. Nonetheless, the cleavage specificity induced by intercalating drugs frequently differed from that expected, based on the binding of drugs to DNA in the absence of enzyme.
A key issue in understanding the mechanism of action of top2-targeting drugs is the determination of where drugs bind in the covalent complex. Important clues can be obtained from the DNA sequence of cleavage sites induced by intercalating drugs. For example, the intercalator amsacrine with human top2␣ exhibited the strongest preference at the ϩ1 base (29). Recent biochemical studies by Kreuzer and colleagues (17) using a photoreactive amsacrine analog demonstrated reactivity only with the Ϫ1 and ϩ1 bases, in agreement with the results suggested from the DNA cleavage pattern.
Only recently have investigators begun to compare the effects of different enzymes on DNA cleavage specificities with the same top2 poison. This problem is of particular interest because mammalian cells express two different top2 isoforms, ␣ and ␤ (2). A recent study compared recombinant forms of human ␣ and ␤ and found similar cleavage specificities for teniposide and the anthracycline 4-demethoxy-3Ј-deamino-3Јhydroxy-4Ј-epidoxorubicin (40). The cleavage specificity was also found to be the same for mouse top2.
Yeast has been commonly used to analyze topoisomerase functions and to study the biochemistry and molecular biology of topoisomerase inhibitors (2,3,22). Of particular importance is the determination of two different structures of the breakage/rejoining domains of the enzyme by x-ray crystallography (41,42). A model for the binding of top2 to DNA has been proposed (43). Although details of specific protein:nucleic acid contacts will require a solution of the structure of the protein bound to DNA, the model is consistent with the notion that residues in the helix-turn-helix domain play key roles in interacting with DNA near the cleavage site and that this domain is also close to sites where top2-targeting drugs interact with DNA (24).
Results reported here showed strong similarities between yeast top2 and recombinant human top2␣ in the cleavage site preferences for several agents. However, several intriguing differences were noted. Interestingly, the non-intercalating agent etoposide showed clear similarities. Both human and yeast top2 have a strong preference for a C at the Ϫ1 position, along with a complementary preference for G at the ϩ5 position. In addition, yeast and human enzymes with homologous mutations in the helix-turn-helix domain (Ser 740 3 Trp and Ser 763 3 Trp for yeast and human, respectively) showed the same change in cleavage specificity, a preference for a C at Ϫ2 (and G at ϩ6) that is independent of the base at the Ϫ1 position. This result is consistent with an etoposide-binding site that is well conserved between the two enzymes.
Clerocidin is a top2 poison that has an action that is analogous to the Ser 740 3 Trp mutant of yeast top2 and the Ser 763 3 Trp mutant of human top2␣. Clerocidin generated heat-and salt-stable covalent complexes with human top2␣ (44) and also heat-stable complexes with yeast top2. 2 The sequence preference for clerocidin with human top2␣ was G at position Ϫ1 (45), 2 J. L. Nitiss, unpublished results. suggesting that interactions between the Ϫ1 base and drug may be an important determinant of the stability of covalent complexes.
The helix-turn-helix domain is also important in drug action with the non-intercalating fluoroquinolones. It is well established that amino acids around Ser 83 of gyrA are the principal site of resistance mutations to fluoroquinolones in E. coli (32,46). Biochemical results also suggested the presence of a quinolone-binding site in the vicinity of Ser 83 (47). Resistance to quinolones has also been observed in yeast mutants with changes in this region (38,48). Interestingly, we detected differences between yeast top2 and human top2␣ in DNA cleavage specificity induced by the fluoroquinolone CP-115,953. For all three topoisomerases examined, there were clear sequence preferences at both the Ϫ1 and ϩ1 bases. The specificities for all three enzymes were somewhat different, but in each case the specificities at the Ϫ1 and ϩ1 positions were pyrimidine and purine, respectively. As was the case for etoposide, complementary preferences, in this case at positions ϩ4 and ϩ5, were also seen.
The most extensive differences between yeast and human top2␣ in cleavage pattern specificities were observed with drugs that intercalate in DNA. For human top2␣, a preference for A at the ϩ1 position was observed, along with a complementary T ϩ4 preference as previously reported (18,29). A sta-tistically significant preference at the Ϫ1 position was not observed. By contrast, yeast showed a clear preference for T at the Ϫ1 position, along with A at ϩ1. Thus, unlike the human enzyme, there is a clear preference with the yeast enzyme at both positions Ϫ1 and ϩ1. This result demonstrates that different drugs cannot be categorized just on the DNA sequence preference around the cleavage site.
The differences between human top2␣ and yeast top2 seen with the intercalating drug mitoxantrone are more complicated. Both the human and yeast enzymes exhibited preferences at both Ϫ1 and ϩ1 positions, but other preferences were also seen, such as A at position Ϫ3 with the human enzyme and C at position ϩ6 with the yeast enzyme. One factor that may contribute to this more complicated pattern is the strong inhibition of cleavage seen at high mitoxantrone concentrations (49 -51). Perhaps the complex pattern that arises for both enzymes may be due in part to the ability of mitoxantrone to inhibit cleavage in a sequence-dependent manner.
A recent model has attempted to explain the similar sequence preferences of different top2 poisons by suggesting that  21  34  34  30  27  30  41  46  22  27  52  28  32  33  26  36  42  41  36  34  C  40  49  48  36  50  43  16  20  70  80  25  44  39  30  20  18  44  47  32  51  G  40  20  25  32  34  30  43  42  22  12  30  38  41  36  70  62  14  18  35  34  T  37  38  34  41  30  36  38  31  26  22  32  31  28  41  24  24  41  36  36  22 Position from cleavage site The reactions were then incubated at 65°C for the indicated times prior to the addition of SDS and proteinase K. Top2, no drug treatment; Control, no top2, no drug treatment. Numbers correspond to genomic positions of the nucleotide covalently linked to top2. h WT, human wild-type top2␣; h S763W, human top2␣ S763W . Connected arrowheads correspond to DNA cleavage sites with a 4-base pair stagger that represent potential DNA double-strand breaks. drugs that have a common sequence preference share a common pharmacophore (52). In this model, a top2 poison is modeled as consisting of two "modules," one that mediates DNA binding, e.g. which intercalates in DNA, and a second module that interacts with the enzyme. By this model, sequence specificity would be determined mainly by the DNA binding module, whereas the potency of the drug would also depend on the second module. The results presented here demonstrate that the same topoisomerase poison with the same DNA substrate exhibits different sequence specificities with different top2 enzymes. Thus, our results require a modification of the model. For example, enzyme binding, DNA cleavage, and strand separation of the 4-base overlap between sites of cleavage may lead to a reorientation of the drug interacting with DNA (or enzyme), and the reorientation may affect the ability of the drug to prevent religation. Such reorientation seems particularly plausible for a drug molecule like amsacrine that intercalates between the Ϫ1 and ϩ1 bases. The reorientation may involve specific contacts between the drug and the enzyme, and these contacts may be different for the yeast and human top2 enzymes.
Recent results from Osheroff and colleagues (19,53) have stressed the importance of the enzyme in determining cleavage site specificity with non-intercalating top2 poisons. The results described here indicate that the enzyme plays a very important role in the cleavage specificity of intercalating top2 poisons as well. Since intercalators bind DNA with a distinct sequence preference, part of the base sequence specificity of top2 poisoning by these agents depends on drug-DNA interactions. The results presented here highlight the importance of interactions of all three components of the trapped covalent complex with protein, DNA, and drug.