Selected Novel Flavones Inhibit the DNA Binding or the DNA Religation Step of Eukaryotic Topoisomerase I*

Topoisomerases are involved in many aspects of DNA metabolism such as replication and transcription reactions. Camptothecins, which stabilize the covalent inter- mediate of topoisomerase I and DNA are effective, though toxic, drugs for cancer therapy. In this study, a new class of topoisomerase I inhibitors was identified, and their mode of action was characterized using re- combinant human topoisomerase I preparations and human HL-60 leukemic cells. Quercetin and the related natural flavones, acacetin, apigenin, kaempferol, and morin, inhibit topoisomerase I-catalyzed DNA religa- tion. In contrast to camptothecin, these compounds do not act directly on the catalytic intermediate and also do not interfere with DNA cleavage. However, formation of a ternary complex with topoisomerase I and DNA during the cleavage reaction inhibits the following DNA religation step. 3,3 (cid:42) ,4 (cid:42) ,7-Tetrahydroxy-substituted flavones stabilize the covalent topoisomerase I-DNA intermedi- ate most efficiently. Enhanced formation of covalent topoisomerase I-DNA complexes was also demonstrated in human HL-60 cells. In contrast, synthetic 3 (cid:42) ,5 (cid:42) -dibromo-4 (cid:42) -hydroxy-3-methylflavones bind selectively to topoisomerase I in its non-DNA-bound form and block the following DNA binding step. As that oligonucleotide substrates the enzyme was further purified to apparent homogeneity by gel filtration using a Supedex 200 column (1 (cid:51) 30 cm, Pharmacia) equilibrated, and developed with 400 m M potassium phosphate, pH 7.0, 1 m M DTT, 1 m M PMSF, 10% glycerol. Pure and partially pure enzyme preparations were stored with 60% glycerol at (cid:50) 20 °C and were stable for at least 5 months. Measurement of Open Circular Plasmid DNA Formation, DNA Re- laxation, and DNA Mobility Shift— 250 ng of pBR 322 plasmid DNA were incubated with 200 units of human topoisomerase I in the pres- ence of camptothecin or various flavonoids at the concentrations indicated. Controls were without drugs or without enzyme. The assay had a final volume of 40 (cid:109) l of reaction buffer (10 m M BisTris-propane, pH 7.9, containing 10 m M MgCl 2 , 10 m M KCl, 0.1 m M DTT, and 10% dimethyl sulfoxide). Incubation at 37 °C for 30 min was terminated by addition of 1% SDS. Samples were then digested with 1 mg/ml protein- ase K at 37 °C for 30 min. For the study of noncovalent binding of protein and DNA by DNA mobility shift, SDS denaturation and pro- teinase K digestion were omitted. Gel electrophoresis was performed at 0.4 V/cm for 12 h in 1% agarose gels with TBE buffer containing 0.5 (cid:109) g/ml of ethidium bromide (for open circular formation and DNA mobility shift measurements) or with Tris acetate-EDTA buffer in the absence of ethidium bromide (for DNA relaxation measurements). In the latter case gels were stained with 0.5 (cid:109) g/ml ethidium bromide after electrophoresis. Fluorescence of ethidium bromide in the gels (excita-tion 302 nm, emission (cid:46) 600 nm) was documented by Polaroid photog- raphy.

Topoisomerases are involved in many aspects of DNA metabolism such as replication and transcription reactions. Camptothecins, which stabilize the covalent intermediate of topoisomerase I and DNA are effective, though toxic, drugs for cancer therapy. In this study, a new class of topoisomerase I inhibitors was identified, and their mode of action was characterized using recombinant human topoisomerase I preparations and human HL-60 leukemic cells. Quercetin and the related natural flavones, acacetin, apigenin, kaempferol, and morin, inhibit topoisomerase I-catalyzed DNA religation. In contrast to camptothecin, these compounds do not act directly on the catalytic intermediate and also do not interfere with DNA cleavage. However, formation of a ternary complex with topoisomerase I and DNA during the cleavage reaction inhibits the following DNA religation step. 3,3,4,7-Tetrahydroxy-substituted flavones stabilize the covalent topoisomerase I-DNA intermediate most efficiently. Enhanced formation of covalent topoisomerase I-DNA complexes was also demonstrated in human HL-60 cells. In contrast, synthetic 3,5-dibromo-4-hydroxy-3-methylflavones bind selectively to topoisomerase I in its non-DNA-bound form and block the following DNA binding step. As a consequence, these synthetic flavonoids are capable of counteracting topoisomerase I-directed effects of camptothecin. Inhibition of DNA binding is obtained by voluminous hydrophobic substituents in 6-position of the flavone structure. Our data show that selective inhibitors of both half-reactions of topoisomerase I can be derived from the flavone structure.
Flavonoids, ubiquitously occurring and widely consumed secondary metabolites of plants, are among the active components in vegetables and fruits that prevent or inhibit cancer development (1,2). Among naturally occurring flavonoids, quercetin (3,3Ј, 4Ј,5,7-pentahydroxyflavone), the lead compound, is most widely studied in vitro and in vivo. It is known to interfere with several pathways of intermediary metabolism and various components of signal transduction cascades, especially those stimulated by the tumor promoter 12-O-tetradecanoylphorbol-13acetate. Quercetin and related flavonoids are known to inhibit the growth of tumour cells (3)(4)(5)(6)(7)(8) and to potentiate the cytotox-icity of DNA-damaging anti-cancer drugs such as cis-platinum (9 -11). However, their cytotoxic effect cannot yet be clearly assigned to a specific cellular target. The cytotoxic action of quercetin is associated with DNA scission in a divalent cationdependent manner (12). Moreover, quercetin, the isoflavone genistein, and some related compounds have been shown to induce topoisomerase II-mediated DNA cleavage in mammalian cells (13,14). Genistein is also known to inhibit type I topoisomerases in vitro (15). Thus, topoisomerase-mediated DNA damage seems to be a candidate mechanism, by which some flavonoids may exert their cytotoxic potential.
Here we show, that among a large number of flavonoids, only certain flavones are potent and selective inhibitors of topoisomerase I-catalyzed DNA religation. Their ability to stabilize the covalent topoisomerase I-DNA complex in vitro and in living cells is similar to that of the known topoisomerase I inhibitor camptothecin, although the mechanism of interaction appears to be different. Moreover, defined molecular substitutions of the flavone structure generate a new class of topoisomerase I-targeted compounds that inhibit selectively the DNA binding of the enzyme. Thus, flavones seem to be an ideal model for the study of structural requirements for selective interaction with DNA cleavage and religation reactions catalyzed by topoisomerase I that may serve for the development of new anti-cancer drugs.

Materials
The mouse monoclonal antibody to human topoisomerase I was a kind gift of Dr. Igor Bronstein, Engelhard Institute, Moscow, Russia. Peroxidase labeled goat anti-mouse IgG was purchased from Jackson Inc. Enhanced chemoluminescence detection reagents (ECL), goldlabeled goat anti-mouse IgG (Auroprobe), and silver enhancement reagents (Intense BL) were obtained from Amersham, Little Chalfont, United Kingdom. Quercetin, other natural flavonoids, and purified oligonucleotides were obtained from Karl Roth GmbH, Karlsruhe, Germany. Synthetic flavonoids were kindly supplied by Dr. K. Irmscher, E. Merck, Darmstadt, Germany. For chromatography we used columns and a FPLC system of Pharmacia Biotech Inc., Uppsala, Sweden. Camptothecin and polyethyleneimine were obtained from Sigma, Mü nchen, Germany. Benzonase® was purchased from E. Merck. PMSF 1 was from Fluka, Neu-Ulm, Germany. All other reagents were of the highest degree of purity commercially available.

Methods
Cell Culture-Human HL-60 cells (American Tissue Culture Collection CCL240, Rockville, MD) were grown in liquid culture (RPMI 1640 ϩ fetal calf serum 5% (v/v), 10 g/liter penicillin/streptomycin) in a humidified atmosphere containing 5% (v/v) CO 2 . Cells were routinely checked to be free of mycoplasms by immunoassays and cultural analysis.
Enzyme Preparation-Human topoisomerase I was heterologously produced in Saccharomyces cerevisiae RS 190: MATa ade2-1 ura3-1 his3-11 trp1-1 leu2-3, 112 can1-100 top1-8::LEU2, transformed with plasmid YEpGAL1-hTOP1, bearing the gene of human topoisomerase I under the inducible yeast promoter PGAL1 as described in Bjornsti et al. (16). After induction, yeast cells were frozen in liquid N 2 and powdered in a mortar. 20 g of yeast powder was rapidly mixed with 20 ml of 50 mM potassium phosphate, pH 7.0, 2 M NaCl, 1 mM DTT, 1 mM PMSF, and 10% glycerol. After sedimentation of the debris (30 min, 36,000 ϫ g, 4°C), the extract was subjected to DNA precipitation with 0.1% polyethyleneimine for 30 min at 4°C, followed by centrifugation (30 min, 36,000 ϫ g, 4°C). Proteins in the supernatant were precipitated with 3 M ammonium sulfate for 30 min at 4°C, sedimented (30 min, 36 000 ϫ g, 4°C), renatured with 50 mM potassium phosphate, pH 7.0, 1 mM DTT, 1 mM PMSF, and 10% glycerol and finally adsorbed to a heparin-Sepharose column (5 ϫ 5 cm). After washing, topoisomerase I was eluted with 400 mM potassium phosphate, pH 7.0, 1 mM DTT, 1 mM PMSF. One l of the eluate typically contained 10 -20 ng of topoisomerase I with a molecular weight of approximately 100 kDa and a specific activity of 5-10 units/ng. One unit (unit of topoisomerase I activity) completely relaxes 250 ng of pBR 322 plasmid DNA within 30 min at 37°C. The purity of the enzyme in the heparin-Sepharose eluate was 15-20%. For experiments involving oligonucleotide substrates the enzyme was further purified to apparent homogeneity by gel filtration using a Supedex 200 column (1 ϫ 30 cm, Pharmacia) equilibrated, and developed with 400 mM potassium phosphate, pH 7.0, 1 mM DTT, 1 mM PMSF, 10% glycerol. Pure and partially pure enzyme preparations were stored with 60% glycerol at Ϫ20°C and were stable for at least 5 months.
Measurement of Open Circular Plasmid DNA Formation, DNA Relaxation, and DNA Mobility Shift-250 ng of pBR 322 plasmid DNA were incubated with 200 units of human topoisomerase I in the presence of camptothecin or various flavonoids at the concentrations indicated. Controls were without drugs or without enzyme. The assay had a final volume of 40 l of reaction buffer (10 mM BisTris-propane, pH 7.9, containing 10 mM MgCl 2 , 10 mM KCl, 0.1 mM DTT, and 10% dimethyl sulfoxide). Incubation at 37°C for 30 min was terminated by addition of 1% SDS. Samples were then digested with 1 mg/ml proteinase K at 37°C for 30 min. For the study of noncovalent binding of protein and DNA by DNA mobility shift, SDS denaturation and proteinase K digestion were omitted. Gel electrophoresis was performed at 0.4 V/cm for 12 h in 1% agarose gels with TBE buffer containing 0.5 g/ml of ethidium bromide (for open circular formation and DNA mobility shift measurements) or with Tris acetate-EDTA buffer in the absence of ethidium bromide (for DNA relaxation measurements). In the latter case gels were stained with 0.5 g/ml ethidium bromide after electrophoresis. Fluorescence of ethidium bromide in the gels (excitation 302 nm, emission Ͼ600 nm) was documented by Polaroid photography. The amounts of open and closed circular pBR 322 DNA were determined by video-amplified fluorescence intensity measurements of the respective DNA bands in each lane of the gel using a video densitometer (Froebel, Wasserburg, Germany). For immunoblot analysis of DNA mobility shift experiments, the residual contents of the sample application slots after electrophoresis were dot-blotted onto nitrocellulose (see two paragraphs further down). Proteins in the agarose gel were electrophoretically transferred to Immobilon membranes (Millipore) by the semi-dry method. Immunostaining of Western blots and dot-blots was carried out as described in the next paragraph.
Immunoband De/Repletion-200 units of human topoisomerase I was incubated with 6 g of calf thymus DNA in the presence of 10 mM MgCl 2 (for the study of noncovalent DNA binding) or with 4 g of calf thymus DNA in the presence of 2 mM MgCl 2 (for the study of covalent complex formation) in a final volume of 40 l of reaction buffer. Incubation at 37°C for 30 min with and without drugs was terminated by addition of 1% SDS. Controls were without drugs, without DNA, with enzyme alone or with a subsequent DNA digestion with 250 units of detergentresistant endonuclease Benzonase® for 10 min at 37°C. For the study of drug effects in intact cells, 10 6 HL-60 cells were cultivated for 30 min at 37°C with and without drugs. In this case the reaction was terminated by sedimentation of the cells (1000 ϫ g, 5 min, 4°C) and subsequent lysis in 1% SDS. Samples were subjected to SDS-polyacrylamide (8%) gel electrophoresis, and proteins that had entered into the gel were electrophoretically transferred to nitrocellulose sheets by the semi-dry method. Immunostaining of immobilized proteins was carried out using a mouse monoclonal antibody of human topoisomerase I, peroxidaselabeled goat anti-mouse IgG, and the ECL-system (Amersham). Migration distances of immunostained protein bands were compared with that of rabbit muscle myosin (212 kDa), ␣ 2 -macroglobulin from bovine plasma (170 kDa), ␤-galactosidase from Escherichia coli (116 kDa), human transferrin (76 kDa), and bovine liver glutamic dehydrogenase (53 kDa).
Immunodot-blot Analysis of DNA-linked Topoisomerase I-400 units of human topoisomerase I was incubated with 3 g of calf thymus DNA in the presence of 2 mM MgCl 2 in a final volume of 500 l of reaction buffer. Incubation at 37°C for 30 min with and without drugs was terminated by addition of 0.2% SDS. Controls were incubated without drugs or were subsequently treated with 250 units of detergent resistant endonuclease Benzonase® for 10 min at 37°C. Samples were filtered through nitrocellulose sheets using a 96-well vacuum dot-blot aparatus (Schleicher & Schü ll), followed by two washes (500 l) with 10 mM Tris-HCl, pH 7.5, 50 mM NaCl. Nitrocellulose sheets were irradiated (254 nm, 2 min), dried at 20°C for 12 h, and finally subjected to immunostaining with mouse monoclonal antibody to human topoisomerase I as described in the Western blot section. Since flavonoids bound to the nitrocellulose interfered with the ECL method, detection of bound antibodies was in this case carried out with gold-labeled rabbitanti-mouse IgG and silver enhancement of the gold label.
Separate analysis of drug effects on topoisomerase I-mediated DNA cleavage and religation was carried out with an oligonucleotide suicide substrate, as described in (17). We used a double-stranded 36-mer oligonucleotide (Fig. 1a) containing a strong topoisomerase I cleavage sequence described in (18). The cleaved strand had a nick two base pairs 3Ј to the major cleavage position and was labeled with [ 32 P] at the 5Ј-end. All other 5Ј-ends including that of the nick were blocked by phosphorylation, in order to prevent religation at these sites. The composition of the substrate is schematized in Fig. 1a. In agreement with (19) we found the nicked fully double-stranded oligonucleotide a much better substrate for topoisomerase I, than the partially double-stranded suicide substrate originally described in Svejstrup et al. (17). For the FIG. 1. Schematic outline of the separate measurement of topoisomerase I-mediated cleavage and religation. A, the oligonucleotide substrate is composed of a 36-mer noncleaved strand, a 20-mer cleaved strand, and a 16-mer duplex-forming strand. Noncleaved and duplex-forming strands were nonradioactive, and the cleaved strand was radioactively phosphorylated on the 5Ј-ends with E. coli T4 polynucleotide kinase. The oligonucleotides were hybridized in a ratio of 1:1.5:2 (cleaved:noncleaved:duplex-forming). B, suicide cleavage of the oligonucleotide was initiated by addition of 200 units of purified human topoisomerase I/0.2 pmol of substrate and was allowed to continue for 30 min at 30°C. Religation was subsequently started by addition of 200 pmol of the biotinylated dinucleotide and carried out in the presence of 330 mM NaCl at 37°C for 30 -60 min. C, electrophoresis of the noncleaved (left), the cleaved (middle), and the religated substrate (right) in a 14% polyacrylamide gel under denaturing conditions. Samples were trypsinized and denatured with formamide and heating (96°C for 5 min) before electrophoresis.
cleavage half-reaction, 0.2 pmol of substrate was reacted with 300 units of purified recombinant human topoisomerase I in a final volume of 20 l of 20 mM potassium phosphate, pH 7.0, at 30°C for 30 min. Under these conditions, the oligonucleotide becomes mainly cleaved in a position 2 base pairs 5Ј to the nick. A second minor (5-10%) cleavage occurs 4 base pairs 5Ј to the nick. The di(tetra)nucleotide that is cleaved off escapes from a further religation reaction by diffusion. Religation to the 5Ј-end distal of the nick is blocked by phosphorylation. Therefore, the cleavage is irreversible, and the cleaved substrate becomes quantitatively converted to a covalent complex with the enzyme (Fig. 1b). For religation, a 1000-fold excess of a 3Ј-biotinylated GA dinucleotide was subsequently added to the topoisomerase I-DNA complex as a religation substrate together with 330 mM NaCl, in order to prevent recleavage of the religated DNA (Fig. 1b). The religation reaction was carried out at 37°C for 30 -60 min. After trypsinizing (4 mg/ml, 10 min, 37°C), the oligonucleotide DNA was precipitated (75% ethanol, 400 mM NaCl, Ϫ70°C, 30 min) and redissolved in 10 mM Tris, 0.1 mM EDTA, pH 7.5. 40% formamide, 0.5% bromphenol blue, 0.5% xylene blue was added to the samples followed by heating to 90°C for 5 min. Noncleaved, cleaved, and religated forms of the labeled oligonucleotide strand were electrophoretically separated on 0.5-mm 14% polyacrylamide gels under denaturing conditions (5 M urea in TBE buffer) and visualized by autoradiography (see Fig. 1c).
Where statistics could not be applied to the data, representative examples of at least two experiments with similar results are shown.

Inhibition of Topoisomerase I-catalyzed DNA Religation by
Quercetin and Related Natural Flavonoids-For the screening of topoisomerase I-targeted drug effects, we used a strategy based on the alteration of the electrophoretic mobility of pBR 322 plasmid DNA by the combined action of human topoisomerase I and inhibiting drugs (20). As shown in Fig. 2a, the mobility of the naturally supercoiled closed circular doublestranded plasmid DNA increases upon topoisomerase I-mediated relaxation, when electrophoresed with ethidium bromide.
In the presence of camptothecin that binds to the covalent topoisomerase I-DNA intermediate and inhibits the religation half-reaction (21), topoisomerase I introduces nicks into one of the DNA strands. The resulting open circular plasmid DNA migrates in a similar position as DNase I-nicked pBR 322. It is slower than the closed circular and the linearized plasmid forms. It should be noted that the closed circular plasmid forms incorporate much less ethidium bromide than the open circular form, because staining of the closed circular DNA forms is limited by topological constraints, whereas the nicked plasmid form can incorporate a maximal number of ethidum bromide molecules. Actually, direct quantitation of the fluorescence of ethidium bromide in the DNA bands showed that opened circular pBR 322 DNA incorporated about 1.8 times more ethidium bromide than an equivalent amount of closed circular pBR 322 DNA. As pointed out by others (21), this observation may explain why decreases in fluorescence of the open circular DNA band are not strictly matched by increases in the closed circular DNA form (seen Fig. 2b). Fig. 2 shows that quercetin alone does not alter the migration of pBR 322 DNA, but in the presence of topoisomerase I it induces formation of open circular plasmid DNA in a fashion similar to camptothecin. It can also be seen that the topoisomerase I-mediated effect of quercetin is much stronger than that of genistein described previously (15). We quantitatively compared topoisomerase I-mediated formation of open circular plasmid DNA for 13 natural flavonoids and 8 synthetic derivatives with that of camptothecin by video densitometric analysis of fluorescence intensity of the electrophoretically separated DNA bands (Table I). Camptothecin increased the intensity of the open circular plasmid form by 3.17-fold, which is equivalent to a complete conversion of the pBR-DNA (compare Table I and Fig. 2a). Among the various flavonoids tested, only quercetin and the closely related compounds, morin, fisetin, kaempferol, acacetin, and apigenin, were able to increase significantly the open circular plasmid DNA form in a topoisomerase I-dependent fashion. Of these substances, quercetin was the most potent compound, giving a more than 2-fold effect (Table I), which is equivalent to capturing more than 50% of the plasmid DNA in a topoisomerase I-complexed form (Fig. 2a). The EC 50 (dose of half-maximal effectiveness) of quercetin for this effect is approximately 300 M (Fig. 2b). It should be noted that there seems to be a quantitative discrepancy between the data in Fig. 2 and Table I. The values for increases of intensity of the open circular DNA band stated in Table I were derived from direct measurements of fluorescence intensity of the respective areas of the gels by video-enhanced imaging. When assessed in this way, the difference between background cleavage with enzyme alone (Fig. 2a, lane 7) and apparently complete conversion of the DNA substrate to the open circular form in the presence of camptothecin (Fig. 2a, lane 3) was about 3.3-fold. When looking at the polaroid picture of the gel (Fig.  2a), a much larger difference between the open circular bands in lane 3 and 7 becomes apparent, because the high-speed black and white polaroid film used (Type 667, ISO 3000/36°C), has a pronounced property of contrast enhancement, decreasing weak signals and enhancing strong ones.
In order to confirm that the generation of nicked plasmid DNA by the combined action of topoisomerase I and quercetin shown in Fig. 2 is indeed due to the stabilization of the covalent topoisomerase I-DNA catalytic intermediate, we performed the immunoband depletion experiment shown in Fig. 3a. The recombinant human topoisomerase migrates as a single protein band of approximately 100 kDa in SDS-polyacrylamide gels. The amount of enzyme that enters the polyacrylamide gel is not decreased by preincubation with quercetin (Fig. 3a, lane 2) or camptothecin alone (not shown). When the enzyme is preincu- bated with 4 g of calf thymus DNA in the presence of 2 mM MgCl 2 , a minor amount becomes trapped in the covalently DNA-linked form so that it cannot enter the gel. Accordingly, the Western blot signal is slightly diminished (Fig. 3a, lane 3). Co-incubation of enzyme and DNA with camptothecin results in a nearly complete disappearance of the Western blot signal, because all the enzyme becomes covalently trapped on the DNA and cannot enter the gel (Fig. 3a, lane 4). The DNA-linked enzyme can be released by digestion with a detergent-resistant endonuclease and subsequently migrates as a broader band with increased molecular weight, because it becomes retarded by covalently linked oligonucleotide adducts of various sizes (Fig. 3a, lane 5). Co-incubation of enzyme and DNA with quercetin induces a similar immunoband depletion as camptothecin (Fig. 3a, lane 6), indicating that both substances stabilize the covalent topoisomerase I-DNA intermediate.
Direct evidence of the drug-induced covalent DNA linkage of topoisomerase I was obtained by the dot-blot analysis shown in Fig. 3b, which makes use of the selective binding of topoisomerase I-DNA complexes to nitrocellulose in the presence of SDS (20). Under these conditions topoisomerase I alone does not bind to the nitrocellulose filter. Also when preincubated with DNA only a minor fraction of the enzyme binds to the filter (Fig. 3b, Control). However, after preincubation with camptothecin and DNA, the amount of enzyme retained on the filter is greatly increased (Fig. 3b, camptothecin). This increase in filter binding reflects the formation of covalent topoisomerase-DNA complexes, because it is dependent on the presence of DNA (not shown) and can be abolished by DNase treatment prior to filtration (Fig. 3b, left column). Most of the flavonoids, which were able to increase topoisomerase I-dependent formation of open circular pBR 322 (Table I) also increased the fraction of topoisomerase I that was retained on nitrocellulose filters in a DNA-dependent manner (Fig. 3b). However, when tested in this way, morin and fisetin showed weaker effects than to be expected from the data in Table I. This difference may be explained by the fact that in the dot-blot assay, 3-fold lower drug concentrations (300 M versus 1 mM) were used, in order avoid artifacts arising from the binding of the drugs directly to the nitrocellulose.
In order to find out whether quercetin (and related natural flavonoids) can also stabilize covalent complexes of topoisomerase I and DNA in intact cells, we carried out immunoband depletion experiments with human HL-60 cells. These data are summarized in Fig. 3c. Clearly, quercetin was the most effective substance. It made the topoisomerase I band disappear almost completely from immunoblots of HL-60 cells, when applied to the incubation medium at 300 M concentrations for 2 h. The extent of topoisomerase I immunoband depletion was similar to that obtained by 20 M of camptothecin. Treatment  of the cells for 2 h with 300 M concentrations of acacetin, kaempferol, or morin also significantly diminished the topoisomerase I signal, however, to a lesser degree than camptothecin or quercetin. Fisetin or apigenin did not cause an immunoband depletion of topoisomerase I in HL-60 cells under similar conditions. This is in contrast to the effectivity of both substances in cell-free assays (Table I, Fig. 3b) and may be explained by differences in cellular uptake, metabolism, or accessibility to topoisomerase I and/or DNA.
Inhibition of Topoisomerase I-DNA Binding by EMD 50 689 and EMD 21 388 -In contrast to quercetin and its natural congeners, none of the synthetic derivatives of the flavone structure tested in this study (see Table I) stabilized the covalent topoisomerase I-DNA complex in a similar fashion as camptothecin or quercetin. However, some of these derivatives effectively inhibited topoisomerase I-catalyzed relaxation of pBR 322 DNA, the most notable being EMD 50 689 (6-cyclopentanyloxy-3Ј,5Ј-dibromo-4Ј-hydroxy-3-methylflavone) (see Fig. 2, lane 5). EMD 50 689 also decreased the amount of open circular pBR 322 DNA by 3-fold as compared with that formed by topoisomerase I alone (Table I). Both observations indicate a specific interaction of EMD 50 689 with the DNA cleavage reaction of topoisomerase I either by an inhibition of the association between enzyme and DNA substrate or by an inhibition of the cleavage reaction itself after noncovalent association of enzyme and DNA. In order to distinguish between these possibilities, we carried out the experiments shown in Fig. 4. Fig. 4a proves that under nondenaturing conditions a noncovalent complex of topoisomerase I and pBR 322 DNA is formed, which is immobile in native agarose gel electrophoresis. To show this, DNA and topoisomerase I were incubated together and subsequently subjected to native agarose gel electrophoresis carried out in TBE buffer containing 0.5 g/ml ethidium bromide (lane 4). The sample of lane 5 was denatured with SDS before electrophoresis. The sample in lane 3 was also digested with proteinase K. Lane 2 shows DNA alone. Lane 1 shows enzyme alone, not treated with SDS. The influence of these different conditions upon the electrophoretic mobility of DNA and topoisomerase I in the agarose gel was compared. After electrophoresis the gel was first stained with ethidium bromide, in order to document migration of DNA (Fig. 4a, top  panel). Subsequently, topoisomerase I remaining in the sample application slot after electrophoresis was visualized by immunoblotting (middle panel). Finally, topoisomerase I that had entered the agarose gel was electroblotted onto an Immobilon membrane and also visualized by immunostaining (bottom panel). Although of poor quality, the electroblot of the agarose gel shows clearly that topoisomerase I, not treated with SDS, and not bound to DNA, enters the gel as a diffuse band (lane 1). Preincubation of topoisomerase I with EMD 50 689 before adding pBR 322 DNA clearly inhibits the formation of the electrophoretically immobile, noncovalent topoisomerase I-DNA complex in a dose-dependent manner (Fig. 4b, lanes   FIG. 3. Evidence for drug-stabilized covalent topoisomerase I-DNA complexes. A, immunoband depletion of purified topoisomerase I: 200 units of human topoisomerase I was incubated with 4 g of calf thymus DNA in the presence of 2 mM MgCl 2 at 37°C for 30 min with and without drug. The reaction was terminated by addition of 1% SDS. Controls were without drugs, without DNA, with enzyme alone, and with a subsequent digestion with 250 units of detergent-resistant endonuclease Benzonase® for 10 min at 37°C. Samples were subjected to SDS-polyacrylamide (8%) gel electrophoresis and proteins that had penetrated into the gel were electrophoretically transferred to nitrocellulose sheets and visualized with mouse monoclonal antibodies to topoisomerase I. B, dot-blot analysis: 400 units of human topoisomerase I was incubated with 3 g of calf thymus DNA in the presence of 2 mM MgCl 2 at 37°C for 30 min with and without drugs. The reaction was terminated by addition of 0.2% SDS. Controls were incubated without drug (line 1) or subsequently treated with 250 units of Benzonase (left column). Samples were filtered through nitrocellulose sheets, UV-irradiated, dried, and stained with mouse monoclonal antibody to human topoisomerase I. Bound antibody was detected with gold-labeled rabbit anti-mouse IgG and silver enhancement. C, immunoband depletion of topoisomerase I in intact human HL-60 cells: 10 6 cells were cultured with or without drugs for 2 h, followed by sedimentation, SDS lysis, and immunoblot analysis, as described in A. [2][3][4][5], although EMD 50 689 does not alter the electrophoretic mobility of the DNA in the absence of enzyme (Fig. 4b, lane 1, see also Fig. 6a). This observation gives a clear indication that EMD 50 689 inhibits the noncovalent binding between topoisomerase I and its DNA substrate with an EC 50 of about 100 M. This observation was confirmed by the experiment shown in Fig. 5, where topoisomerase I was incubated with 6 g of calf thymus DNA and 10 mM MgCl 2 . Under these conditions a major fraction of the enzyme present in the assay binds to the DNA substrate and upon SDS denaturation does not enter the polyacrylamide gel, as can be monitored by depletion of the Western blot signal. When EMD 50 689 was included, DNA binding of the enzyme was blocked, and, accordingly, the Western blot signal reappeared. A similar effect could also be observed with EMD 21 388, but only when the drug was preincubated with the enzyme before DNA was added. EMD 49 223 was ineffective under all conditions tested. This points to a potential role of a voluminous hydrophobic substituent in position 6 of the phenol ring A of the flavone structure (see Table I) in the inhibition mechanism.
In summary, the data shown in Figs. 4 and 5 give a clear indication that the inhibition of the catalytic activity of topoisomerase I by EMD 50 689 observed in Fig. 2a (lane 5) is due to an inhibition of the DNA binding of the enzyme. This effect of EMD 50 689 could be likewise obtained by intercalation into partially relaxed plasmid DNA or by a direct interaction of the drug with the enzyme itself. In order to distinguish between these possibilities, we carried out the experiments shown in Fig. 6. When pBR 322 plasmid DNA was first completely relaxed by topoisomerase I and then incubated with EMD 50 689, the electrophoretic mobility of the relaxed plasmid DNA (determined in the absence of ethidium bromide) did not change, indicating that EMD 50 689 does not intercalate into, or bind to, the DNA. On the other hand, the inhibitory effect of EMD 50 689 on topoisomerase I-mediated DNA relaxation can be enhanced by preincubation with the enzyme before addition of the DNA plasmid. Upon preincubation with enzyme, effective inhibition of topoisomerase I-catalyzed pBR 322 DNA relaxation could already be observed at 100 M concentrations, in contrast to 300 M without preincubation (Fig. 6b). Taken together, these data suggest that EMD 50 689, and to a lesser degree, EMD 21 388 (see Fig. 5) bind directly to topoisomerase I. In the drug-bound form the enzyme cannot bind and cleave its DNA substrate. As these steps are a prerequisite for the action of drugs like camptothecin, that inhibits the subsequent DNAreligation step, EMD 50 689 should be able to antagonize camptothecin and protect the enzyme from this drug. This is actually shown in Fig. 7, where camptothecin-and topoisomerase I-mediated formation of open circular DNA is abolished by a preincubation of the enzyme with EMD 50 689. However, this antagonism implies that a putative effect of EMD 50 689 on the second half-reaction of topoisomerase I cannot be measured by assays, in which the enzyme is allowed to go repeatedly through the complete catalytic cycle of DNA cleavage and reli- Drugs were either added together with the DNA or the enzyme was first incubated with the drugs for 10 min at 37°C where after DNA was added. The reaction was terminated by addition of 1% SDS. Controls were without drugs, without DNA, and with enzyme alone. Samples were subjected to SDS-polyacrylamide (8%) gel electrophoresis and proteins that had penetrated into the gel were electrophoretically transferred to nitrocellulose sheets by the semi-dry method and visualized with mouse monoclonal antibodies to topoisomerase I. gation. Moreover, catalytic assays do not allow a precise chronological dissection of the inhibitory mechanism in relation to the catalytic cycle. Thus, the step at which an inhibitor needs to enter the catalytic cycle, and the step at which it becomes effective in trapping or inhibiting the enzyme cannot be differentiated. In order to overcome these problems, we have made use of an oligonucleotide suicide substrate of topoisomerase I, which restricts the enzyme to a single round of cleavage and religation and allows for addressing the two half-reactions separately (for details, see Fig. 1). As shown in Fig. 8a, the cleavage reaction could be effectively blocked by EMD 50 689 in this system when the enzyme was preincubated with the drug, but not when drug and enzyme were simultaneously added to the substrate, thus confirming that EMD 50 689 needs first to interact with the enzyme in order to inhibit subsequent substrate binding. However, EMD 50 689 did not inhibit the religation of a topoisomerase I suicide substrate complex that had been formed previously in the absence of the drug (Fig. 8b). A similar result was also obtained when the enzyme-substrate complex was preincubated with the drug before the religation subtrate was added (not shown). Thus, EMD 50 689 appears to be a selective inhibitor of the free enzyme and does not target topoisomerase I in its DNA-bound form.
As expected (21), camptothecin effectively inhibited religation of the cleaved suicide complex (Fig. 8b), but not the cleavage reaction itself (Fig. 8a). Similarly, quercetin neither affected the cleavage reaction when added together with the enzyme (Fig. 8a) nor when preincubated with the enzyme (not shown). However, in contrast to camptothecin, quercetin also failed to inhibit the religation reaction, when added to an enzyme-substrate complex that had been previously formed in the absence of the drug (Fig. 8b). Quercetin also did not inhibit topoisomerase I-mediated DNA religation, when preincubated with the enzyme-substrate complex before addition of the religation substrate (not shown). Inhibition of the religation reaction by quercetin was only observed, when the drug had already been included into the cleavage reaction (Fig. 8b).

Different Modes of Inhibition of Topoisomerase I
As summarized in Fig. 9, three distinct mechanistic concepts of topoisomerase I inhibition, can be deduced from our data. (i) EMD 50 689 and EMD 21 388 bind selectively to topoisomerase I in its free, not DNA-bound, form. When bound to the enzyme, these drugs block the access to the DNA subtrate and, thus, by trapping the enzyme in its free form prevent all subsequent steps of the catalytic cycle. This mode of inhibition closely resembles that reported of acidic phospholipids (22) and of tyrphostins, like AG-555 (23). (ii) Camptothecin binds selectively to the covalent enzyme-DNA post-cleavage complex, inhibits the subsequent religation step, and, thus, stabilizes the catalytic intermediate (21). (iii) Quercetin and some closely related natural flavone derivatives also stabilize the covalent enzyme-DNA post-cleavage complex, but by a mechanism different from camptothecin. In order to stabilize the catalytic intermediate, these drugs need to form a ternary complex with topoisomerase I and the DNA subtrate during the cleavage reaction, which does not affect the cleavage itself, but prevents a subsequent religation reaction. Ternary complex formation has also been shown for the inhibition of topoisomerase II by aminoacridines (24).

Structure-Function Relationships of Flavonoids Relevant for Topoisomerase I Inhibition
Inhibition of topoisomerase I is restricted to the group of the flavones, whereas chalcones, aurones, flavanones, and isoflavanones are inactive (Table I). The ability of these compounds to form a planar, conjugated A-C-ring system appears to be essential for inhibition (compare dihydroquercetin versus quercetin). Flavones can either stabilize the catalytic topoisomerase I-DNA intermediate (Fig. 9c) or inhibit the DNA binding of the free enzyme (Fig. 9a). These two effects, which are selective with respect to the catalytic step and antagonistic to each other, can be related to certain substitutions of the flavone structure.
Stabilization of Cleavable Complex-The potency of quercetin and related compounds for stabilizing the covalent topoisomerase I-DNA intermediate requires hydroxyl substitutions in positions 5 and 7 and a OH or OCH 3 substituent in the 4Ј position (see Table I). Flavones lacking -OH in the 4Ј, 5, or 7 position (chrysin and the synthetic EMDs) are inactive. However, -OH in the 5 position is not essential, because Fisetin is inhibitory. The same reasoning excludes intramolecular bonding between the 4-keto group and the 5-OH group or chelation of divalent cations by these two substituents (1) as mechanism of inhibition. Inhibitory effects do also not correlate with structures known for participation in antioxidant or iron-chelating effects of flavonoids (25,26). Hydroxylation in the 2Ј position (compare morin to quercetin or kaempferol or EMD 47 020 to apigenin) or substituents in the 3 position (genistein) reduce the inhibitory potency. Thus, a planar A-C benzopyron structure (27) seems to be important in forming ternary complexes with enzyme and DNA. The 3Ј-4Ј-ortho-dihydroxy substitution pattern in the B-ring of quercetin could also indicate a catecholtype mechanism of inhibition known from inhibition of other enzymes (28,29). A similar ortho-dihydroxy-substituted phenolic ring is also essential for inhibition of topoisomerase I by certain tyrphostins (23). However, flavones with methoxy substituents preventing catechol-like reactions (acacetin and EMD 20 940) are also effective inhibitors.
Inhibition of Topoisomerase I-DNA Binding-Binding to free topoisomerase I and blocking of its access to DNA can be assigned to the general structure of 6-substituted, 3-methyl,3Ј,5Ј-dibromo,4Ј-hydroxyflavone. The tetra-substituted phenolic B-ring per se is not inhibitory (EMD 47 020 is inactive). However, the bulky bromo substituents and the 3-methyl group are forcing these compounds into nonplanar structures (the B-ring becoming perpendicular to the plane of the A-C-ring system (30, 31)) that will not intercalate into the DNA (as actually shown in Fig. 6a). Effectiveness of these compounds decreases when the 6-substituent is negatively charged: EMD 50 689 (6-cyclopentyloxy-) is more effective than EMD 21 388 (6-hydroxy-); EMD 49 233 (6-phosphate-) and EMD 49 224 (6-sulfate-) are ineffective. This suggests that a voluminous hydrophobic substituent in the 6-position is essential for interaction with topoisomerase I. Similar reasoning might also explain the effectiveness of EMD 47 059, carrying two phenolic substituents at the chromanone structure. Moreover, the tyrphostin AG-555 carrying a similar phenylpropane substituent has been reported to be the most potent inhibitor of topoisomerase I catalytic activity, as compared with its congeners (23). The effectiveness of completely different aromatic or heterocyclic ring compounds suggests the presence of an essential spacy hydrophobic pocket in the topoisomerase I molecule in or close to the site involved in noncovalent DNA binding.

Potential Value of Flavone Inhibitors of Topoisomerase I
Inhibition of topoisomerase II by substances that stabilize the covalent enzyme-DNA intermediate is a longstanding and important therapeutic concept in cancer therapy (32,33). In contrast, drugs such as camptothecin that stabilize the covalent topoisomerase I-DNA intermediate have only just recently been introduced into cancer chemotherapy, although camptothecin is among the most effective anti-cancer drugs. However, the clinical use of camptothecin and its water-soluble derivatives is compromised by the chemical instability of a mechanistically crucial lactone ring under neutral and slightly alkaline conditions (34). Recently, the fungal toxine saintopin (35,36) and the 7H-benzo-pyrido-indole intoplicine (37) have been re- FIG. 9. Model of selective drug interactions with topoisomerase I reactions: the catalytic cycle of topoisomerase I is divided into DNA binding, cleavage, and religation. Each of these partreactions is marked by a double arrow, inhibition of the respective step is indicated by a solid cross-bar. Diffusional movements are indicated by single arrows. A, EMD 50 689 and 21 388 (EMD) bind to free topoisomerase I and subsequently inhibit binding of the free enzyme to the DNA cleavage site, thus preventing all following steps of the catalytic cycle. B, camptothecin (CPT) binds to the covalent topoisomerase I-DNA complex (after cleavage) and inhibits subsequent religation. C, quercetin (QCT), when captured during the cleavage reaction within the covalent topoisomerase I-DNA complex, inhibits subsequent religation, but does not interfere with the DNA binding of the enzyme or the cleavage reaction itself. ported to stabilize the catalytic intermediate of topoisomerase I and topoisomerase II. Early clinical trials (38,39) show that these substances may form a new class of anti-tumor drugs, which are active on a variety of solid tumors and escape crossresistance to drugs that target solely one type of topoisomerases. Quercetin is already known (13) to act on topoisomerase II religation activity in a similar way, as shown here for topoisomerase I. Although the mechanism of action of quercetin appears to be different from that of camptothecin, we failed to obtain clear experimental evidence of an additive action of the two drugs on topoisomerase I-mediated DNA cleavage. However, in view of the relatively low toxicity of quercetin and related flavonoids and the known anti-tumor activity of these substances (2), their use as anti-cancer drugs seems feasible. Our data defining some of the structural features, which are decisive for the topoisomerase I-directed effects of these substances, can be used to set out the course for a systematic search for more potent inhibitors targeting both types of topoisomerases. Moreover, the flavone structure also provides a template for the creation of a second class of topoisomerasetargeted drugs, which share the common structure of 6-substituted 3Ј,5Ј-dibromo-4Ј-hydroxy-3-methylflavones and selectively inhibit the cleavage half-reaction of topoisomerase I. In contrast to other substances known to inhibit the cleavage reaction of topoisomerase I, such as chebulagic acid (20) and ␤-lapachone (40), 6-substituted 3Ј,5Ј-dibromo-4Ј-hydroxy-3methylflavones do not inhibit the DNA cleavage reaction itself, but interfere with the the binding of the enzyme to DNA.